1

Wind-Driven Rain Resistant Window Installation Options for Hurricane-Prone Areas

Cory T. Salzano
1
, J ason T. Smith
2
, Forrest J . Masters
3
, and J ames D. Katsaros
4

1
Graduate Research Assistant. Department of Civil and Coastal Engineering,
University of Florida, 365 Weil Hall, Gainesville, Florida, 32611, csalzano@ufl.edu
2
Graduate Research Assistant. Department of Civil and Coastal Engineering,
University of Florida, 365 Weil Hall, Gainesville, Florida, 32611, jasonsmith@ufl.edu
3
Corresponding Author. Assistant Professor of Civil and Coastal Engineering, University of Florida, 365 Weil
Hall, Gainesville, Florida, 32611, masters@ce.ufl.edu, Phone: (352) 392-9537 ext. 1505, Fax: (352) 392-3394
4
Flashing Systems Development Leader, DuPont BuildingInnovations, 5401 Jefferson Davis Highway,
Richmond, VA 23234, James.D.Katsaros@usa.dupont.com

Abstract Water ingress is one of the most critical, recurring issues during hurricane impacts.
Although most homes and businesses survive structurally, a significant number experience
enough rain penetration to cause damage to the interior and a loss of building contents. This
issue came to the forefront after the 2004 hurricane season, when it was clearly shown that
homes built to the new Florida Building Code standards suffered minimal physical damage, yet
incurred substantial insured losses due to water intrusion. This paper provides an overview of
experimental research conducted to investigate the performance of various window installation
options for single family homes in extreme wind-driven rain conditions. Static, pulsating as well
as amplitude- and frequency-modulated sinusoidal pressure load sequences were applied under
simulated wind-driven rain conditions to 18 wall assemblies that uniquely varied by their
fenestration, installation methodology, wall system construction, and exterior finish.

Keywords: Fenestration, Installation, Hurricane, Water Barrier, Drainage Method
1. Introduction
Current water penetration resistance test methods (e.g., ASTM E331-00, AS/NZS 4284:1995,
J IS A 1517) for fenestration evaluate products in isolation and not as a component integrated into
a wall system. Thus the efficacy of the various window installation techniques employed in the
field is not well understood, and less is known about the leakage paths originating from the wall
or the window-wall interface. Moreover, it is unclear how effective these water penetration
control strategies work in the extreme buffeting loads associated with the turbulent wind action
in a surface layer hurricane wind field.
The objective of this study is to determine which installation guidelines, as well as particular
details, are most effective at preventing water intrusion and/or directing water out of the building
assembly. Eighteen wall specimens were constructed using either the traditional exterior barrier
interface prescribed in Method A1 of ASTM E2112 or the drainable installation practices
detailed in the FMA/AAMA 100-07 and the draft FMA/AAMA 200 Standard Practices. A
unique combination of fenestration, installation and finished wall system comprised each wall.
Section 2 explains the difference between the installation approaches, and Section 3 provides
details on the wall design and construction. The subject walls were wetted and subjected to four
rounds of pressure loading. Static, pulsating as well as amplitude- and frequency-modulated
sinusoidal pressure load sequences were applied, followed by a repeat of the static test. Details
may be found in Section 4. Results from this study are currently being vetted by the research
oversight committee. Information on how to locate the results from this study is provided in
Section 5.
2

