Assessment of Thermal Bridging of Fasteners Through Insulated Roof Assemblies

16 • IIBEC Interface January 2024
Interface articles may cite trade, brand,
or product names to specify or describe
adequately materials, experimental
procedures, and/or equipment. In no
case does such identification imply
recommendation or endorsement by the
International Institute of Building Enclosure
Consultants (IIBEC).
Assessment of Thermal
Bridging of Fasteners through
Insulated Roof Assemblies
By Sarah Rentfro, PE; Georg Reichard, PhD,
PE; Elizabeth Grant, PhD, AIA; Jennifer
Keegan, AAIA; Eric Olson, PE; Cheryl
Saldanha, PE, CPHD; and Thomas J. Taylor
This paper was originally presented at the 2023
IIBEC International Convention and Trade Show.
FASTENERS THROUGH ROOFING assemblies,
composed of metal screws and plates, are
thermal bridges that bypass the thermal
insulation and create points of increased heat
flow. Figure 1 shows an example of this effect
on snow-covered roofs.
As energy code requirements for thermal
insulation have become more stringent,
quantifying that loss through the building
enclosure due to thermal bridging has become
more relevant. For this reason, some energy
codes and performance standards require
documenting thermal bridges and quantifying
their influence through detailed analysis. The
impacts of point thermal bridges (such as
fasteners) can be numerically simulated with
software tools; however, such simulations are
often time-consuming and sometimes need
laboratory tests as validation.
This study provides a relative comparison
of various roofing configurations with and
without fasteners. The authors compare the
thermal performance of a physical assembly,
tested under controlled laboratory conditions,
with a detailed three-dimensional (3-D)
computer simulation of the same assembly.
By incrementally increasing the complexity of
the assemblies in the tests and simulations,
the authors seek to better understand the
limitations of simulations, with the ultimate
goal of developing an experimentally validated
computer simulation approach that will enable
the evaluation of a broader range of roof
assemblies and roof fastener configurations.
Figure 1. Examples of thermal bridging at fasteners on snow-covered roofs.
Feature
January 2024 IIBEC Interface • 17
The authors have approached this in two
phases, the first of which is covered within
this paper. This preliminary phase focused
on the change in heat flux or flow (converted
into a thermal resistance or R-value) through a
series of test roof assemblies with and without
fasteners. The authors did not compare surface
or internal temperatures of the test assemblies
to computer simulations and so cannot yet
comment on the validity or accuracy of the
simulations and the experimental results. The
authors intend to publish a separate follow-up
study comparing the experimental data to twoand
three-dimensional computer simulations
to review the accuracy of computer modeling
methods commonly used for calculating and
accounting for thermal bridges in the design and
construction industry.
RELATED STUDIES
Several simulation studies, discussed in further
detail below, have estimated the thermal penalty
attributable to fasteners in roofing assemblies.
A Heat Transfer Analysis of Metal
Fasteners in Low-Slope Roofs
In an early finite-difference method simulation,
Burch et al.1 found an increase of 3% to 8%
in heat loss due to fasteners. They examined
fasteners modeled as cylinders with metal caps
in low-slope roofs with metal and wood decks at
a density of 0.5 fasteners/ft2 (5.4 fasteners/m2)
in insulation ranging from 1 to 6 in. (25 to 150
mm) thick. The increase in heat loss rose with
insulation thickness; the 8% increase in heat
loss corresponded to the assembly with 6 in. of
insulation. They found that burying fasteners
below the top layer of insulation reduced their
thermal effect to one-fourth that of the case
where fasteners penetrated both layers of
insulation. Further, fasteners had twice the effect
in metal decks as compared to wood decks, and
replacing the metal fastener caps with plastic
caps reduced thermal loss per fastener by 44%.
Effects of Mechanical Fasteners and
Gaps between Insulation Boards on
Thermal Performance of Low-Slope
Roofs
Petrie et al.2 conducted laboratory experiments
on roof assemblies incorporating three different
types of fasteners through two 2 in. (51 mm)
layers of polyisocyanurate (polyiso) insulation
and found that fasteners, on average, reduced
the thermal resistance of the roof assembly by
7% at a mean insulation temperature of 75°F
(24°C) compared to the same roof assembly
with no fasteners. Petrie et al.’s steady-state
simulation with HEATING 7 showed a 12%
reduction in roof R-value for fasteners with steel
plates, while only a 3% reduction in roof R-value
when using specially designed steel fasteners
with plastic heads extending through the top
layer of insulation. Findings were extended to
determine their impact on simulated heating
and cooling loads in six locations representing
varying climates in the US.
Roofing Research and Standards
Development: ASTM STP1590
In a simulation of a roof assembly with fastener
plates placed above the cover board and
fasteners penetrating the insulation into a
steel roof deck, Olson et al.3 found that while
the system nominally met the insulation
requirements of the International Energy
Conservation Code,4 it failed to meet these
requirements when the conductive effect of
industry-standard fasteners was considered.
Olson et al. showed that, in a simulated roof
assembly, the thermal penalty of fasteners in
a temperate climate may exceed that of other
penetrations such as roof drains, equipment
supports, and roof vents in a typical installation.
They explained that this was due to the
large number of fasteners as compared to a
typically smaller number of other, larger roof
penetrations.
Olson et al. used the 3-D, finite-difference
software HEAT3 to find a roughly 17% increase in
heat loss caused by fasteners, assuming exposed
metal plates over gypsum cover board at 1
fastener per 2 ft2 (0.2 m2) or 0.5 fasteners/ft2
(5.4 fasteners/m2).
