Structural Behavior of Tie Connections for Residential Brick Veneer Construction
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Structural Behavior of Tie Connections for
Residential Brick Veneer Construction
James M. LaFave 1* and Dziugas Reneckis2
Wood frame structures with anchored brick masonry play a key role in the overall behavior and performance
veneer are a common form of residential construction of residential brick veneer wall construction.
throughout the United States, particularly in regions of
moderate seismicity and/or high wind. Brick veneer wall Prescriptive Construction Requirements and
construction is valued for its pleasant appearance, excellent Recommendations
thermal performance, and ability to prevent water penetra-
tion [Drysdale et al. (1999)]. Residential brick veneer Currently, prescriptive installation requirements for
construction has often performed effectively during mod- corrugated sheet metal ties used to connect brick veneer
erate earthquakes and severe wind storms, by presenting to wood frame backup construction are specified in the
greater structural resistance than that of bare wood framing Masonry Standards Joint Committee (MSJC) Code [MSJC
[Hamilton et al. (2001); IMI et al. (1990); McGinley et al. (2005)], the International Residential Code (IRC) for
(1996)]; brick masonry walls can also protect residential One- and Two-Family Dwellings [ICC (2003)], and the
structures and their contents from airborne debris during Brick Industry Association (BIA) Technical Notes 28 [BIA
severe wind storms [McGinley et al. (1996)]. However, as (2002)], as summarized in Table 1. The tie connections
a result of inappropriate construction materials and meth- should ideally satisfy a set of performance requirements
ods, the structural performance of residential brick veneer such as: a) sufficient strength and stiffness (in tension and
construction can sometimes be significantly compromised compression) to transfer lateral loads to the backup, b)
(to the point of even including out-of-plane collapse of the adequate transverse flexibility to accommodate differential
veneer, as described in more detail below). vertical movements between exterior and interior walls,
and c) resistance to corrosion and moisture transfer across
Residential Brick Veneer the air cavity [BIA (2003)]. As can be seen from Table 1, it
Construction is typically assumed that these performance requirements
are met if a properly spaced grid of galvanized 22 ga.
Residential brick veneer wall systems typically com- corrugated sheet metal ties are employed, attached to the
prise an exterior masonry wall and an interior wood frame wood backup with at least 8d galvanized nails and also
backup (separated by an air cavity), both of which are adequately embedded into the mortar joints.
supported vertically on a foundation. Regularly spaced
corrugated sheet metal ties are used to connect the brick For construction in moderate seismic and/or severe wind
masonry to the backup through the cavity, which acts as regions of the U.S., codes often specify reduced tributary
a thermal barrier and provides for drainage as well as wall areas per each tie connection (closer tie spacings), as
weather resistance. In a veneer wall system, the wood also noted in Table 1. Additionally, for construction in some
framing is designed to carry all lateral and gravity loads, higher risk seismic zones, the MSJC (2005) requires the
except for masonry self-weight [Drysdale et al. (1999)]. In use of horizontal joint reinforcement in the brick masonry;
reality, the brick veneer does, however, carry a portion of however, this requirement has been questioned by Bennett
the lateral load, due to its much higher stiffness than that of and Bryja (2003), based on satisfactory performance of
the wood backup [BIA (2002)]. Lateral loads from exterior unreinforced brick veneer construction during some earth-
masonry walls are transferred to the wood backup via the quakes and severe wind storms, as well as on experimental
tie connections, so the properties of these connections test results and an assessment of the relative hazard of
earthquakes vs. severe wind events in certain regions of the
U.S. Other suggestions that have been proposed for improv-
1
Assoc. Prof., Dept. of Civ. and Env. Engnrg., 3108 ing residential brick veneer construction are to use ring or
Newmark Lab, Univ. of Illinois at Urbana-Champaign, screw-shank nails for attaching ties to the wood framing (to
205 N. Mathews Ave., Urbana, IL 61801, 217-333- increase nail pullout resistance [FEMA (2004)]), and also to
8064, 217-265-8039 (Fax), jlafave@uiuc.edu (* – cor- use adjustable wire ties such as those typically employed in
responding author). light-commercial construction [Krogstad (2003)]; however,
2
Grad. Res. Asst., Dept. of Civ. and Env. Engnrg., 3139 these measures are seldom prescribed or used in actual
Newmark Lab, Univ. of Illinois at Urbana-Champaign, residential brick veneer construction, possibly due in part
205 N. Mathews Ave., Urbana, IL 61801, 217-333- to perceived higher construction costs.
8322, renetski@uiuc.edu.
TMS Journal December 2005 105Table 1. Prescriptive installation requirements for corrugated sheet metal ties (1 in. = 25.4 mm; 1 ft2 = 0.0929 m2)
Construction details MSJC (2005) ICC (2003) BIA (2002)
Tie thickness (gage) [min.] 22 22 22
Tie width (in.) [min.] 0.875 0.875 0.875
Wall area per tie (ft2) [max.] 2.67 2.67 2.67
Horizontal spacing (in.) [max.] 32 24 24
Vertical spacing (in.) [max.] 18 24 24
Wall area per tie in seismic zones (ft ) [max.]
2 a
2.00 2.00 n/a
Wall area per tie in severe wind zones (ft2) [max.] 1.87b 2.00c n/a
Fastener to wood backup [min.] 8d nail d
n/a n/a
Distance from fastener (tie eccentricity) (in.) [max.] 0.5 n/a 0.5
half the veneer
Embedment length into mortar (in.) [min.] 1.5 n/a
thickness or 1.5
Mortar cover on outside face (in.) [min.] 0.625 n/a 0.625
Air gap (in.) [min. and max.] 1.0 1.0 1.0
a
MSJC (2005) also requires horizontal joint reinforcement.
b
New requirement for construction in areas where basic wind speed is between 110 and 130 mph (177 and 209 km/h);
also, max. horizontal spacing of ties is reduced to 18 in.
c
Wind regions of more than 30 psf (1.45 kPa) design pressure.
d
… or fastener having equivalent/greater pullout strength; should also be corrosion resistant.