2. Window Installation: Water Barrier vs. Drainage Method
Water penetration performance of windows in residential construction is highly dependent on
the moisture control strategy used to integrate the window into the water-shedding surface of the
surrounding building envelope. Window installation methods can be broken up into two
approaches based on these moisture control strategies. The first is the water barrier method,
which seeks to prevent water migration through the external interface of the window and the wall
system by creating a water shedding barrier that is coincident with the exterior moisture barrier
and air barrier (RDH 2002). The second approach is the drainage method, which places the water
shedding barrier inside the interface to redirect any leakage to the drainage plane of the wall
(Katsaros & Hardman 2007, Lstiburek 2008).
In the traditional water barrier method installations, such as those outlined in ASTM E2112
prior to the most recent revision, a continuous bead of sealant is applied to the interior surface of
the mounting fin/flange of the window to provide a moisture and air barrier at the external
interface of the window opening. While this method is currently considered common practice for
the installation of fenestration products, several shortcomings have been identified in the
literature (e.g., Katsaros and Hardman 2007, Lstiburek 2008). This installation technique does
not make provisions to control leakage that may occur during the windows service life nor does
it account for initial construction defects. Its effectiveness is defined by the ability of either an
external wall cladding system or a concealed barrier to prevent water ingress into the building.
In drainage method installations, such as FMA/AAMA 100-07, the interior perimeter of the
window opening is sealed at the head and jambs, but gaps are left at the sill in the exterior seal
located between the mounting flange and the sill pan flashing. In contrast to the exterior water
barrier method, the drainage method creates a moisture and air barrier at the interior interface
between the fenestration product and the rough opening. Water that leaks into the opening
through the window or the adjacent wall assembly is blocked from infiltrating to the interior of
the building at this barrier and is drained through the openings at the sill to the water-resistive
barrier of the wall. The drainage method assumes that the exterior cladding of the wall assembly
is not the only moisture barrier of the wall system and takes a practical approach to handling any
water that may seep into the opening. For this reason, the most recent version of ASTM E2112
(2007) recommends the use of pan flashings, which constitute a drainable installation, for all
windows and doors.
3. Specimen Design and Construction
The test matrix consisted of a representative sample of common residential wall assemblies
from the coastal southeastern United States. A total of ten wood frame and eight concrete
masonry unit (CMU) wall sections were constructed. Each wall measured 2.44 m by 2.44 m (8 ft
by 8 ft) and featured a unique combination of wall type, exterior finish and window interface.
The window interface was determined by the installation approach, which followed either an
exterior water barrier or drainage method. A matrix of the wall combinations used for each
specimen is shown in Table 1. In order to isolate the leakage paths of the window interface, the
test matrix initially consisted of fixed windows only. It was anticipated that the operable
components of the window specimens would begin to leak at 15% of their design pressure rating,
which is the current requirement stipulated in most building codes and standards (e.g.,
AAMA/WDMA/CSA 101/I.S.2/A440). A companion conference proceeding authored by Lopez
et al. provides more detail on the performance of operable windows. Logistical issues later
necessitated the need to allow the incorporation of two awning windows and a single hung
3

window into this project. These substitutions did not interfere with any of the results or testing
procedures, however.

Table 1. Test Specimen Matrix. (CMU walls are designated by the shaded rows)

Wood Frame Walls. The wood wall specimens were framed using 2x4 stud members at
40.64 cm (16 in) o.c. and sheathed with 1.11 cm (7/16 in) oriented strand board. The window
openings were framed with a double 2x4 sill plate and a double 2x8 header. The bottom plates of
the wall specimens were affixed with a steel C4x7.5 channel to facilitate the transport of the
specimens and the attachment to the testing rigs. Bolts with plate washers were fastened through
the channel at 70 cm (24 in) o.c. to secure the wall to the channel in order to simulate a typical
slab-on-grade connection.
A polymer based house wrap material was installed on all wood wall sections to serve as the
water-resistive barrier for the wall assembly. The windows were then installed by certified
window installers and flashed. For exterior water barrier method installations, the flashing
sequence of ASTM E2112 Method A1 was followed, which is the basis of most window
manufacturers installation instructions. Self-adhered extendable sill pan flashing was also
installed in these wall openings, which although not specified in the ASTM standard for exterior
barrier installations, provides additional protection to the wood sill area.
The FMA/AAMA 100-07 Standard Practice was used as the basis for the drainage method
installations. This standard has the option to use 10.16 cm (4 in) self-adhered flashing or a 22.86
cm (9 in) mechanically attached flashing to flash the exterior jambs and head of the window and
allows the use of a low expansion foam or backer rod and gunnable sealant to provide the
4

interior moisture and air seal. Different combinations of these installation options were tested.
All sealants used in these installations met the requirements of either or both ASTM C920-05
and AAMA 808.3-05 for gunnable sealants or AAMA 812-04 for low expansion foams.