Olson et al. explored the protective effect
of using an insulating cover board in lieu of a
gypsum cover board, and also the impact of
placing fasteners below the cover board and fully
adhering the cover board with adhesive. Their
simulations showed that even when adhering an
insulating cover board over fastened insulation,
there was still a 10% reduction in effective
thermal resistance compared to a roof assembly
with no fasteners.
Toward Codification of Energy
Losses from Fasteners on
Commercial Roofing Assemblies
& Development of Chi-Factors
Towards Codification of Thermal
Bridging in Low-Slope Roofing
Assemblies
Moletti and Baskaran5 and Moletti et al.6 tested
a range of common roofing assemblies in a
horizontal guarded hot box apparatus and
found that thermal bridging attributable to
roof fasteners increased with fastener density
and with increasing thermal resistance of the
insulation they penetrate. They reported a loss
in effective R-value ranging from 4.4% to 13.3%
across design assemblies rated R-26 through
R-36. They also found that covering the fastener
heads with the top layer of insulation led to 30%
to 70% reductions in thermal bridging compared
to fasteners extending through the cover board
and insulation, with more favorable results
derived from a thicker top layer of insulation.
Building Envelope Thermal Bridging
Guide V. 1.6
Developed by Morrison Hershfield7 and industry
partners, this guide includes a catalog of
common building enclosure details incorporating
thermal bridges. The reported values were
calculated using a 3-D finite-element analysis
(FEA) heat transfer software package developed
by Siemens PLM Software. The catalog includes
multiple roof details, two of which (10.1.9 and
10.1.13) incorporate exterior-insulated, low-slope,
mechanically fastened roof assemblies over
metal decks similar to (although not directly
comparable to) the assemblies included in this
study. Detail 10.1.9 includes a fastener density
of 0.3 fasteners/ft2 (3.4 fasteners/m2) with #10
and #14 fasteners embedded at different depths
of a roofing assembly with various insulation
thicknesses. Detail 10.1.13 includes a fastener
density of 1 fastener/ft2 (10.8 fasteners/m2)
with #14 fasteners through the entire depth of
the roofing assembly with various insulation
thicknesses.
Significance
As evidenced by the range of conclusions
garnered from these studies, more physical
experiments and computational simulations
addressing fastened roof components in their
various permutations are needed to understand
how thermal bridging from fasteners numerically
impacts the overall thermal performance
of roofing assemblies. These studies are
necessary to support design efforts and the
future development of building codes, industry
standards, and energy performance certifications.
Figures or tables containing data on the point
transmittance of roof fasteners, based on their
dimensional characteristics and the parameters of
the roof assemblies in which they are used, such
as those published within the Building Envelope
Thermal Bridging Guide7 by Morrison Hershfield,
allow a practical and simple way to estimate
overall thermal performance. These data could, in
turn, lead the roofing industry to develop more
thermally efficient assembly technologies.
18 • IIBEC Interface January 2024
Table 1. Naming protocol for study
Fastener Code Fastener Configuration
A No fastener
B #12 fastener, 6 in. long
Assembly Code Assembly Type
I Single 4 in. polyisocyanurate board
II 4 in. polyisocyanurate covered with 0.5 in. high-density polyisocyanurate
cover board
III 4 in. polyisocyanurate on steel deck
IV 4 in. polyisocyanurate on steel deck covered with 0.5 in. high-density
polyisocyanurate cover board
Abbreviations Full Term
PIR Polyisocyanurate board
HDB High-density polyisocyanurate cover board
SD Galvanized steel deck
Note: 1 in. = 25.4 mm.
Table 2. Assembly cases
Fastener
Code
Assembly
Code Case Assembly Components Diagram
A I A-I 4 in. PIR
A II A-II 0.5 in. HDB
4 in. PIR
B I B-I #12 fastener
4 in. PIR
B II B-II
0.5 in. HDB
#12 fastener
4 in. PIR
A III A-III 4 in. PIR
SD
A IV A-IV
0.5 in. HDB
4 in. PIR
SD
B III B-III
#12 fastener
4 in. PIR
SD
B IV B-IV
0.5 in. HDB
#12 fastener
4 in. PIR
SD
Note: HDB = high-density polyisocyanurate cover board; PIR = polyisocyanurate board; SD =
galvanized steel deck. 1 in. = 25.4 mm.
METHODOLOGY
The following section summarizes the
methodology used as a basis for this study
in both the physical experiment and the 3-D
computer simulation.
Study Setup
The roofing assembly builds in complexity, in the
stepwise fashion shown in Tables 1 and 2.
In the B-cases, a 6 in. (150 mm ) long #12
fastener penetrated the insulation layer and the
top flute of the galvanized steel deck (SD), (where
applicable). This resulted in an approximately
2 in. (51 mm) portion of the fastener that was
exposed below the top flute of the SD. Insulation
retention plates with a 3 in. (76 mm) diameter
and 0.019 in. (0.48 mm) thickness were used
with the fastener.
Physical Experiment
The authors tested the simplified roof assemblies
as depicted in Tables 1 and 2 in a controllable
climate test chamber. The climate chamber
configuration is shown in Fig. 2. Experimental
tests were conducted in triplicate series to
permit a baseline for statistical evaluation of
measurements.