Common Construction Practice Deficiencies In older construction, however, connection failures by tie
fracture were quite common, sometimes in part as a result of
Over the years, residential brick veneer wall damage corrosion due to inadequate moisture drainage out of the wall
(including cracking, relative movement, and even collapse cavity. Connection failures by tie pullout from the mortar joint
of masonry under out-of-plane loading) has been observed were usually a product of poor mortars and/or too short of a
on a number of occasions resulting from strong wind events tie embedment length into the mortar joint (tie embedment
and moderate earthquakes [Bryja and Bennett (2004); length can be reduced significantly when veneer walls are
FEMA (1999); McGinley et al. (1996); Sparks (1986); constructed with the air cavity in excess of the code specified
EERI (1996); IMI et al. (1990); Page (1991)]. During such value [FEMA (1999)], as well as when ties were installed at
events, out-of-plane wall damage is most likely to occur as excessive slopes across the cavity due to tie misalignment
the brick veneer moves away from the backup (for instance, with the mortar joint [DeVekey et al. (1988)]).
as a result of wind suction pressures during severe wind
storms). This places a high demand on the tensile force Another leading source of damage noted in residential
(and displacement) capacity of the tie connections, which brick veneer construction was the low withdrawal strength
typically ultimately exhibit one of three types of failure: of the tie connection fastener (nail) at its attachment to the
tie fracture, tie pullout from the mortar joint, or tie fastener wood frame backup. As was noted in most of the post-
(nail) pullout from the wood backup. Veneer wall damage disaster damage surveys, and in a separate case study by
has often been explained by improper material use and/or Thomas (1988), low fastener withdrawal strength can be
poor workmanship during construction, particularly as attributed to nails driven into only exterior wood sheathing
relates to the installation of the tie connections. (and not the studs), nails driven into edges of studs, and/or
the use of inadequately sized nails. In particular, follow-
A majority of the collapsed veneer walls referred to ing the destructive Tennessee wind storms of the Fall of
above revealed failed tie connections, which were often 2002, Bryja and Bennett (2004) noted the widespread use
spaced further apart than permitted by codes. In the more of thin 28 ga. ties attached only with 1¼ in. (32 mm) long
modern construction, tie connections typically failed in a tie roofing nails to the wood framing, which was identified
or nail pullout mode at their point of attachment (to either the as a fairly common construction practice in that region.
masonry or the wood backup, respectively) before develop- Consequently, almost all observed veneer wall failures
ing the full tensile strength capacity of the ties themselves. were accompanied by tie connection nail pullout from the
106 TMS Journal December 2005wood backup. Similarly, following the midwestern U.S. brick veneer on wood frame wall systems (when subjected
tornadoes in the Spring of 1999, FEMA (1999) reported to seismic and/or wind loading) by addressing current
on the common use of shorter 6d nails for brick veneer widespread residential construction practice. One phase
tie attachment to the wood backup. Recent visits to local of the study involves laboratory testing of brick-tie-wood
construction sites in central Illinois further confirm the connection subassemblies comprising two bricks with a
common use of 28 ga. ties (as primarily distributed by corrugated sheet metal tie either nail- or screw-attached
local suppliers) with 1¼ in. roofing nails for attachment; to a wood stud. The subassemblies are subjected to in-
however, it was also noted that reduced tie spacings (such plane and out-of-plane loads, permitting the evaluation of
as 16 in. x 16 in. (406 mm x 406 mm)) have sometimes stiffness, strength, and failure modes for a local portion
been adopted. As a result of all these observations, the of a veneer wall system (rather than just of a single tie by
study described herein was undertaken to further explore itself). Some of the preliminary tie connection subassembly
tie connection and overall brick veneer wall behavior as it test results have already been presented elsewhere [Choi
may be affected by various common code compliant and and LaFave (2004)]; this current paper presents addi-
non-compliant construction practices. tional related results in the context of the earlier tests and
other ongoing work. Another phase of the study involves
Previous and Ongoing Studies laboratory testing and analytical development of detailed
of Brick Veneer Construction finite element (FE) models for full-scale brick veneer wall
panel specimens; the out-of-plane performance of solid
Behavior walls has been evaluated, when subjected to static and
dynamic loading, with different combinations of the tie
Over the years, a number of experimental and ana- connections [Reneckis et al. (2004); Reneckis and LaFave
lytical studies have been conducted on brick veneer walls (2005)]. Additional experimental and analytical research
subjected to wind and earthquake loads, to understand is underway at UIUC to explore the effects of different
the interrelationship between the masonry, the ties, and architectural features on veneer wall performance and to
the backup framing (metal or wood). A majority of these better understand the response of brick veneer to wind
studies were related to veneer walls with steel stud backup, loads. Wall specimens with a window opening and a gabled
mainly evaluating the relationship between backup deflec- region (like those found at the ends of many wood-frame
tions and masonry wall cracking [Brown and Arumala homes with brick veneer) are being tested, and additional
(1982); Kelly et al. (1990); Wilson and Drysdale (1990)]; FE studies are being conducted.
in addition, McGinley et al. (1988) experimentally studied
the ultimate behavior of such walls as limited, in part, by As part of the ongoing experimental study of residential
the tie connection strength. brick veneer construction at UIUC, more brick-tie-wood
connection subassemblies have been tested (to accompany
Some of the earliest dynamic tests on brick veneer those already carried out by Choi and LaFave (2004)).
walls with wood backup framing, including un-reinforced These tests further explore tie connection behavior, primar-
and reinforced masonry, were performed by Priestley et al. ily when loaded in tension, for various code compliant and
(1979); they evaluated veneer wall response and dynamic non-compliant tie installation methods (selected primarily
properties in conjunction with failures of the masonry, the by consideration of the reports of inadequate attachment to
stud backup, and the ties. Wind pressure tests on a full-scale the wood frame backup, as described above). Detailed de-
wood frame home structure with masonry veneer were scriptions of the types of tie connection test specimens, test
performed by DeVekey (1987); these tests showed, among setups, and testing procedures used, as well as a summary
other things, that the tie connections may carry unevenly of the most important experimental results, are presented
distributed forces when anchored at different locations to the in the sections to follow. And then, finally, a parametric
wood frame. Testing (including retrofit methods) of older study of the out-of-plane (suction) pressure vs. displace-
(turn of the 20th century) construction was done by Paquette ment behavior of a solid brick veneer wall system with
et al. (2001) for the case where the masonry was originally wood backup is presented as a function of tie installation
anchored to the wood backup with nails only (nail head method, using a finite element (FE) model developed in
embedded into the mortar). And finally, some shake table part based on experimental wall system behavior.