After the windows were installed, the walls were finished with either fiber cement board or
stucco. The fiber cement board was attached by a professional installer in accordance with the
manufacturers installation instructions. The fiber cement siding planks were blind nailed to
wood studs using siding nails (2.28 mm shank x 5.61 mm HD x 5.08 cm long) installed 9.53 mm
(3/8 in) from the edge of the planks and 2.54 cm (1 in) from the top. A minimum 3.18 cm (1
in) overlap was maintained between planks. The windows, as well as the top and the sides of the
walls were framed with fiber cement trim pieces to provide a clean finish on all the fiber cement
siding planks. Horizontal flashing was provided at the head of each window.
The stucco was applied over lath in a three coat application consisting of a scratch coat,
brown coat, and finish coat in compliance with ASTM C926-06 and ASTM C1063-06.
Expansion joints were placed at the four corners of the window on all wood wall specimens
receiving stucco to allow for the proper movement of the stucco while curing. It was necessary
to provide 10.16 cm (4 in) decorative banding around the windows to accommodate the large
projection from the integral mounting fin. It should be noted that these bands are purely
decorative and were installed onto a scratched surface around the window after the proper stucco
application had already been applied.
The stucco was allowed a minimum of 7 days cure time to reach a pH level below 13 before
the application of paint to prevent alkali burn. All wall sections, fiber cement board and stucco,
received a three coat paint finish. The first coat was an alkali resistant primer applied to a
thickness of 2 mil. The second and third coat was a construction grade acrylic paint with a slight
tint applied to a thickness of 4 mil wet. A minimum cure time of three hours was given between
coats and the walls were not tested until a minimum of 10 days after the finish coat.
Masonry Walls. The CMU walls were constructed on-site in accordance the requirements of
Sections 2104.1.1 through 2104.5 of the 2004 Florida Building Code and ACI 530.1-05/ASCE 6-
05/TMS 602-05 A licensed, professional masonry contractor built the walls directly on top of a
20.32 cm (8 in) steel MC8 X 21.4 channel. Number 5 rebar was welded to the base channel prior
to construction to provide the vertical reinforcing steel for the wall as well as to prevent the wall
sections from overturning while moving them between testing sites. The wall specimens were
constructed from normal weight 20.32 cm (8 in) CMUs (ASTM C90-01a) in a typical running
bond with a Type S mortar (ASTM C270-01a). Down-cells on both sides of the window, and at
both sides of the wall were filled with a coarse grout (ASTM C476-01) in each wall specimen to
provide flexural rigidity. Bond beams were poured at the top and the base of the wall. To enable
the viewing of any water that was leaking into the wall system without compromising the
window and wall interface; the block faces on the interior side of the wall above the poured base
beam were removed, leaving open gaps on the interior block surface.
The rough openings and windows were sized to accommodate 1.91 cm ( in) pressure
treated wood bucking. Due to the slight variations in the construction and window sizes, some
flexibility was allowed; a maximum of 2.54 cm (1 in) bucking was accepted. Bucking was not
required at the sill of the opening because the CMU walls were constructed with a precast
concrete sill. Two precast sill designs were utilized: some sills were flush with the exterior block
surface (flush sill) and others featured a protruding edge for enhanced drainage (face sill).
The windows were installed following the guidelines of ASTM E 2112 for the traditional
exterior barrier installations, or the draft FMA/AAMA 200, which details a modified exterior
5