The climate chamber consists of a warm
side (interior condition) chamber and a cold
side (exterior condition) chamber. It allows for
January 2024 IIBEC Interface • 19
controlling temperature and relative humidity
on both sides of the test assembly and
capturing temperature, relative humidity, heat
flux, and air velocity measurements as needed
depending on study requirements. The climate
chamber was customized with an assembly
frame to allow for horizontal mounting of the
test assembly to include gravitational impacts.
Since relatively small local heat flux differences
had to be assessed, a guarded meter box
approach was designed for these experiments.
Figure 3 shows an open view of the climate
chamber including the meter box within the
lower guard box (interior chamber).
All tests were conducted under steadystate
conditions and did not consider the
temperature dependence of the insulating
materials. A 2 × 2 ft (0.6 × 0.6 m) area of the
test assembly was monitored and the heat flux
across the test assembly was measured. The
exterior chamber was held at 50°F (10°C), and
the interior chamber was held at 100°F (38°C),
resulting in a mean insulation temperature of
75°F (24°C). For the I- and II-cases, the nearsurface
airflow was maintained at 50 ft/min
(0.25 m/s) in the exterior chamber and 70 ft/
min (0.36 m/s) in the interior chamber. For
the III-and IV-cases, the near-surface airflow
was maintained at 50 ft/min (0.25 m/s) in the
exterior chamber and 40 ft/min (0.20 m/s) in
the interior chamber. These velocities were used
to create a homogeneous condition across both
sides of the test assembly. Adding the SD in the
III- and IV-cases changed the airflow rate in the
interior chamber.
The test sequence was developed to
minimize the number of times the test chamber
needed to be opened and closed and the
samples manipulated, and to enable the same
4 in. (100 mm) polyiso board (PIR) specimen
to be used throughout an entire series of
tests, thereby eliminating variation in PIR as
a potential error source. The roof assemblies
studied in this analysis incorporated the
following modifications/simplifications from a
typical roofing assembly that may be observed
in the field (that is, on a construction site):
• The roofing membrane was omitted since
the membrane’s contribution to thermal
resistance is negligible and adhering a
membrane could introduce potential error
between assemblies.
• The adhesive layer (for example, low-rise
spray foam adhesive) was omitted between
the high-density cover board (HDB) and PIR
to facilitate removing the HDB between tests.
Foam weatherstripping tape was applied
to the top perimeter of the PIR (beneath
the HDB) to achieve an air seal between the
two layers, which resulted in an air gap of
approximately 0.1 in. (2.5 mm) between the
two layers.
• One layer of 4 in. (100 mm) PIR was used
in lieu of multiple PIR layers to avoid
discrepancies caused by imperfect contact
between the layers and between staggered
boards. These imperfections are not
considered in computational simulations
and are also difficult to replicate with each
test case. The 4 in. PIR does not meet current
prescriptive energy code requirements in
most of the continental US (for example, per
the 2021 International Energy Conservation
Code).8 However, 4 in. thick boards are
consistently produced and can therefore be
expected to have a reliable R-value.
• Foam weatherstripping tape was applied to
the top perimeter of the SD, beneath the PIR,
to air seal between the two layers, which
resulted in an air gap of approximately 0.19
in. (4.76 mm) between the two layers. Foam
flute plugs were also utilized at the open ends
of the metal deck.
The above-noted modifications were
included in the corresponding detailed 3-D
computer simulation (see next section).
Computer Simulation
The authors performed a detailed steady-state
thermal analysis of the same roof assemblies
tested in the physical experiment (see Table
2) using the 3-D FEA tool ANSYS, developed
by ANSYS, Inc. ANSYS simulates heat flow
through materials, components, and systems
based on a defined geometry and interior/
exterior environmental conditions, referred to
as boundary conditions.
Figure 2. Climate chamber configuration for roof system testing.
Figure 3. Open view of assembly frame and meter boxes (left) and guard box with climate
control (right) within the climate chamber.
20 • IIBEC Interface January 2024
Geometry
The finite-element method utilized in the
detailed ANSYS computer simulation allows for
a more accurate representation of the fastener
geometry than the finite-difference method
used in past research (reference Related Studies
above) since it can mesh irregular (that is, nonrectilinear)
shapes.
The model geometry (Fig. 4) was developed
as described in Tables 1 and 2, with the following
clarifications: The fastener manufacturer
provided a detailed 3-D SolidWorks model of the
fastener and fastener plate geometry, including
ribbed plate and fastener threads. Several minor
simplifications, which are considered to have
negligible impacts on the overall heat flow,
were made to the fastener threads and head to
facilitate meshing.
Material Properties
The authors utilized a two-dimensional (2-D)
FEA tool, THERM by the Lawrence Berkeley
National Laboratory (LBNL), to determine the
effective thermal conductivity of the small,
enclosed air cavities within the model (for
example, between the top of the insulation and
the bottom of the cover board) for input into
ANSYS.
Thermal conductivities and their sources for
the solid model components are as follows:
• High-density polyiso board (HDB): 0.017 Btu/
hr-ft-°F (0.029 W/m-K), from manufacturer’s
published product data
• Polyiso insulation board (PIR): 0.015 Btu/
hr-ft-°F (0.026 W/m-K), from manufacturer’s
published product data
• Fastener and plate (carbon steel): 29 Btu/hr-ft-
°F (50 W/m-K), from fastener manufacturer
• Galvanized steel deck (SD): 36 Btu/hr-ft-°F (62
W/m-K), from THERM material database
• Air cavities: vary, from THERM model
Boundary Conditions
Each case was modeled with steady-state
boundary conditions applied to the outermost
surfaces (Fig. 5). The side faces of the assembly
(that is, cut surfaces at the perimeter of the
assembly) were assigned as adiabatic boundary
conditions, which represent boundaries across
which there is no heat flow. Simulations were
based on 2 × 2 ft (0.6 × 0.6 m) assembly
dimensions.