tests were recently performed by Beattie (2004) to evaluate
the behavior of brick veneer at wall corners. Overall then, for
wood frame construction these studies have mainly focused TIE CONNECTION TESTING
on brick veneer wall systems built using older construction PROGRAM
practices, and they have not fully explored the strength limits
of the tie connections themselves. Tie Connection Test Specimens
A study recently undertaken at the University of Illinois Sets of tie connection subassemblies (totaling more
(UIUC) endeavors to evaluate the structural behavior of than 240 specimens) were tested to evaluate the structural
TMS Journal December 2005 107Figure 1—Section View of a Typical Brick-Tie-Wood Subassembly (1 in. = 25.4 mm)
behavior of corrugated sheet metal brick veneer ties of dif- 21/2 in. (64 mm)), while the other end was attached by a
ferent thicknesses installed in various ways representing nail (or screw) fastener to a short length of 2x4 wood stud
common residential construction practice. The tests were (Standard Grade Spruce-Pine-Fir), typically through a strip
conducted monotonically and cyclically in tension, com- of 7/16 in. (11 mm) thick oriented strand board (OSB) APA
pression, and shear. This paper emphasizes tie connection Rated 24/16 wall sheathing. Eight different groups of brick-
tensile strength and stiffness results gleaned from more tie-wood connection subassemblies were constructed
than 155 monotonic and cyclic tests. Detailed results for and tested. Construction variables for the subassemblies
45 tie connection compression tests are given elsewhere included tie thickness, tie attachment method (to the wood
[Choi and LaFave (2004)] and will only be summarized stud), and tie eccentricity (at the connection to the wood
briefly herein. (Detailed results for more than forty mono- stud).
tonic and cyclic tie connection shear tests are also given
elsewhere [Choi and LaFave (2004)]; those tests and Three tie thicknesses were studied: 22 ga. (0.031 in.
results are outside the scope of this paper.) (0.79 mm)), 28 ga. (0.015 in. (0.38 mm)), and 16 ga. (0.062
in. (1.57 mm)). In accordance with typical prescriptive
To completely characterize the local connection be- design recommendations and code requirements for brick
havior of a brick veneer wall system consisting of exterior veneer on wood backup framing, 22 ga. ties should be used
brick masonry attached to wood studs by a series of cor- [MSJC (2005); ICC (2003)], so they were employed in
rugated sheet metal ties, brick-tie-wood subassemblies the majority of the subassemblies. Thinner 28 ga. ties are
were used (rather than simply just testing the ties). The quite often used in actual residential construction practice,
subassemblies typically consisted of two standard bricks which warranted their study as well, whereas 16 ga. ties
connected to a wood stud with a corrugated sheet metal tie are seldom used in residential construction, so they were
(see Figure 1). This type of test specimen (as well as the simply included to explore what the effect would be of
testing method) differs from that specified by American using ties exceeding the minimum specifications.
Society for Testing and Materials (ASTM) E 754 [ASTM
(1998)]. The ASTM E 754 method is applicable only for Nails and wood screws were each investigated as
(tension) pullout resistance of ties embedded in mortar methods for attaching corrugated veneer ties to the wood
joints, whereas the tests reported herein include tension, backup. Galvanized 8d nails (2½ in. (64 mm) long)
compression, and cyclic loading, as well as important tie were used in the bulk of the subassemblies, per typical
connection failure modes other than just tie pullout. prescriptive code requirements [MSJC (2005)]. Galva-
nized roofing nails of two different lengths (21/2 in. and
The bricks used were 31/2 in. x 75/8 in. x 21/4 in. (89 mm 11/2 in. (38 mm)) were also studied since such fasteners
x 194 mm x 57 mm) standard modular three-hole “Colonial (in lengths even as short as 11/4 in. (38 mm)) are often
Reds” joined together by professional masons using Type “substituted” for 8d nails in practice. A few subas-
N mortar (cement:lime:sand = 1:1:6). In each subassembly, semblies (including all of those with 16 ga. ties) were
one end of a bent corrugated sheet metal tie was embedded constructed using #8 x 21/2 in. bugle head galvanized
into the mortar joint (a minimum distance of approximately deck screws instead of nails.
108 TMS Journal December 2005Table 2. Brick-Tie-Wood Connection Subassembly Types and Average Tension Test Results (1 lb = 4.45 N; 1 in. =
25.4 mm)
Monotonic Tension Tests Cyclic Tests (Tension Results)
Ultimate Ultimate
Secant Stiffness Secant Stiffness
Specimen Type Load Load
(lbs/in.) (lbs/in.)
(No.) (lbs) (No.) (lbs)
N(8d)22min 9 118 3,140 4 164 4,510
N(8d)22ecc 20 164 640 10 159 500
N(8d)28min 5 168 2,280 4 162 2,180
N(8d)28ecc 5 155 280 4 146 230
N(2.5)22min 6 106 3,240 5 91 1,400
N(1.5)22min 6 80 1,280 5 72 1,190
S(–)22ecc 10 406 670 5 418 720
S(–)16min 5 397 1,610 5 418 2,350
Most of the subassemblies were constructed with the mum code-prescribed installation requirements with respect
corrugated sheet metal tie bent 90-degrees right over the to either tie thickness or fastener type/length. Finally, the last
head of the nail or screw fastener, which represents the two categories of tie connections (S(–)22ecc and S(–)16min)
“best-case” installation situation from the standpoint of each exceed the minimum code-specified installation re-
(minimum) tie eccentricity at the bend (with respect to the quirements for tie thickness and/or fastener type.
