EPDM Roof System Performance: An Update of Historical Warranty Service Costs

September 2003 Interface • 29
INTRODUCTION
One of the most important steps in roof design involves selecting
the best roof system for specific building conditions and building
owner needs. In order to make the optimum selection, the
designer must evaluate the costs and benefits of the many options
available. While initial costs of roofing systems can be calculated
with reasonable accuracy, it is much more difficult to accurately
calculate the longer term warranty service and maintenance costs
for different roofing systems. In order to make the best decision,
the designer needs statistically accurate and reliable information
based on actual historical performance of roofing systems.
The warranty service records of roofing manufacturers offer an
interesting opportunity to expand the industry’s understanding of
roof system performance. Because warranty records typically
include thousands of roofs covering millions of square feet, an
analysis of the information contained in these records may be
more statistically reliable than information gathered from limited
sources. Because warranty service records also identify when a
roof was installed and when warranty service was performed, they
can provide an accurate chronology of roof performance. Finally,
because warranty records provide the actual cost of any roof service
performed, roof performance can be analyzed from an economic
perspective.
THE ROOF SERVICE DATABASE
The database used in this study is significant both in size and
scope. In regard to size, the database covers over three billion
square feet of EPDM roofs installed on over one hundred fifty
thousand roofs from 1982 to the present. Regarding the scope of
detail provided in the database, the file for each roof contains
extensive background information, including:
1. Construction type (new construction/re-roof/recover)
2. System type (ballasted/adhered/mechanically attached)
3. Geographic location (city/state or province)
4. Installing contractor (identification and current license status)
5. Warranty information (date completed/warranty term)
Because all roofs received a material and workmanship warranty,
the database also contains detailed information regarding
the timing, cost, and type of service performed or authorized by
the manufacturer during the warranty period. The typical warranty
period of the majority of roofs in the database is ten years, and
the service records for each roof identify the cost and date of all
service expenditures made by the manufacturer during this period.
These expenditures include both emergency leak calls as well
as more comprehensive repairs when necessary. In addition, each
service activity is “coded” based on the principal roof system component
that required service, allowing for a detailed analysis of
the causes of service expense.
Given the level of detail available in the database, a number of
trends can be analyzed, including the relative performance of
roofs over time, the average frequency and timing of warranty service
activities, and the comparative performance of different roof
systems, components, and construction types.
DEFINITIONS AND METHODOLOGY
1. Historical warranty service cost is defined as the manufacturer’s
actual recorded spending during a specified period
of service. For this study, costs are expressed as actual
spending during each year of roof service following the initial
installation.
2. Warranty service costs for any given segment of the database
are divided by the total installed surface area of that
segment, in order to allow for comparison between different
segments of the database. As a consequence, the
annual unit warranty service costs can be compared for
any year.
3. In order to adjust for inflation, spending is stated in constant
units, using 1982 as base year and adjusting each
year based on the U.S. National GDP Deflator as calculated
by the United States Bureau of Labor Statistics.
4. In order to maintain the confidentiality of the manufacturer’s
actual dollar costs in the database, warranty service
costs are indexed against a baseline year.
FOREWORD
This paper provides a five-year update of the article, “EPDM Re-Roofing Versus Recover:
A Comparison of Historical Maintenance Costs,” originally published in Interface in July 1998.
30 • Interface September 2003
EPDM ROOF SYSTEM PERFORMANCE:
HISTORICAL TRENDS
Overall EPDM Warranty Service Cost Trends
At least ten years of actual historical warranty service costs
are now available for all roofs installed in 1993 or earlier. Chart 1
illustrates the overall change in warranty service spending during
the first ten years of service for EPDM roofs installed each year
from 1982 to 1993. As illustrated by the chart, unit warranty service
costs declined 85% from 1982 to 1987. As will be discussed
later in this paper, the significant drop in costs is most likely
attributable to changes in product technology introduced during
those years. Although the improvement trend begins to flatten out
slightly from 1987 through 1993, cost reductions continue each
year, and even these smaller incremental improvements account
for another 60% reduction in warranty service cost.