The boundary conditions (indicated in Table
3) incorporate near-surface temperatures and air
flows and the emissivity of the adjacent visible
surfaces (that is, interior surfaces of the testing
chamber) measured in the physical experiment.
The authors matched the computer models’
boundary conditions to the experimental setup
rather than utilizing standard ASHRAE boundary
conditions, to eliminate a possible source of
difference between the experimental and
computer simulation results.
To calculate the convective film coefficient,
the authors followed the methodology for forced
convection (utilizing external flows over a flat
plate) outlined in chapter 4 of the 2017 ASHRAE
Handbook: Fundamentals.9 Properties of air were
obtained from papers by Baumgartner et al.10
and Kadoya et al.11 The convective film coefficient
does not incorporate natural convection as it is
expected that the size of the test chamber limits
the ability for natural convection to develop.
Simulation
The simulated heat flow in ANSYS was
converted into a U-factor (and associated R-value)
using the projected area of the assembly in the
horizontal (that is, projected-X) plane. Figure 6
shows the typical temperature output from
ANSYS.
RESULTS
The following section summarizes results from
both the physical experiment and the 3-D
computer simulation.
Figure 4. General geometry of three-dimensional simulated assembly configurations.
January 2024 IIBEC Interface • 21
Physical Experiment Results
Figure 7 shows the average calculated thermal
resistance R-values and the range of individual
test results from the three laboratory tests for
each test assembly configuration. Figure 8
shows the percent change from the A-case
R-values to the B-case R-values.
The cases with no fastener (A-cases) show the
following trends:
• Case A-I to A-II: The R-value increased by
10.8% when adding the cover board (HDB) to
the PIR.
• Case A-I to A-III: The R-value increased by
13.5% when adding the steel deck (SD) to the
PIR.
• Case A-I to A-IV: The R-value increased by
19.4% when adding the HDB and SD to the
PIR.
• Case A-II to A-IV: The R-value increased by
7.7% when adding the SD to the PIR and HDB.
• Case A-III to A-IV: The R-value increased by
5.2% when adding the HDB to the PIR and SD.
The cases with a #12 fastener (B-cases) show
the following trends:
• Case B-I to B-II: The R-value increased by
13.1% when adding the HDB to the PIR.
• Case B-I to B-III: The R-value increased by
5.4% when adding the SD to the PIR.
• Case B-I to B-IV: The R-value increased by
17.0% when adding the HDB and SD to the
PIR.
• Case B-II to B-IV: The R-value increased by
3.4% when adding the SD to the PIR and HDB.
• Case B-III to B-IV: The R-value increased by
11.0% when adding the HDB to the PIR and SD.
The cases with no fastener (A-cases) and the
cases with a #12 fastener (B-cases) show the
following trends relative to one another:
• A-cases to B-cases overall: Adding a #12
fastener in the B-cases reduced the thermal
resistance by a range of 2.2% (A-II to B-II)
to 11.0% (A-III to B-III) when compared to
the same condition in the A-cases with no
fastener.
• III-cases vs. I-cases: The III-cases with PIR and
an SD had a greater relative drop in thermal
resistance when the #12 fastener was added
(11.0%) compared to the I-cases with just PIR
(4.2%).
• IV-cases vs. II-cases: The IV-cases with PIR,
an HDB, and an SD also had a greater relative
drop in thermal resistance when the #12
fastener was added (6.1%) compared to the
II-cases with just PIR and an HDB (2.2%).
• II-cases vs. I-cases: The II-cases with PIR and
an HDB had a lesser relative drop in thermal
resistance when the #12 fastener was added
(2.2%) compared to the I-cases with just PIR
(4.2%).
• IV-cases vs. III-cases: The IV-cases with PIR,
an HDB, and an SD also had a lesser relative
drop in the thermal resistance when the #12
fastener was added (6.1%) compared to the IIIcases
with just PIR and SD (11%).
Computer Simulation Results
Figure 9 below shows calculated R-values from
the computer simulation of each test assembly
configuration, and Fig. 10 shows the percent
change from the A-case R-values to the B-case
R-values.
The cases with no fastener (A-cases) show the
following trends:
• Case A-I to A-II: The R-value increased by
10.2% when adding the HDB to the PIR.
• Case A-I to A-III: The R-value increased by
0.8% when adding the SD to the PIR.
Table 3. Boundary conditions used for computer simulation in ANSYS
Surface Temperature Convective Film Coefficient Emissivity
ºF ºC Btu/hr-ft2-ºF W/m2-K
Bottom, warm side
(I- and II-cases)
100 38 0.54 3.08 0.95
Top, cold side
(I- and II-cases)
50 10 0.44 2.49 0.95
Bottom, warm side
(III- and IV-cases)
100 38 0.41 2.33 0.95
Top, cold side
(III- and IV-cases)
50 10 0.44 2.49 0.95
Sides, adiabatic N/A N/A N/A N/A N/A
Note: N/A = not applicable.
Figure 5. Example boundary conditions for cases A-I (top) and B-I (bottom).
22 • IIBEC Interface January 2024
• Case A-I to A-IV: The R-value increased by
11.4% when adding the HDB and SD to the
PIR.