centerline of the fastener). For specimens with 8d nails or
wood screws, this resulted in a small eccentricity (see Fig- In addition to the 108 regular tie connection monotonic
ure 1) of approximately 5/32 in. (4 mm); in the cases where and cyclic tension tests reported in Table 2, eighteen more
roofing nails were used, this resulted in a slightly larger tension tests were conducted on eccentric subassemblies
eccentricity of approximately 1/4 in. (6 mm), due to the where a relative offset displacement was then also intro-
larger nail head. To investigate the effect of variability in tie duced between the bricks and the wood stud at the onset
eccentricity (possibly due to sloppy tie installation and/or of testing. This different type of test, which was conducted
poor tie alignment with the brick veneer mortar joint), some on subassemblies representing just three of the specimen
specimens were tested with a greater eccentricity at the bend types, is briefly described further in the section below re-
of 1/2 in. (13 mm), as shown in the Figure 1 inset, which is garding the test setup and testing procedures. There were
the maximum eccentricity permitted by the MSJC. also 35 regular and 10 offset specimens tested in monotonic
compression, representing six of the eight different speci-
A listing of the eight different types of brick-tie-wood men types described above.
connection subassemblies constructed and tested (per the
construction variables described above) is provided in the Finally, in order to simply evaluate the relation-
first column of Table 2 (where the number of test specimens, ship between tie connection tension pullout strength vs.
as well as summary average tensile strength and stiffness tie embedment length, 30 brick-tie specimens (without
data, is also presented for all groups of regular subassemblies wood studs) were prepared and tested in tension (10 sets
subjected to either monotonic tension or cyclic loading). The of specimens in triplicate) with different tie embedment
shorthand notation used to designate the categories of test lengths into the mortar joint. The 22 ga. and 28 ga. ties
specimens is of the form: F(f)##e, where “F(f)” represents were evaluated for embedment lengths of 11/2 in. (38 mm),
the type and size of fastener [Nail (8d or 2.5 in. roofing or 2 in. (51 mm), 21/2 in. (64 mm), and 3 in. (76 mm), while
1.5 in. roofing) vs. Screw (–)], “##” represents the thickness the 16 ga. ties were only tested with embedment lengths of
of the tie (22 ga. vs. 28 ga. vs. 16 ga.), and “e” represents the 11/2 in. and 2 in. (which was almost the maximum possible
eccentricity of the tie at the bend from the fastener centerline embedment for this type of “pre-bent” tie, accounting for
(minimum eccentricity vs. 1/2 in. eccentricity). the presence of the minimum air cavity between the tie
and the bricks). A 11/2 in. tie embedment length into the
The first two categories of tie connections (N(8d)22min mortar joint represents the minimum embedment permitted
and N(8d)22ecc) both meet the minimum code-specified by the MSJC, while 3 in. would be about the maximum
installation requirements described above. The next four practical embedment length (for the size of bricks used in
categories of tie connections (N(8d)28min, N(8d)28ecc, this study) in keeping with the MSJC minimum mortar
N(2.5)22min, and N(1.5)22min) all fail to meet the mini- cover requirement beyond the end of a tie.
TMS Journal December 2005 109Figure 2—Brick-Tie-Wood Connection Subassembly Test Setup
Test Setup and Testing Procedure cell in the testing frame was connected to the computer via
the controller to allow for both measuring and controlling
Experiments on the brick-tie-wood connection the load and overall (actuator) displacement. Two LVDTs
subassemblies were conducted using a universal testing were attached directly onto every test specimen (one on
machine in the Newmark Structural Engineering Labora- each side of the wood stud) to measure just the displace-
tory (NSEL) at UIUC. The test setup is shown in Figure ment occurring from the face of the bricks to the face of
2; all tests were performed at least 28 days after specimen the wood stud; these LVDTs were also connected to the
fabrication. High-strength gypsum was applied to the bot- computer via the controller. All data were simultaneously
tom of the bricks to level the surface where they contacted recorded every 0.5 sec.
the testing machine bed, thereby avoiding shear loads on
the mortar joint. Steel plates and C-clamps held the bricks The monotonic tension and compression tests to failure
to the testing machine bed, with rubber sheets installed were actuator displacement controlled at a rate of 0.1 in./
between the plates and the bricks to ensure a tight fit. C- min (2.5 mm/min). The cyclic tests were also controlled
clamps were also used to grip the wood stud in the upper by actuator displacement, with a total of 24 intended cycles
part of the testing machine. as shown in Figure 3. (The repeat same amplitude cycles
and the small displacement cycles were included to explore
An Instron 8500 Plus controller, a 20 k (90 kN) MTS strength and stiffness degradation.) Cyclic displacements
Systems Universal Testing Frame, and a PowerMac com- were applied at the rate of 1 cycle/min; after 24 cycles,
puter with National Instruments analog-to-digital acquisi- additional displacement cycles were occasionally applied
tion board and custom written LabView software were used in increments of 0.05 in. (1.3 mm), as needed to produce
to control the tests and to collect the data. A small load subassembly failure.
To simulate possible vertical differential wall move-
ment across the cavity (from long-term moisture and tem-
perature effects), a few additional eccentric subassemblies
were tested monotonically and cyclically after being given a
substantial initial 1/4 in. (6 mm) relative offset displacement
(perpendicular to the face of the tie) between the bricks and
the wood stud. Such offset tests were only conducted for
three of the eight different types of tie connection subas-
semblies (and typically in sets of only three specimens),
so those results in tension are not tabulated in Table 2.
Detailed results for the monotonic and cyclic offset tests
are provided elsewhere [Choi and LaFave (2004)]; general
behavior trends from the offset tests will be noted below
whenever they are significant in comparison with the rest
Figure 3—Planned Displacement History for Cyclic of the tie connection subassembly test data reported.