barrier installation as well as a drainage installation. Under the barrier installation provided by
the draft FMA/AAMA 200 standard, a continuous exterior seal between the window and the wall
system is required as well as an interior seal around the entire window perimeter. The interior
seal may be achieved using either a backer rod and gunnable sealant or a low expansion foam,
per the specifications described above for the wood wall systems. Both of these options were
tested. The drainage method described in this standard follows the same installation
procedures as the barrier method with the exception that a discontinuous exterior seal at the sill
is provided. This discontinuous seal may be accomplished with 5.08 cm (2 in) voids in the
sealant near the ends of the sill, a weep screed or another wicking mechanism.
The most distinguishable characteristic in the draft FMA/AAMA 200 standard is the
application of a liquid applied flashing on the CMU in the pre-installation procedures. The liquid
applied flashing is a breathable waterproofing material that is applied onto the CMU wall 22.86
cm (9 in) around the perimeter and in the return of the rough opening to enhance the water
resistance of the masonry wall system. The adhesion of a stucco application applied directly to
the liquid applied flashing is of some concern; therefore, three different application methods for
the liquid applied flashing were tested on the six specimens that follow the draft FMA/AAMA
200 installation method. On two specimens, a non-emulsifiable bonding agent was applied to the
liquid applied flashing treated area to promote adhesion of the stucco. In two other specimens, a
wire lath was applied over the liquid applied flashing treated area before the application of
stucco. The remaining two FMA/AAMA 200 walls were treated with the liquid applied flashing
on the interior surface of the wall surrounding the window rough opening, including the rough
opening return, and down to the base of the wall.
After the windows were installed into the rough openings of the CMU walls, the exterior
finishes were applied. All the wall sections received either a 2.22 cm (7/8 in) three coat stucco
application or a decorative cementious coating (DCC) consisting of a stucco skim coat applied
directly over the block with a thickness that varied from a paint thickness to approximately 6.35
mm (1/4 in). The stucco was allowed to cure under moist ambient conditions until the
appropriate pH level was achieved to accept paint. The paint was applied in the same three coat
application as the wood walls and allowed to dry for a minimum of ten days before the wall
specimens were tested.
4. Experimental Procedure
In order to test the effectiveness of the exterior barrier method and the drainage method
approaches to the installation of fenestration products, each window-wall specimen was cycled
through a series of four tests. In order, specimens were evaluated under a static test, a pulsating
test, a dynamic test, and then re-evaluated under the initial static test. Wind driven rain was
simulated in all three test in accordance with Section 9 of ASTM E 331-00. A minimum of two
days was allotted between each round of testing to allow for adequate drying of the specimen
before being subjected to the next test.
The static test was derived from Procedure A in ASTM E 1105-00. Each specimen was
loaded using a pressure chamber with an initial pressure of 137 Pa (2.86 psf, per ASTM E 331-
00) for five minutes, and then the pressure was linearly increased over 15 minutes to half of the
windows design pressure (DP) rating.
The pulsating test was derived from a combination of ASTM E 2268 and AAMA draft of
the Voluntary Specification for Rating the Severe Wind Driven Rain Resistance of Windows,
Doors and Unit Skylights. The pulsating test utilized the same testing chamber used for the static
test. ASTM E 2268 defines a median pressure of 137 Pa (2.86 psf) in which specimens are to be
6

pulsated at a pressure difference equal to 50% and 150% of the median test pressure. The
AAMA 200 specification provides a range of performance levels from 1 to 10, where level one
pulsates from 239 to 718 Pa (5 to 15 psf) and level ten pulsates from 670 to 2011 Pa (14 to 42
psf). For the specimens tested in this study, a pulsating pressure schedule was developed with
ranges lower than specified in AAMA draft specification (to accommodate for ASTM E 2268)
up to the highest performance level whose upper pressure was half the manufacturers design
pressure. For example, a wall system with a DP60 window was loaded to the performance level
that pulsed from 479 to 1436 Pa (10 to 30 psf). Each test consisted of the varying pressure levels
cycling 60 times every 1.5 second continuously.
The dynamic test consisted of the specimen being loaded into a mock wood frame house to
simulate flow characteristics, as well as realistic roof-to-wall and base connections, for common
residential housing. Extreme wind events were created with the implementation of the University
of Floridas Hurricane Simulator.
UF has constructed a 2800 hp hurricane simulator capable of replicating turbulent wind and
rain loads on a full-size low-rise structure (shown in Figure 1). Four 700 hp engines spin eight
hydraulic actuated vaneaxial fans to produce 1676+Pa (35 psf) stagnation pressures. Air passes
through specially designed venture inlets forcing the air to travel perpendicular to the fan disc for
maximum efficiency. Air accelerates through the contraction and passes through a series of
custom designed neutral shape NACA airfoils designed to discharge water at the trailing edge to
simulate wind-driven rain. The airfoils are connected to a hydraulic rotary actuator, which
creates the changes in wind directionality. To recreate hurricane conditions an active computer
control system modulates wind speed by varying fan RPM, creates directional effects by
articulating the airfoils at the exit, and injects water into the flow field to simulate rain. The
control system utilizes a fast running PID-control operated in the Labview 8.5 environment.