• Case A-II to A-IV: The R-value increased by
1.2% when adding the SD to the PIR and HDB.
• Case A-III to A-IV: The R-value increased by
10.5% when adding the HDB to the PIR and SD.
The cases with a #12 fastener (B-cases) show
the following trends:
• Case B-I to B-II: The R-value increased by
11.0% when adding the HDB to the PIR.
• Case B-I to B-III: The R-value decreased by
0.4% when adding the SD to the PIR.
• Case B-I to B-IV: The R-value increased by
10.5% when adding the HDB and SD to the
PIR.
• Case B-II to B-IV: The R-value decreased by
0.4% when adding the SD to the PIR and HDB.
• Case B-III to B-IV: The R-value increased by
11.0% when adding the HDB to the PIR and SD.
The cases with no fastener (A-cases) and the
cases with a #12 fastener (B-cases) show the
following relative trends to one another:
• A-cases to B-cases overall: Adding a #12
fastener in the B-cases reduced the thermal
resistance by a range of 2.7% (A-II to B-II) to
4.6% (A-III to B-III) when compared to the same
condition in the A-cases with no fastener.
• III-cases vs. I-cases: The III-cases with PIR
and SD had a greater relative drop in thermal
resistance when the #12 fastener was added
(4.6%) compared to the I-cases with just PIR
(3.4%).
• IV-cases vs. II-cases: The IV-cases with PIR,
an HDB, and an SD also had a greater relative
drop in thermal resistance when the #12
fastener was added (4.2%) compared to the
II-cases with just PIR and an HDB (2.7%).
• II-cases vs. I-cases: The II-cases with PIR
and an HDB had a lesser relative drop in the
thermal resistance when the #12 fastener was
added (2.7%) compared to the I-cases with just
PIR (3.4%).
• IV-cases vs. III-cases: The IV-cases with PIR,
an HDB, and an SD also had a lesser relative
drop in the thermal resistance when the #12
fastener was added (4.2%) compared to the IIIcases
with just PIR and SD (4.6%).
Comparative Results
Figure 11 shows calculated R-values from the
physical experiment compared to the computer
simulation for each test assembly configuration,
and Fig. 12 shows the percent change from the
A-cases’ R-values to the B-cases’ R-values for the
two procedures.
The two procedures show the following
notable differences for the cases with no fastener
(A-cases):
• Case A-I to A-III: The R-value increased by
13.5% in the physical testing when adding
the SD to the PIR, while the corresponding
R-value in the computer simulation stayed
relatively constant (0.8% increase).
• Case A-II to A-IV: The R-value increased
by 7.7.% in the physical testing when
adding the SD to the PIR and HDB, while
the corresponding R-value in the computer
simulation increased only slightly (1.2%
increase).
• Overall comparison: The difference
between the physical experiment results
and the computer simulation results varies
between 1.2% and 5.9% by case (utilizing the
physical experiment data as a baseline).
The two procedures show the following
notable differences for the cases with a #12
fastener (B-cases):
• Case B-I to B-III: The R-value increased by
5.4% in the physical testing when adding
the SD to the PIR, while the corresponding
Figure 6. Color temperature output for case B-IV at fastener (section and isometric views).
Figure 7. Physical experiment R-value for each assembly.
January 2024 IIBEC Interface • 23
R-value in the computer simulation stayed
relatively constant (0.4% decrease).
• Case B-II to B-IV: The R-value increased by
3.4.% in the physical testing when adding the
SD to the PIR and HDB, while the corresponding
R-value in the computer simulation stayed
relatively constant (0.4% decrease).
• Overall comparison: The difference between
the physical experiment results and the
computer simulation results varies between
0.8% and 6.7% by case (utilizing the physical
experiment data as a baseline).
Generally, the physical testing and computer
simulation show similar trends in the relative
change in thermal resistance between the A-cases
and B-cases (with the addition of a #12 fastener).
However, the two procedures show the following
notable differences:
• III-cases: In the assembly with the PIR and SD,
the physical testing showed a much greater
relative drop in the thermal resistance (11%)
compared to the computer simulation (4.6%).
• IV-cases: In the assembly with the PIR,
HDB, and SD, the physical testing showed a
somewhat greater relative drop in the thermal
resistance (6.1%) compared to the computer
simulation (4.2%).
DISCUSSION AND
CONCLUSIONS
In this section, the results from the previous
section are discussed in detail. The discussion is
divided into three sections: physical experiment
conclusions, computer simulation conclusions,
and conclusions related to the comparison
between computer simulation and experimental
results. A discussion on the comparison to past
research by others (that is, related studies) is also
included.
Physical Experiment Conclusions
The experimental results show that adding a
fastener reduces the thermal resistance of the
roofing assembly in all cases. By incrementally
adding layers, the results show the following:
1. The insulating effect of HDB:
• In cases without a fastener, adding the SD
to the PIR (A-III relative to A-I) has a better
R-value than adding the HDB alone (A-II
relative to A-I). However, in the cases with
a fastener, adding the HDB to the PIR (B-II
relative to B-I) is more effective than adding
the SD alone (B-III relative to B-I) because the
HDB reduces the thermal bridging from the
fastener.
• In cases with a fastener, in assemblies with
PIR and SD alone compared to PIR alone
(B-III relative to B-I), there is increased
radiant exchange to the interior because the
SD with a fastener acts as a radiator. Adding
the HDB (B-IV) insulates the fastener,
increasing the fastener and SD temperature
and reducing the radiant heat exchange to
the interior.