Subassembly Tests (1 in. = 25.4 mm)
110 TMS Journal December 2005Figure 4—Monotonic Tension Load-Displacement Figure 5—Nail Pullout From Wood During Cyclic
Curves for All Five Type N(8d) 28 min Subassemblies Testing of a Type N(1.5)22min Subassembly
(1 lb = 4.45 N; 1 in. = 25.4 mm)
The brick-tie embedment specimens were all tested loading was negligible; the maximum reduction in aver-
in monotonic tension up to pullout failure, using a slightly age tensile strength when subjected to cyclic loading for
modified version of the apparatus described above. The test any of the eight types of tie connections tested was only
setup and testing procedure were similar to those specified about 10%. (In some cases, there was a more significant
by ASTM E 754, except that the ties were directly grasped cyclic loading effect on the connection failure modes
by custom-made screw grips in the testing machine. and/or on the tie connection average tensile stiffness, as
will be described in more detail below.) Furthermore, dif-
Subassembly Tension Test Results ferent tie eccentricities were found to typically only affect
tie connection average subassembly tensile strength by
Table 2 lists the number of test specimens, along with about 15% or less for otherwise identical groups of test
average tensile strength and stiffness data, for all groups specimens (although additional movements on the order of
of regular tie connection subassemblies subjected to about 0.1 to 0.2 in. (2.5 to 5.1 mm) were usually required
monotonic tension or cyclic loading. An example of one to develop the equivalent tensile strength in the eccentric
set of monotonic tension load-displacement curves, for tie connections). Therefore, to best summarize the detailed
the N(8d)28min subassemblies, is shown in Figure 4. The average monotonic and cyclic tensile strengths presented
tensile strength for each subassembly is simply defined as in Table 2, certain aggregate values (as a function of only
the maximum tension load achieved during testing. (The tie thickness and fastener type) can be computed and
coefficient-of-variation for the average tensile strengths of compared, as follows.
the different groups of test specimens was typically between
10% and 20%.) The tensile stiffness for a subassembly is For all tie connection subassemblies with 8d nails
defined as the secant stiffness of the load vs. average LVDT and meeting the other minimum prescriptive installation
displacement curve up to a tension load of 100 lbs (445 N) specifications, the average tie connection tensile strength
(or up to an opening displacement of 0.05 in.(1.27 mm) in was 153 lbs (681 N), from 43 total tests, whereas for
the rare case when a particular specimen never reached a all of the tie installations that used thinner (28 ga.) ties,
tensile load of 100 lbs). Computing the stiffness to a load the average tie connection tensile strength was 158 lbs
of 100 lbs was in part selected because this load represents (703 N), from 18 total tests. This indicates that using a
approximately two-thirds of the average maximum tension thinner tie does not necessarily compromise the strength
load achieved in subassembly types just meeting minimum of a typical brick-tie-wood connection (assuming that no
code-specified installation requirements. Furthermore, us- tie deterioration has occurred). However, when 8d nails
ing the secant stiffness to 100 lbs allows for easy compari- were replaced with similar length (21/2 in. (63.5 mm))
son with the BIA Technical Note recommendation that a roofing nails in subassemblies with 22 ga. ties, the average
minimum tie connection load of 100 lbs should be achieved tie connection tensile strength was 99 lbs (441 N), from
at a deflection of 0.05 in. [BIA (2003)], which corresponds 11 total tests; the average tie connection tensile strength
to a secant stiffness of 2,000 lbs/in. (350 N/mm). was only 76 lbs (338 N) when 11/2 in. roofing nails were
used (also from 11 total tests). (For reference, this tie
Tie Connection Tensile Strength connection tensile strength of 76 lbs (338 N) corresponds
to a uniform local wall suction pressure of about 28.5 psf
With respect to average tie connection subassembly (1.35 kPa) applied over a wall area of 2.67 ft2 (0.25 m2).)
tensile strength, the effect of cyclic loading vs. monotonic This indicates that using short roofing nails (instead of 8d
TMS Journal December 2005 111nails) to attach veneer ties to the wood backup can result for N(8d)28min (from 9 total tests), 2,400 lbs/in. (420 N/mm)
in as much as a 50% or more reduction in tie connection for N(2.5)22min (from 11 total tests), and 1,980 lbs/in.
tensile strength. (347 N/mm) for S(–)16min (from 10 total tests). The one
exception to this was tie connection subassembly group
The predominant tie connection failure mode observed N(1.5)22min, which had a somewhat lower overall aver-
in the monotonic tension tests of nailed subassemblies was age tensile stiffness of 1,240 lbs/in. (217 N/mm), from 11
nail pullout from the wood stud, which helps to explain total tests, due to the relatively early onset of nail pullout
why tie thickness had no effect on average tie connection contributions to connection flexibility since very short
tensile strength for these specimens, while nail type had a roofing nails were used.
significant effect. During cyclic testing of subassemblies
with nails, various failure modes were observed, including
nail pullout (see Figure 5), tie fracture, yield around the tie
hole (permitting the head of the nail to pass through), and
tie pullout from the mortar joint. Finally, the maximum
effect of an initial offset displacement on average tensile
strength of tie connections with nail fasteners was found
to be only about a 15% reduction in strength.
When 21/2 in. (63.5 mm) long wood screws were
used to replace 8d nails in subassemblies with 22 ga. ties
or as the principal fastener in subassemblies with 16 ga.
ties, the average tie connection tensile strength was 409
lbs (1,820 N), from 25 total tests, an increase of more
than 150% over the average strength of tie connection (a) Nail Attached 28 ga.
subassemblies just meeting the minimum prescriptive
installation requirements. The predominant tie connec-
tion failure mode observed in both monotonic and cyclic
tension tests of subassemblies with screw fasteners was
tie pullout from the mortar joint, with a few occurrences
of either yield around the tie hole (permitting the screw
head to pass through) or tie fracture. The deleterious effect
of initial offset displacement on the tensile strength of tie
connections with screw fasteners was more pronounced
than in subassemblies with nails, but the strength of con-
nections with wood screws subjected to initial offset was
still always much higher than that of any category of tie
connection using nails.
(b) Nail Attached 22 ga.
Tie Connection Tensile Stiffness
With respect to average tie connection subassembly
tensile stiffness, the effect of cyclic loading vs. monotonic
loading was more apparent than it was for tensile strength,
but the reduction in average tensile stiffness when sub-
jected to cyclic loading for most types of tie connections
was still typically less than about 15%. However, varia-
tion in tie eccentricity was consistently found to have a
considerable effect on tie connection average subassembly
tensile stiffness, while fastener type also had an effect.