Figure 1. University of Florida Hurricane Simulator
The loads were designed using 10 minute wind speed observations collected by the Florida
Coastal Monitoring Program that were converted to velocity pressures. It was conservatively
assumed that there was perfect aerodynamic admittance between the free stream velocity
pressure and the stagnation pressure on the windward wall. Records with a mean velocity >20
m/s were extracted and detrended. The longitudinal velocity component was calculated and
passed through nine bandpass filters in 0.1 Hz passband increments. The peak amplitude for each
passband was recorded and divided by the records 10 minute mean velocity to get a peak
amplitude / mean ratio. Data was stratified into three turbulence intensity regimes, of which the
middle turbulence range (0.15 0.20) was used. The 50
th
percentile peak values were employed
to construct a sinusoidal loading pattern at three different velocity thresholds that correspond to
239, 479, and 718 Pa (5, 10 and 15 psf).
7

5. Presentation of Results and Conclusions
Results have been compiled but cannot be released until the allotted presentation time at the
conference. This research was conducted in collaboration with and under the oversight of an
industry and government task force that consists of architecture, engineering, homebuilding,
product manufacturing, insurance, product certification and code development interests. The
results and the conclusions of the study will be reviewed on August 18, 2008, and presented
publicly for the first time at this conference. Supplemental information will be provided at the
event. Readers are also welcome to contact the corresponding author for this information.
6. References
AAMA/WDMA/CSA 101/I.S.2/A440 American Architectural Manufacture Association Standard specification for
windows, doors and unit skylights, 2005.
AAMA 808.3-92 American Architectural Manufacture Association Voluntary Specification for Exterior Perimeter
Sealing Compound, 1992.
AAMA XXX-XX (Draft) Voluntary specification for rating the severe wind driven rain resistance of windows
doors and unit skylights, American Architectural Manufacture Association, 1827 Walden office square, Suite
550, Schaumburg, Illinois 60173-4268.
ASTM C920-05 American Society for Testing and Materials, Standard Specification for Elastomeric J oint
Sealants, 2005.
ASTM C1063-06 American Society for Testing and Materials Standard Specification for Installation of Lathing
and Furring to Receive Interior and Exterior Portland Cement-Based Plaster 2006.
ASTM E331-00 American Society for Testing and Materials Standard Test Method for Water Penetration of
Exterior Windows, Skylights, Doors, and Curtain Walls by Uniform Static Air Pressure Difference, 2000.
ASTM E2112-07 American Society for Testing and Materials Standard Practice for Installation of Exterior
Windows, Doors and Skylights 2007.
ASTM E2268-04 American Society for Testing and Materials Standard Test Method for Water Penetration of
Exterior Windows, Skylights, and Doors by Rapid Pulsed Air Pressure Difference 2004.
AS/NZS 4284:1995 Australian Standards Testing of building facades 1995.
FMA/AAMA 100-07 Fenestration Manufacturers Association Standard Practice for the Installation of Windows
with Flanges or Mounting Fins in Wood Frame Construction 2007.
FMA/AAMA 200-06 Fenestration Manufacturers Association Standard Practice for the Installation of Windows
with Flanges for Masonry Construction, 2006.
J IS A 1517-1984 J apanese Industrial Standard Test Method of Water Tightness for Windows and Doors.
Katsaros, J. and Hardman, B., 2007. Failed fenestration: New materials require new techniques. Thermal
Performance of the Exterior Envelopes of Whole Buildings X International Conference Proceedings.
Lstiburek, L.W., 2008. Drainage, Holes and Moderation. Insight, Vol. 4.
RDH Building Engineering Limited. 2002. Water penetration resistance of windowsStudy of manufacturing,
building design, installation and maintenance factors.
7. Acknowledgements
The authors wish to thank the research oversight committee, which include the Alside,
American Architectural Manufacturers Association, American Forest & Paper Association,
APAThe Engineered Wood Association, Architectural Testing, Inc., Atrium Companies, Inc.,
Castcrete Corporation, C.B. Goldsmith and Associates, Inc., CEMEX Corporation, Certified Test
Labs, Do Kim & Associates, DuPont, Fenestration Manufacturers Association, Florida Building
Commission, Florida Home Builders Association, General Aluminum Windows, Institute for
Business and Home Safety, J ames Hardie, J BD Code Services, J ELD-WEN Windows and
Doors, Lawson Windows, Marvin Windows, MI Windows, Nu-Air, Painter Masonry, PGT
Industries, PPG Industries, Protecto Wrap, Silverline Windows and Doors, Simonton Windows,
TRACO, and WCI Group, Inc. This research was supported by the National Science Foundation
(CMMI- 0729739) and the Florida Building Commission.