• In cases with a fastener, both cases that
include the HDB (B-II and B-IV) had a much
smaller drop in R-value relative to their
corresponding A cases than those without a
HDB (B-I and B-III).
2. The insulating effect of SD air spaces:
• In cases with and without a fastener, the
addition of an SD (A-III and B-III relative
to A-I and B-I, respectively, and A-IV and
B-IV relative to A-II and B-II, respectively)
increases the thermal resistance, which
is likely due to the enclosed air pockets
within the flutes, since trapped air is an
insulator.
• In the cases adding an SD to PIR without a
fastener (A-III), the SD with air pockets has
a higher R-value than HDB and PIR (A-II).
This trend is not the same when a fastener
is added, as B-III has a lower R-value than
B-II. The fastener may introduce enough
thermal bridging to counteract the benefit
of the insulating air spaces with this specific
configuration.
3. The impact of SD on thermal bridging:
Figure 8. Physical experiment R-value percent change from A-cases to B-cases.
Figure 9. Computer simulation R-value results for each assembly.
24 • IIBEC Interface January 2024
• The SD cases without an HDB had a greater
relative drop in thermal resistance compared
to their respective A-cases when fasteners
were added (B-III relative to A-III) than those
with PIR alone (B-I relative to A-I). This is
likely because the SD acts as a thermal
radiator.
• The SD cases had a greater relative drop
in thermal resistance compared to their
respective A-cases when fasteners were
added (B-III and B-IV relative to A-III and
A-IV, respectively) than did the cases without
an SD (B-I and B-II relative to A-I and A-II,
respectively).
In summary, the physical experiment results
demonstrate that a roof assembly with an HDB
adds insulating value from the board itself while
also reducing thermal bridging from the fastener.
Adding an SD also adds insulating value from the
enclosed air pockets, but it concurrently amplifies
the thermal bridging from fasteners.
It is worth noting, however, that various
aspects of the experimental setup proved difficult
to maintain and replicate, which likely impacted
the results to an extent (as indicated by the
variation of R-values across samples for each
assembly reported in Fig. 7). Additional testing
(that is, gathering of additional data points to
serve as the basis for a statistical analysis) needs
to be performed to evaluate potential outliers in
the dataset.
Computer Simulation Conclusions
The computer simulation results also show
that adding a fastener reduces the thermal
resistance of the roofing assembly in all cases. By
incrementally adding layers, the results show the
following:
1. The insulating effect of HDB:
• In cases without a fastener, adding the
HDB to the PIR (A-II relative to A-I) is more
effective than adding the SD alone (A-III
relative to A-I). The same trend can be seen
when a fastener is added (B-II relative to B-I
versus B-III relative to B-I). This diverges from
the experimental result trend for the same
cases.
• In cases with a fastener and PIR, with or
without SD alone (B-III relative to B-I), there
is minimal difference in R-values. Adding the
HDB (B-IV) insulates the fastener, increasing
the fastener and SD temperature and
reducing the radiant heat exchange to the
interior.
• In cases with a fastener, both cases that
include the HDB (B-II and B-IV) had a
smaller drop in R-value relative to their
Figure 10. Computer simulation R-value percent change from A-cases to B-cases.
Figure 11. Comparative R-value results.
January 2024 IIBEC Interface • 25
corresponding A-cases than those without
an HDB (B-I and B-III).
2. The insulating effect of SD air spaces:
• Adding the SD changes the R-value
minimally with or without a fastener (A-III
relative to A-I, A-IV relative to A-II, B-III
relative to B-I, and B-IV relative to B-II).
This indicates that air space modeling
and contact resistance between layers
requires further study. Note that the metal
deck and insulation were not modeled
in contact with one another since the
physical experiment incorporated a small
(0.19 in. [4.76 mm]) air gap between the
two layers.
3. The impact of SD on thermal bridging:
• The SD cases without an HDB had a
greater relative drop in thermal resistance
compared to their respective A-cases
when fasteners were added (B-III relative
to A-III) than those with insulation alone
(B-I relative to A-I). This may be because
the SD acted as a thermal radiator.
• The SD cases had a greater relative drop
in thermal resistance compared to their
respective A-cases when fasteners were
added (B-III and B-IV relative to A-III and
A-IV, respectively) than did the cases
without SD (B-I and B-II relative to A-I and
A-II, respectively).
In summary, similar to the physical
experiment, the computer simulation results
demonstrate that a roof assembly with an HDB
adds insulating value from the board itself
while also reducing thermal bridging from
the fastener. Also, adding an SD amplifies the
thermal bridging from fasteners. In contrast
to the physical experiment, the computer
simulation demonstrates, perhaps incorrectly,
that adding an SD has minimal impact
on overall thermal resistance rather than
increasing the thermal resistance, indicating
that the way the models account for air spaces
should be further reviewed.
Comparison of the Results
of Physical Experiments and
Computer Simulations
When comparing the results of the physical
experiments and computer simulations on a
case-by-case basis, the difference between
them ranges from 0.8 to 6.7%. The authors
intend, through ongoing work, to review the
correlations in more detail. As shown in the
discussion above, some trends observed by
both approaches were similar. The diverging
trends that warrant further review include the
following:
• The trends when adding the SD do not
match. This may indicate that assumptions
with air space modeling and contact
resistances are inaccurate. Computer models
assume each layer is in perfect contact.
• The nature of the physical experiment
introduces a potential for outliers; however,
it is difficult to perform statistical analyses
on small sample sizes.