It should be noted that when this study was first published in
1998, complete historical data were available only for the first five
years of service for many of these roofs. As a consequence, the
ten-year warranty service costs were
estimated by applying regression analysis
to the five-year trends available at
that time. Five years later, it is very gratifying
to report that the actual ten-year
trends indeed follow the 1998 prediction.
In order to demonstrate the closeness
between the 1998 projections and
actual results today, a line indicating
the original projection has been added to
Chart 1. It should be noted that any
slight differences in the projected and
actual cost are statistically insignificant
at a 99.5% level of reliability.
Warranty Service Costs by
Component
Chart 2 provides a breakdown of tenyear
costs by the type of component
requiring warranty service. For this study, components
were divided into three broad categories:
1) field seams; 2) perimeter flashings,
such as base tie-ins, edge metal flashings, and
terminations; and 3) other components, including
curbs, penetrations, and membrane attachment.
During the period from 1982 to 1993,
warranty service costs for both field seams and
perimeter flashings improved significantly. The
trends in this chart appear to reinforce the
commonly held opinion that much of the
improvement in EPDM system performance is
due to changes in seaming and attachment
technology. Costs associated with field seams
and perimeter flashings have declined 92% and
95%, respectively. It should be noted that this
improvement coincides with the introduction of
several important component technologies by
the roofing manufacturer:
Year of Intro: Technology:
1985 -1986 Butyl-based splice adhesive replaces Neoprenebased
adhesive.
1985 – 1986 EPDM-based wall flashings replace Neoprenebased
flashings.
1987 – 1988 Tape laminates replace adhesive seams at roof
edges and battens.
1988 – 1989 Metal battens and screw fasteners replace wood
nailers and nails.
1991 – 1992 Reinforced perimeter fastening strips
introduced.
1992 – 1993 Seam tape with high-solids primer replaces
seam adhesive.
Warranty Service Costs by System Type
Chart 3 provides a breakdown of ten-year costs by type of
EPDM roof system. For this study, EPDM systems are divided into
three major categories: 1) ballasted, 2) fully adhered, and 3)
mechanically attached. The chart clearly illustrates the relative
performance of these three basic system types has also changed
September 2003 Interface • 31
significantly over time. While both
mechanically attached and fully
adhered roofs under performed ballasted
roofs in the early years of the
database, all three systems now provide
almost identical performance.
This performance undoubtedly is related
to previously identified improvements
in field seams as well as the
relative number of seams required for
each of these systems.
While ballasted roofing systems
typically use EPDM membrane panels
up to 50 feet wide by 100 feet long,
fully adhered and mechanically
attached systems frequently use smaller
roof panels requiring more field
seams. Given the relatively high warranty
service costs associated with
field seams in the early years of the
database, it is reasonable to infer that
roofs with a large number of field seams will be affected the most
by seam performance. As the performance of field seams has
improved, however, the number of seams in the roofing system
has become a less critical factor, thus narrowing differences in
system performance.
Ten-Year Warranty Service Costs Versus Initial
Installed Costs
Although the cost indexing employed in this study allows for
comparison of warranty service costs over time, this indexing provides
no immediate comparison to the initial cost of the roof.
However, using a similar indexing approach, the estimated initial
cost of the roofs in the database can be compared to the actual
ten-year warranty service costs. Based on cost data derived from a
nationally published construction cost estimating workbook,
indexed installation costs were calculated for several types of
EPDM systems in the database1. As
illustrated in Chart 4, warranty service
costs for EPDM roofing systems
account for a very small percentage of
total roof costs, even for systems
installed in the early 1980s. While tenyear
warranty service costs for EPDM
roofs installed in 1982 amounted to
less than 10% of the original installed
cost, the same ten-year warranty service
cost for EPDM roofs installed in
1993 is barely more than 1% of the
original installed cost.
EPDM Recover Versus
Re-roofing
One of the most important steps in
designing a roofing system for an
existing building involves the evaluation
of the existing roof. Even before
the specification for a new roof system
can be developed, the designer must
first decide whether to remove or retain the existing roof.
Although it is generally acknowledged that removal of the existing
roof system is the most conservative design practice, environmental
and economic concerns make complete tear-off increasingly
difficult. Before deciding whether to tear off the existing roof, the
designer must analyze the costs and benefits of complete removal.