In terms of overall average tensile stiffness (for the
monotonic tests plus the cyclic tests), all but one group of (c) Screw Attached 22 ga. and 16 ga.
tie connections that were constructed with the minimum
possible tie eccentricity from the fastener at the tie bend Figure 6—Idealized Monotonic Tension Force-
had average tensile stiffness values close to or in excess of Displacement Relationships for Brick-Tie-Wood
2,000 lbs/in. (350 N/mm) – 3,560 lbs/in. (624 N/mm) for Connection Subassemblies (1 lb = 4.45 N; 1 in. =
N(8d)22min (from 13 total tests), 2230 lbs/in. (391 N/mm) 25.4 mm)
112 TMS Journal December 2005On the other hand, all three groups of tie connections length of 11/2 in. (38 mm), instead of the 21/2 in. (64 mm)
that were constructed and tested with the larger 1/2 in. tie that was actually used, it is unlikely that there would have
eccentricity from the fastener at the tie bend had overall been any significant changes in the average subassembly
average connection tensile stiffness values of only about tensile strength and stiffness values presented above.
one-third or less of 2,000 lbs/in. (350 N/mm) – 600 lbs/in.
(105 N/mm) for N(8d)22ecc (from 30 total tests), 690 Summary of Subassembly Compression Test
lbs/in. (121 N/mm) for S(–)22ecc (from 15 total tests), and Results
260 lbs/in. (46 N/mm) for N(8d)28ecc (from 9 total tests).
When monotonic and cyclic tension tests were conducted All of the brick-tie-wood subassemblies that were
including initial offset displacements, there was typically a tested in compression failed by flexural buckling of the tie,
small additional decrease in connection stiffness for these regardless of the type of test specimen. The compression
groups of specimens with tie eccentricity, by as much as load typically decreased after buckling and then increased
about 20%. again at very large displacements as the tie began to be
crushed in the air cavity. For 22 ga. ties, the average subas-
As a comparative measure of the overall tension behav- sembly compressive strengths at buckling typically ranged
ior (including both strength and stiffness) for the various anywhere from about two-thirds to even greater than the
types of brick-tie-wood connection subassemblies tested, tensile strengths for similar nailed subassemblies, while the
idealized multi-linear average load-displacement curves average compressive strengths were typically only about
have been prepared and plotted together in Figure 6 for all one-fourth of the average tensile strengths for subassem-
specimen types. Details of the methodology used for pre- blies with the thinner 28 ga. ties (and for the much thicker
paring these average curves are given elsewhere [Choi and 16 ga. ties, subassembly compressive strengths were usu-
LaFave (2004)]. The plots shown in Figure 6 are based only ally about twice their tensile strengths). Furthermore, subas-
on the monotonic tension tests; however, it has been shown sembly compressive stiffness values were lowest when 28
that overall average multi-linear tension envelope curves ga. ties were used and highest when 16 ga. ties were used.
based on cyclic tests are typically well matched with the This indicates that tie thickness is the main determinant
idealized curves for monotonic tension [Choi and LaFave for both subassembly compressive strength and stiffness,
(2004)]. The effects of tie connection tensile strength and while the presence of an initial offset displacement and/or
stiffness on overall brick veneer wall system performance cyclic loading was also found to have a small detrimental
will be explored briefly in a later section. effect on subassembly behavior in compression.
Tie Embedment Tests Detailed subassembly compressive strength and stiff-
ness data are given elsewhere, along with more information
The average maximum mortar joint tensile pullout about subassembly failure modes in compression [Choi
strengths from the brick-tie connection embedment tests and LaFave (2004)]. However, it should be understood
as a function of tie thickness and embedment length are that all of the compressive strength and stiffness values
provided in Table 3. For any particular embedment length, from such subassembly testing are lower bounds for actual
thicker ties generally had greater pullout strengths, and as brick veneer walls, owing to the invariable presence of
would be expected, longer embedment lengths generally some “mortar droppings” in the air cavity that effectively
resulted in greater pullout strengths as well. The sets of increase both the compressive strength and stiffness attrib-
embedment length test specimens typically exhibited higher utable to any one tie connection in the system. Approximate
average strengths than did the corresponding groups of modeling strategies to deal with this effect, at least in terms
brick-tie-wood subassemblies that had similar or even lon- of tie connection compressive stiffness, are summarized
ger embedment lengths, because the strength of those sub- elsewhere [Reneckis and LaFave (2005)]; relying on this
assemblies was usually controlled by other failure modes effect a priori with respect to tie compressive strength may
(at lower loads), such as pullout of the nail from the wood not be advisable, however, which further points up the
stud. Therefore, even if the brick-tie-wood subassemblies desirability of using at least 22 ga. ties.
had been constructed with the MSJC minimum embedment
BRICK VENEER WALL PARAMETRIC
Table 3. Average Pullout Strengths for Brick-Tie
Embedment Specimens (1 lb = 4.45 N; 1 in. = 25.4 mm) ANALYSIS EXAMPLE
Embedment Average Ultimate Load (lbs)
Three-dimensional finite element (FE) models have
Length (in.) 28 ga. 22 ga. 16 ga. been developed to represent one-story wall panels of
1 /2
1
174 263 384 residential brick veneer on wood frame construction, as
2 174 410 470 described in detail by Reneckis and LaFave (2005), based
21/2 243 500 - on full-scale experimental wall specimen tests [Reneckis
3 285 594 - et al. (2004)] and brick-tie-wood subassembly test results
TMS Journal December 2005 113such as those described herein. The FE wall panel models siding material is used on perpendicular exterior walls).
are calibrated to capture the static and dynamic experimen- Brick termination at edges is also common in practice at
tal brick veneer wall behavior at different response levels, wall corners with control joints and near large window
up to and including tie connection damage and even wall and door openings, which permit individual sections of
collapse. The models are then used to perform parametric the veneer to move independently of one another.
studies evaluating the effects of certain combinations of tie
connections and layouts on the out-of-plane performance of As shown in Figure 7, the wall panel specimen com-
brick veneer walls subjected to static and dynamic loading. prised a standard single-story wood frame stud wall with
In the analysis example described below, the experimen- exterior OSB sheathing and interior gypsum wallboard,
tal and modeling procedures for the full-scale wall panel resting on a partial floor diaphragm supported by a rein-
specimens are briefly summarized, followed by parametric forced concrete (RC) foundation. Across the top, the wall
studies of the out-of-plane behavior of brick veneer walls panel supported partial roof/ceiling framing (connected
subjected to uniform (suction) pressures, with various tie to the wall top-plate with metal rafter ties), which in turn
connection installation methods. braced the wall across its length. The interior ends of both
the partial floor and roof/ceiling framing were supported
Brief Description of Brick Veneer Wall Panel by a steel reaction frame, to represent the rest of a wood
Experiments and FE Models frame house. The brick veneer walls were constructed by
professional masons using identical masonry materials
Summary of Brick Veneer Wall Test Specimens as for the brick-tie-wood subassemblies described above.