General Comparisons to Past Work
On the experimental side, Moletti et al.6 reported
a 4.4% decrease in effective thermal resistance for
an R-26 system on a steel deck with #14 fasteners
penetrating a fiberglass mat gypsum roof cover
board and two layers of insulation at a density of
0.25 fasteners/ft2 2.69 fasteners/m2. While not
an exact match, case B-III in this study (R-23.6
insulation on steel deck with #12 fasteners at
a density of 0.25 fasteners/ft2 [2.69 fasteners/
m2] and no cover board) is similar and showed
an 11.0% R-value reduction in the physical
experiments.
On the simulation side, the results from
the present study can be compared in general
terms to Olson et al.’s finite-difference method
simulation.3 Case B-IV in the present study is the
most similar, with Olson et al.’s study utilizing
a simplified representation of the metal deck,
fastener, and fastener plate and including an
additional gypsum substrate board between the
metal deck and 4.5 in. (110 mm) thick PIR. With
a #12 fastener head and plate buried below the
HDB cover board, at the same fastener density
as considered in this study, Olson et al. found a
5.9% reduction in effective R-value compared to
an assembly with no fasteners. This can be loosely
compared to the 4.2% R-value reduction found in
this study for case B-IV relative to case A-IV.
General Conclusions
The authors conducted physical experiments and
computer simulations in a stepwise fashion to
isolate the influence of the different layers in the
Figure 12. Comparative percent change from A-cases to B-cases.
26 • IIBEC Interface January 2024
assembly and to see where physical modeling
and computer simulation converge and diverge.
Both physical experimentation and computer
simulation are simplifications of reality, and
there are errors inherent in both approaches.
The results of this study identify some diverging
trends that warrant further analysis. The value of
computer simulation, once validated by physical
experimentation, is its ability to quickly extend
results to a wide range of possible scenarios.
CONTINUATION
The next steps of this study include a review of
the temperatures at different locations of each
test assembly and the determination of point
transmittances for the roof fasteners. With these
data, the authors can determine more precisely
where the computer simulations are diverging
from the physical experiment. A sensitivity
analysis can also be performed to determine the
relative impact of air space modeling and the
effect of contact resistances on the computer
simulation results. Future work may include
performing additional physical testing to produce
a statistically significant sample size.
The authors also intend to perform simplified
2-D FEA evaluations, which are often used
by practitioners for determining overall roof
assembly thermal performance (due to increased
efficiency and overall lower cost to perform
the analysis), to review the potential negative
consequences inherent with analyzing a 3-D
problem in two dimensions.
ACKNOWLEDGMENTS
The authors would like to acknowledge the RCIIIBEC
Foundation for supporting and funding this
study, OMG Roofing Products for contributing
physical fasteners and plates and associated 3-D
geometry of these materials for the computer
simulation, and GAF for supplying polyiso and
high-density polyiso boards.
REFERENCES
1. Burch, D. M., Shoback, P. J., and Cavanaugh, K.
1987. “A Heat Transfer Analysis of Metal Fasteners
in Low-Slope Roofs.” ASTM International. West
Conshohocken, PA: ASTM International.
2. Petrie, T. W., Atchley, J. A., Desjarlais, A. O., and
Christian, J. E. 2000. “Effects of Mechanical
Fasteners and Gaps between Insulation Boards on
Thermal Performance of Low-Slope Roofs.” Journal
of Thermal Envelope and Building Science. Vol. 23,
No. 4. pp. 292-317.
3. Olson, E. K., Saldanha, C. M., & Hsu, J. W. 2015.
Roofing Research and Standards Development:
ASTM STP1590. West Conshohocken, PA: ASTM
International.
4. International Code Council (ICC). 2012. International
Energy Conservation Code. Brea, CA, USA: ICC.
5. Molleti, S. and Baskaran, B. 2020. “Towards
Codification of Energy Losses from Fasteners on
Commercial Roofing Assemblies.” Interface. IIBEC.
Raleigh, NC: pp. 14-24.
6. Moletti, W., van Reenan, D., & Baskaran, B. 2021.
“Development of Chi-Factors Towards Codification of
Thermal Bridging in Low-Slope Roofing Assemblies.”
Energy & Buildings. 231, 110559. https://doi.
org/10.1016/j.enbuild.2020.110559.
7. Morrison Hershfield Limited. 2021. Building Envelope
Thermal Bridging Design Guide v. 1.6. Vancouver, BC:
BC Hydro Power Smart.
8. ICC.
9. American Society of Heating Refrigeration and Air
Conditioning Engineers (ASHRAE). 2017. 2017 ASHRAE
Handbook: Fundamentals I-P and SI Editions. Atlanta,
GA: ASHRAE.
10. Baumgartner, M., R. Weigel, A. Harvey, F. Plöger,
U. Achatz, and P. Spichtinger. 2020. “Reappraising
the Appropriate Use of a Common Meteorological
Quantity: Potential Temperature.” Atmospheric
Chemistry and Physics. Vol. 20.
11. Kadoya, K., N. Matsunaga, and A. Nagashima. 1985.
“Viscosity and Thermal Conductivity of Dry Air in the
Gaseous Phase.” Journal of Physical and Chemical
Reference Data. Vol. 14, No. 4.
