Exterior Fire Performance of Low-Slope Roof Assemblies

May 15, 2004

22 • Interface February 2004
This article will focus on fire testing of roof
assemblies as described in ASTM E 108, NFPA
256, and UL 790 standards, all initially adopted
and published by their respective organizations
in 1958. The three standards are virtually identical
in terms of the fire testing procedures used
to evaluate roof assemblies, so, for simplicity,
only the E 108 Standard will be referenced in
this document.
The ASTM E 108 Test Standard was developed
based on fire testing procedures utilized
by Underwriters Laboratories Inc. (UL) since
1914 to classify roof assemblies. The purpose of
this fire test evaluation is to provide comparative
laboratory flammability characteristics of
roof assemblies in order to comply with building
code requirements. The E 108 Standard evaluates
the entire assembly, not individual components
or membranes. Roof assemblies are
tested using newly manufactured materials,
representative of those utilized in actual building
construction. The test data are developed
from simulated fire sources reproduced in a
laboratory environment as described in the
standard, which would originate outside the ASTM E 108 Test Apparatus
February 2004 Interface • 23
building. However, the
E 108 Standard is not
intended to predict the
expected performance of
roof assemblies under all
actual fire conditions.
The standard does not
apply to aged samples
removed from existing
roofs. Removal of samples
from the field, with the
intention of testing per
E 108, would require a
detailed sampling protocol
to control items such as
the sample selection
process, verification of the
materials used in the
installation, verification of
the initial fire rating of the
assembly, and conditioning
of the specimen prior to
testing. The possibility of
testing field-obtained samples
is discussed in greater
detail later in this article.
The ASTM E 108
Successful testing following
the ASTM E 108
procedure will result in a
fire classification of Class A, Class B, or Class C. Class A assemblies
are considered to be effective against severe fire test exposures.
Class B assemblies are considered to be effective against
moderate fire test exposures. Class C assemblies are considered
to be effective against light fire test exposures.
Testing is conducted on unaged roof assemblies installed over
combustible (wood) or non-combustible (concrete, gypsum, or
metal) roof decks. Combustible deck assemblies require successful
performance for all of the following tests: Spread of Flame,
Intermittent Flame, and Burning Brand. Non-combustible deck
assemblies require successful performance for only the Spread of
Flame test. The assembly is mounted in the test apparatus and
inclined between 1/8 inch and 5 inches per foot (as desired by the
material manufacturer) before testing. Classification is obtained at
the maximum slope tested if all required tests are passed (successful
testing at a 5-inch slope gains classification without any
limitation on slope). The following section summarizes these three
testing procedures.
Spread of Flame Test
The Spread of Flame Test measures the potential for flame
spread across the surface of the roof assembly using a constant
flame source.
Class A is achieved if the maximum flame spread does not
exceed 6 feet. Class B is achieved if the maximum flame spread
does not exceed 8 feet. Class C is achieved if the maximum flame
spread does not exceed 13 feet. For all classifications, the fire
cannot spread laterally to both edges of the test deck, burning or
glowing particles cannot fall to the floor and continue to burn or
glow, and a combustible roof deck cannot be exposed (visible) at
the conclusion of the test.
Intermittent Flame Test
The Intermittent Flame Test measures the potential for fire to
penetrate from the outside of the roofing assembly to the underside
(inside the building) of the combustible roof deck using a
variable flame source.
Class A Spread of Flame Test
24 • Interface February 2004
The flame temperature and wind speed are identical to those
listed in the Spread of Flame Test table on page 23. Successful
results are achieved provided there is no sustained flaming (a
continuous burning flame) of the underside of the test deck and
burning or glowing particles do not fall to the floor and continue
to burn or glow.
Burning Brand Test
The Burning Brand Test measures the potential for fire to
penetrate from the outside of the roofing assembly to the underside
(inside the building) of the combustible roof deck using a
burning brand fire source.
The brands are positioned on the top surface of roof assembly
at locations as described in the standard. Successful results are
achieved provided there is no sustained flaming (a continuous
burning flame) of the underside of the test deck and burning or
glowing particles do not fall to the floor and continue to burn or
Sample Preparation
The roofing assembly to be tested is installed over the test
deck as follows. The insulation or membrane underlayment is
attached to the deck using adhesives or mechanical fasteners, or
is loose-laid. The membrane is also attached to the insulation or
underlayment using adhesives or mechanical fasteners, or is
loose-laid. (The actual application depends on the assembly type
being tested). The membrane may be trimmed along the edges of
the test deck, or may be wrapped over the edges and nailed to the
test frame. The intent of the membrane application (and the wrapping
or not wrapping), as stated in Section 6.5 of ASTM E 108, is
to prevent air from traveling under the material being tested.
Wrapping the membrane edges creates a simulated field installation
where no exposed edges are encountered.
Testing Variation Potential
The potential for variation in test results does exist with ASTM
E 108. The flame calibration temperature and the wind speed calibration
ranges shown in the Spread of Flame Table on the previous
page provide for the possibility of a 100ºF temperature
variation and a 1 mph wind speed variation from one day to the
next (calibrating every day). Testing at the high end of the calibration
window one day can produce results different than those
obtained from testing at the lower end another day.
Roof assembly installation methods can influence the test
results. Adhered membrane assemblies behave differently than
mechanically fastened or loose-laid membrane assemblies. Adhered
membranes do not allow air to pass between the membrane
and the substrate layer, thus helping to retard flame spread.
Loose membranes allow air to pass under them, creating a more
severe condition. All roof assemblies should be tested as intended
for installation in the field.