Chart 5 shows the ten-year performance trend for two
approaches to roof replacement: 1) re-roof, or the tear-off of the
existing roof and installation of a new roof, and 2) recover, or the
installation of a new roof directly over the existing roof. For comparison
purposes, the chart also provides similar information for
roofs incorporated into new construction work. Although the tenyear
warranty service costs for new construction and re-roofing
projects installed during the early 1980s were almost equal, the
costs of recover projects in the early 1980s were significantly
higher than either new construction or re-roof warranty service
costs. However, this difference in costs narrows each year. By
32 • Interface September 2003
1988, the gap in warranty service cost between recover and reroofing
has virtually disappeared. In fact, for roofs installed
between 1988 and 1993, the difference in ten-year repair costs
between EPDM recover and re-roofing is statistically insignificant.
What caused the performance difference between EPDM reroofing
and recover in the early 1980s, and why did this gap close
so dramatically by the end of the decade? In order to better
understand this trend, we must look back at the performance
trends of the basic design and material components of EPDM roofing
systems. Although improvements in components such as field
seams and perimeter attachments obviously reduced warranty
service costs for both re-roofing and recover, it would appear the
effect on recover applications has been more favorable. Undoubtedly,
much of this improvement can be attributed to the elimination
of Neoprene-based seam adhesives, which were commonly
recognized to be more sensitive to moisture drive than butyl or
EPDM-based adhesives and seam tapes. Assuming a typical
recover project installed over an existing roof may contain more
latent moisture than a roof installed over a complete tear-off,
improvement in the moisture resistance of any component will
obviously contribute to lower service costs. Improvements in
perimeter securement also may have contributed to the relative
improvement of recover systems, since attachment to hidden substrates,
which is typical practice for recover projects, is obviously
more difficult than attachment to substrates that have been fully
exposed by the removal of an existing roof.
EPDM ROOF SYSTEM PERFORMANCE:
SUMMARY OF FINDINGS
Based on the warranty records of the three billion square feet
of EPDM roofs in this study, EPDM roofing performance has
improved dramatically over the past 20 years. In fact, the
improvement in performance now makes the cost of servicing an
EPDM roof for ten years almost negligible. Given the extremely low
cost of warranty service through the first ten years of service, it is
also likely that the great majority of these roofs will continue to
offer superior service for many years
beyond the ten years of this study.
It also appears this improvement has
been most significant for EPDM recover systems.
Given the escalating costs associated
with roof tear-off and the ever-present
financial constraints on building owners, a
roofing designer certainly can take comfort
knowing that a properly selected EPDM
recover system may provide an acceptable
service life at a reasonable cost. Of course,
the data presented in this article should not
be used to endorse recovers in all situations.
Existing roofs that are saturated with
water or seriously deteriorated are obviously
candidates for a complete tear-off. However,
the data in this study clearly suggest that
when an EPDM recover system is properly
installed over a suitable existing roof, it will
provide many years of acceptable service
with minimal warranty service expense. 
Footnote:
1. Installed cost data derived from 2002 National Construction
Estimator (Craftsman Books). Installed cost is based on the
application of a ten-year warranted EPDM roofing system
installed over an exposed metal or concrete roof deck and
a minimum R-20 polyisocyanurate roof insulation. Indexed
unit costs are adjusted for inflation during the period 1982
through 1993 based on the U.S. National GDP Deflator as
published by the U.S. Bureau of Labor Statistics. Indexed
unit costs are available only for ballasted and fully adhered
systems. Because mechanically attached EPDM systems
may vary significantly in system design, it was not
possible to develop a reliable unit cost using the general
system descriptions in the National Construction Estimator.
Jim Hoff (hoffjames@firestonebp.com)
has served in a variety of technical
and management roles in the construction
industry for over twenty-five
years. Currently, Hoff is vice president
of marketing for Firestone
Building Products Company and
serves as chairman of the board of
the Polyisocyanurate Insulation
Manufacturers Association (PIMA).
Hoff received an A.A.S. in architectural
technology from Indiana Vocational
Technical College, a B.A. in psychology from Indiana University,
a M.S. in management from Indiana Wesleyan University, and
currently is completing his doctoral dissertation for a D.B.A. in
management from the University of Sarasota.
ABOUT THE AUTHOR
JIM HOFF