ABOUT THE AUTHORS
SARAH RENTRFO, PE
Sarah Rentfro, PE, is a
senior consulting
engineer in Simpson
Gumpertz & Heger Inc.’s
(SGH) Building
Technology group. She
specializes in the design
and integration of
complex building
enclosure systems
including roofing,
waterproofing, air/water
barriers, cladding assemblies, and fenestration
with an emphasis on performance efficiency and
constructability. She is also actively engaged in
SGH’s Building Science practice group and has
experience using various modeling software tools
to simulate heat, air, and moisture flow through
building enclosure systems. Her work includes
new design consulting, building science modeling,
construction administration, and investigation and
rehabilitation of existing building enclosures.
GEORG REICHARD,
PHD, PE
Georg Reichard, PhD,
PE, is a professor and
department head of
building construction and
an associate director in
the Myers-Lawson School
of Construction at
Virginia Tech in
Blacksburg, VA. His
research deals with
experimental and
numerical methods,
simulation, and data modeling, in particular in
the area of building sciences related to building
enclosures and environmental systems. In his
current research, he focuses on building
performance, enclosure durability, disaster
resilience, energy efficiency, and integrated
decision-making for retrofit solutions in
connection with different control strategies and
building materials. Reichard holds a master’s
and a doctoral degree in civil engineering from
Graz University of Technology, Austria.
ELIZABETH GRANT,
PhD, AIA
Elizabeth Grant, PhD,
AIA, is the building &
roofing science research
lead at GAF. She
supports GAF’s efforts
within the commercial
roofing community
through engagement
with design
professionals, providing
technical guidance in
their design and
specification processes. She serves on the board
of directors of the IIBEC Virginia Chapter and is a
member of AIA, RICOWI, and NWIR. Before
joining GAF, she was an associate professor at
Virginia Tech’s School of Architecture + Design,
publishing papers, conducting studies, and
offering courses in architectural design,
environmental design research, and
environmental building systems. She focuses on
the building enclosure and sustainable solutions
to architectural and environmental problems.
JENNIFER KEEGAN,
AAIA
Jennifer Keegan, AAIA
is the director of
Building and Roofing
Science for GAF,
focusing on overall roof
system design and
performance. She has
over 20 years of
experience as a
building enclosure
consultant specializing
in building forensics,
assessment, design, and remediation of building
enclosure systems. Keegan provides technical
leadership within the industry as the chair of the
ASTM D08.22 Roofing and Waterproofing
Subcommittee and the IIBEC Education
Committee chair, and as an advocate for women
within the industry as an executive board
member of National Women in Roofing and a
board member of Women in Construction.
January 2024 IIBEC Interface • 27
ERIC OLSON, PE
Eric Olson, PE, is a
principal in Simpson
Gumpertz & Heger
Inc.’s (SGH’s) building
technology group,
based in Houston, TX.
He specializes in
evaluating and
investigating building
enclosures, including
windows and
curtainwalls, cladding
and veneer systems, roofing, and plaza and
below-grade waterproofing. Olson provides
design consulting services related to new
building construction and building
rehabilitation projects. He also specializes in
forensic evaluations involving hurricane and
storm-related damage assessments, design and
construction claims, and litigation related to
these matters. Olson has written papers and
presented on several building enclosure–
related topics, including thermal bridging
through roofing systems.
CHERYL SALDANHA,
PE, CPHD
Cheryl Saldanha, PE,
CPHD leads Simpson
Gumpertz & Heger
Inc.’s Building Science
and Passive House
Practices and
specializes in
designing and
evaluating building
enclosures for new
construction projects
and existing building
enclosure renovations. Her experience includes
curtainwall rainscreen facade, roofing, and
waterproofing systems on a range of building
types. Saldanha is adept at using multiple
computer software packages to simulate
building systems and details for thermal,
condensation, whole-building energy, and
daylighting analysis. She has co-chaired the
New York City Chapter of the International
Building Performance Simulation Association
and participated on the NYC Commercial Energy
Code Technical Advisory Committee for the last
two code cycles. Saldanha was awarded
Building Design + Construction magazine’s Top
40 under 40 for 2022.
Please address reader comments to
chamaker@iibec.org, including
“Letter to Editor” in the subject line,
or IIBEC, IIBEC Interface,
434 Fayetteville St., Suite 2400,
Raleigh, NC 27601.
THOMAS J. TAYLOR
Thomas J. Taylor has
over 25 years’
experience in the
building products
industry, all working
for manufacturing
organizations in a
variety of new product
development roles. He
is now a consultant,
focused on single ply
membranes and
polyiso insulation. He received his PhD in
chemistry and holds approximately 35 patents.
Publish in IIBEC Interface is seeking submissions for the following
issues. Optimum article size is 2,000 to 3,000 words,
containing five to ten high-resolution graphics. Articles may
serve commercial interests but should not promote specific
products. Articles on subjects that do not fit any given
theme may be submitted at any time.
XXX
xxx
FP
colour
p. 1
Waterproofing
Challenges in
Hydrostatic Conditions
The Technical
Journal of the
International
Institute of
Building Enclosure
Consultants
APRIL 2023 | Vol XLI No. 4 | $15.00
Leak Diagnosis
ISSUE SUBJECT SUBMISSION DEADLINE
May/June 2024 IIBEC International Convention January 15, 2024
and Trade Show
July/August 2024 Restoration/Forensics March 15, 2024
September 2024 Climate Adaptation May 15, 2024
The Technical
Journal of the
International
Institute of
Building Enclosure
Consultants
MARCH 2023 | Vol XLI No. 3 | $15.00
Energy Issues
Practical Considerations
for Whole-Building
Air-Leakage Testing