The brick masonry wall rested on the RC foundation (with
The FE models described herein were developed a layer of vinyl flashing) and was connected to the wood
to represent the structural behavior of brick veneer wall frame backup with 28 ga. corrugated sheet metal ties (at-
panel specimens that were tested on a shake table at UIUC tached to the wood frame with 8d nails) spaced apart 24
[Reneckis et al. (2004)]. The full-scale brick veneer and in. (610 mm) vertically (in five rows) and 16 in. (406 mm)
wood frame test structure, shown in Figure 7, was designed horizontally (on every stud). In the first wall test speci-
and constructed to represent as closely as possible typical men, all of the ties were installed following the “best case”
residential construction practice, in general conformance construction practice, where the tie bend was located at the
with MSJC (2005), ICC (2003), and BIA (2002). The 11 ft nail (i.e. N(8d)28min type connections), and in a second
(3.37 m) length of the wall panel specimen was governed wall test specimen (where the partially collapsed portion of
by the size of the shake table; similar length (and longer) brick veneer from the first wall specimen was rebuilt and
solid exterior walls are often found in residential construc- reconnected to the backup) the top two rows of ties were
tion, particularly at garages. Also, the brick veneer walls installed at the maximum code permitted bend eccentricity
tested had free edges (open ends), similar to those found in of 1/2 in. from the nail (i.e. N(8d)28ecc). During testing of
residential construction with “front face” veneer walls only the second veneer wall specimen, a possible repair tech-
(where the masonry is terminated at a corner and some other nique was also evaluated by temporarily post-installing four
mechanical anchors across the upper wall region, between
Figure 7—Full-Scale Brick Veneer Wall Panel Test Specimen (1 in. = 25.4 mm)
114 TMS Journal December 2005the top two rows of ties, as indicated in Figure 7. The top Table 4. Brick Veneer Wall Panel FE Model Material
of the masonry was free (comparable to the wall edges) Properties (1 ksi = 6.89 MPa)
in all tests, assuming that any local restraint that could be Modulus of
provided in actual construction by a nearby soffit or other Poisson’s Ratio,
Material Elasticity, E
trim-work would be negligible. υ
(ksi)
Wood Studs 1,200 0.4
The brick veneer wall test specimens were subjected
to static and dynamic out-of-plane loading, and their dy- OSB Sheathing 930 0.4
namic properties and ultimate behavior were evaluated, as Brick Masonry 2,000 0.2
described in detail elsewhere [Reneckis et al. (2004)]. In
general, displacement measurements were taken along the The wood frame stud wall and exterior OSB sheathing
wall centerline and at a top corner, on both the wood stud geometries were modeled using beam and shell elements,
and the brick veneer sides; these measurements provided respectively, as shown in Figure 8a. The wood frame wall
veneer and backup displacements relative to the shake table, panel model had simple supports with rotational springs
as well as relative displacements between the veneer and (stiffness = 1,000 k-in./rad (110 kNm/rad) per spring)
the backup (tie deformations). Finally, out-of-plane ac- across the base, representing the experimentally calibrated
celerations were measured on the shake table and at three rotational resistance of the stud wall panel at the wood floor
locations along the centerline of the wall during dynamic and RC foundation boundary; across the top, translational
testing. The static and dynamic tests captured the overall springs (stiffness = 1.2 k/in. (210 kN/m) for interior springs,
performance of the wall system, including interaction and and 4.8 k/in. (840 kN/m) for the two corner springs) were
load-sharing between the brick masonry veneer, the tie implemented, representing the experimentally calibrated
connections, and the wood frame backup. Dynamic test resistance provided by the rafter ties as well as the deforma-
results for the wall specimens corresponded to three levels tion effects of the roof/ceiling framing and steel reaction
of response and damage, which were labeled as: elastic frame (Figure 8b). The brick masonry veneer was modeled
(no visible damage), intermediate (onset of tie connection using shell elements, simply supported across the base
and brick veneer damage), and ultimate (accumulation of (able to pivot without much restraint under out-of-plane
tie and veneer damage sufficient to lead to collapse). It was loading during testing) at a fixed distance away from the
found that tie connection stiffness and strength significantly backup; therefore, in the FE model (as in the physical wall
affected the veneer wall panel performance at all stages of structure) only the tie connections provided lateral stability
behavior. The progression of system damage (particularly to the brick veneer wall.