Sample preparation methods can also influence the results.
Mechanically fastened and loose-laid roof assemblies typically
have the membrane wrapped over the sides of the test deck and
nailed to the frame when tested. These type assemblies, tested
without wrapping the membrane, allow air to blow under the
membrane and out the sides, creating a tunneling affect not
encountered on the roof. Attachment of these assemblies must be
consistent to achieve accu rate results.
Variability is also possible between different laboratories. The
placement of the test apparatus within the room and
the room’s dimensions can influence the test. The location,
size, and operating speed of the smoke exhaust
system can alter the airflow within the test room and
create different burn patterns.
Fire Performance of Aged or
Weathered Roof Assemblies
There are no provisions in the current E 108
Standard with respect to testing aged or weathered lowsloped
asphaltic or
single-ply roof assemblies. Typically, the fire performance
characteristics of aged roof assemblies have not
been a concern within the fire community as long as
the assembly is maintained and not compromised by
factors such as damage, pollutants, or deterioration.
Roof assemblies utilizing asphalt, asphaltic membranes,
polymeric membranes, and EPDM membranes will actually
exhibit improved fire resistance over time since they
February 2004 Interface • 25
lose combustible volatiles with aging. This
phenomenon has been demonstrated with
E 108 fire tests conducted by UL on
weathered low-sloped roof assembly specimens.
Underwriters Laboratories completed a
study where multiple sample assemblies
were prepared in its laboratory, tested
unaged, and tested after at least two years
of outside aging. The study revealed that
the fire performance of these assemblies
either remained unchanged or improved.
Recent attempts by roofing industry
groups to evaluate the fire performance
characteristics of roof assemblies by using
weathered test specimens not prepared in
the laboratory, but obtained from actual
field installations, have not been validated
through the ASTM consensus process.
Field Sampling
If the ASTM E 108 testing procedure
were to be used for field-obtained roofing
assembly testing, the sample selection
methodology would be critical. Specific
guidelines would need to be established to
ensure meaningful results. Careless sampling
will compromise the usefulness of
the data. Some examples of the guidelines
that would need to be followed are highlighted
• First and foremost, the existing
roofing assembly must have a documented
fire rating (Class A, B, or
C) as installed. Testing a
non-fire-rated assembly would
generate useless data since the
assembly, if not fire classified,
would not be expected to pass any
fire test.
• The entire roofing assembly must
be removed without disturbing the
membrane or its attachment method.
Disturbing the membrane
assembly (for example, by peeling
an adhered membrane off a substrate
and testing it in a loose-laid
condition) would again invalidate
the resultant data.
• The sample must be shipped as
removed and protected from damage
during transport. Removing a
membrane and folding it for shipment
would create wrinkles that
would influence burning characteristics.
Non-Folded Membrane
Folded Membrane
26 • Interface February 2004
• The exposed surface of the membrane must be
cleaned to remove dirt and other accumulated contaminants.
The cleaning products chosen and the thoroughness
of cleaning are key elements to ensure
meaningful results.
• The condition of the extracted sample is also vitally
important. Materials in a deteriorated condition will
not burn in the same manner as non-deteriorated
materials. Also, trapped moisture within the assembly
will influence burning characteristics.
• The mounting of the sample for testing must replicate
as nearly as possible the method used in the original
tests, otherwise valid comparative data will not be
• Testing should be conducted at the agency that performed
the original assembly test to ensure
The ASTM E 108 Standard provides a means for determining
the fire performance of roof assemblies. The procedure
was developed by Underwriters Laboratories in the early
1900s for testing new representative specimens in a laboratory
environment. Variations in test results are possible due
to variations in temperature and wind calibration ranges,
sample installation methods, sample preparation methods,
and interlaboratory
equipment and facilities.
Aged assembly testing, previously conducted by UL, supports
the physics-based logic that roof assemblies utilizing
asphalt, asphaltic membranes, and polymeric membranes
will actually exhibit improved fire resistance over time since
they will lose combustible volatiles with aging.
There is currently no concern within the fire community
in regard to life-safety concerns on in-place fire rated roof
assemblies, supporting the need for fire testing of aged field
samples. If an industry group insists on testing fieldobtained
samples for fire resistance, testing should not be
conducted unless recognized sampling and testing procedures
are developed and followed. ■
Kenneth D. Rhodes, PE, has been with
Underwriters Laboratories for 37 years,
where he has been involved with fire
testing and wind resistance testing of
building materials and roofing products.
He is a senior staff engineer at UL, and a
registered professional engineer in the
state of Illinois, having received a degree
in civil engineering from the University
of Illinois. Rhodes is a member of the
Society of Fire Protection Engineers.
Joseph Malpezzi has been employed
with Carlisle SynTec Incorporated for the
past 21 years. He has held various positions
in his career with Carlisle, but has
worked primarily in the technical and
research and development departments.
Joe has a BS in civil engineering from
Penn State University and is currently
the manager of code approvals for
Greg Brandt has been employed by
Firestone Building Products Company
for the last 19 years and currently holds
the position of product manager for
codes and standards, a position he has
held for the past three years. Prior to
attaining this position, he was the manager
for the Codes Department at
Firestone for 13 years. He has been a
member of BOCA, ICBO, and SBCCI,
and is currently a member of ICC. He is
also the technical committee chairperson
at SPRI, a member of ASTM E5, and
sits on STP panels at Underwriters Laboratories for UL 580, UL
790, UL 1256, and UL 1897.