at the tie connections) was noted up until partial collapse of
the veneer walls; tie connection failure modes were closely The FE brick veneer and wood frame backup models
related to those found during subassembly testing, as well were linked together with axial bar elements representing
as those observed in actual veneer wall damage following the tie connections and assigned nonlinear inelastic mate-
some of the natural disasters described above. rial properties in tension and linear elastic in compression,
to specifically capture different tie connection features (as
Brick Veneer Wall FE Modeling Summary a function of tie thickness and installation method). The
idealized monotonic tension force-displacement responses
The FE wall models were developed using the analysis (shown in Figure 6) were converted into equivalent stress-
software ABAQUS (HKS 2002) and the pre-/post-processor strain relations, by considering the axial connection model
software MSC.Patran (2001) to represent the brick veneer element cross-sectional area and length. To represent cyclic
wall panel specimens and their experimental behavior, as behavior (under dynamic loading during FE model valida-
fully described by Reneckis and LaFave (2005). The wall tion), the monotonic subassembly test results were also
models consisted of the wood frame wall panel and the used as simple estimates of the backbone curves for the tie
brick masonry veneer, assigned the linear elastic material connection material hysteresis rule in tension (as noted by
properties listed in Table 4. Other surrounding “bound- Choi and LaFave (2004), average envelope curves for the
ary” components of the (experimental) test structure were tension portion of subassembly cyclic behavior were quite
implemented as linear elastic spring support conditions, similar to the companion subassembly monotonic tension
with properties carefully calibrated in conjunction with test results). On the other hand, tie connection behavior in
wall panel experimental results from the extensive static compression was assumed as linear elastic, in part due to the
and dynamic laboratory testing. The corrugated sheet inevitable tie constraint provided by some mortar bridging
metal tie connections, which played such a key role in the across the brick veneer wall cavity. More details about the
overall wall specimen behavior, were assigned nonlinear tie connection nonlinear hysteretic tensile and linear elastic
inelastic “material” properties. The properties of these tie compressive behaviors (as well as model material density
connection models were primarily based on the idealized and damping properties) can be found elsewhere [Reneckis
force-displacement relationships from the brick-tie-wood and LaFave (2005)]; these particular properties are not
subassembly tests (as shown in Figure 6). relevant to the limited FE wall model static parametric
TMS Journal December 2005 115(a) (b)
Figure 8—Brick Veneer Wall Panel FE Model (1 in. = 25.4 mm)
studies reported herein, which principally depend on the Brick Veneer Wall FE Analysis Parameters
monotonic tension behavior of the tie connections. and Results
The brick veneer FE wall models have been calibrated In the analysis example, seven of the tie connection
and validated per experimental results to effectively capture behaviors (all but the 16 ga. ties) shown in Figure 6 (from the
the brick veneer wall behavior when subjected to varying subassembly tests) are implemented in the brick veneer FE
magnitude dynamic and static point (pull) loads [Reneckis wall models. Furthermore, two different tie spacing layouts
and LaFave (2005)]. Various degrees of brick veneer wall are studied, based on typical construction practice. The first
damage observed experimentally (onset of tie failure, layout has a tie spacing of 24 in. (610 mm) vertically and 16
spread of tie failures across the top row, and ultimately in. (406 mm) horizontally (at every stud), with a wall area
collapse) were able to be captured by considering whether per tie of 2.67 ft2 (0.25 m2), as in the experimental wall panel
tie connections at key locations in the models exceeded specimen and model shown in Figures 7 and 8, respectively
their ultimate load (and/or displacement) capacities. These (and in general conformance with the maximum MSJC,
“working” FE wall models are then used to perform para- IRC, and BIA tie spacing requirements listed in Table 1).
metric studies to determine the performance of veneer walls The second layout involves a tie spacing of 16 in. by 16 in.
with different types of ties and tie installation methods, as (406 mm by 406 mm), corresponding to a reduction in wall
described below. area per tie of 33% (to 1.78 ft2 (0.17 m2)), which even meets
(a) 5 Rows and 9 Columns (24 in. x 16 in.) (b) 7 Rows and 9 Columns (16 in. x 16 in.)
Figure 9—FE Wall Model Uniform Suction Pressure Pushover Curves per Tie Layout (1 psf = 47.9 Pa; 1 in. = 25.4 mm)
116 TMS Journal December 2005the new 2005 MSJC maximum wall area requirement for model with the tie spacing of 24 in. x 16 in. (610 mm x
construction in very high wind load regions. 406 mm) in Figure 9a exhibits a uniform pressure capac-
ity below 30 psf (1.45 kPa). This capacity is less than the
Static pushover analyses were performed on the four- minimum design (unfactored) leeward wind pressure often
teen different FE wall models by subjecting the veneer used for wall components and cladding of typical residential
wall outer shell elements to uniformly distributed pull structures for exposures in the coastal regions of the U.S.;
(outward) loads of increasing magnitudes, representing a even in non-coastal regions, this capacity can be less than
wind (suction) pressure load. (As mentioned previously, the appropriate factored design wind suction pressure. The
residential veneer wall damage often occurs due to such veneer wall strength with a reduced spacing of this type of
wind pressures placing high demand on the tensile per- tie connection is still less than that of a wall utilizing 22 ga.
formance of the tie connections.) The suction pressure vs. ties attached with 8d nails at the wider spacing. These are
veneer top displacement responses for the FE wall models indications of how “sub-standard” tie connection behavior
are shown in Figure 9, clearly demonstrating the effects can impact overall brick veneer wall performance at the
of the different types of tie connections. These overall systems level during extreme loading events. The type of
wall model pressure-displacement curves up to the peak experimental and analytical methodology described above
load points accurately reflect the effects of individual tie could prove useful for assessing in a rational fashion the
unloading (post-peak) behavior; displacement values for implications of suggested changes in prescriptive brick
wall model curves well past the ultimate wall load points veneer construction guidelines.
(into the overall wall unloading range), however, may
be somewhat arbitrary because masonry veneer cracking SUMMARY, CONCLUSIONS, AND
(not represented by the models) could occur at these large RECOMMENDATIONS
displacements, after the onset of tie damage. (The curves
for the S(–)22ecc ties end abruptly because the individual Residential anchored brick veneer construction, which
tie connection post-peak behavior exhibits a relatively is typically designed and built based on prescriptive code
steep drop in load (see Figure 6c), causing some analysis requirements, has sometimes exhibited distress resulting
difficulties; such a veneer wall (with “high-strength” screw from strong wind events and moderate earthquakes. The
fasteners) would most likely have its strength governed at damage has often been attributed to the performance of
very high pressures by other failure modes than at the tie the corrugated sheet metal tie connections used to con-
connections anyway, like masonry cracking.) nect the brick veneer to the wood backup, so a study was
undertaken to better understand the structural behavior of
The relative pre-peak stiffnesses of the veneer wall these tie connections. Laboratory testing of brick-tie-wood
models (in Figure 9) are closely related to the initial connection subassemblies was conducted to explore the
tension stiffness of the type of tie connections used (per effects of tie thickness, tie attachment method to the wood
Figure 6). However, the relative ultimate strengths of the stud, tie eccentricity at the connection to the wood stud,
wall models are not always directly related to the relative and tie embedment length on the strength and stiffness of
ultimate strengths of the individual tie connections used. veneer tie connections. Results from the tie connection tests
For example, the wall model with N(8d)28ecc ties is sig- and related brick veneer wall panel experiments were then
nificantly weaker than the other walls with ties attached by used in analytical models of brick veneer walls to assess the
8d nails, having only about the same strength as the wall effects of different tie connection installation procedures
utilizing weaker N(2.5) 22min ties. In general then, the ul- on wall system performance.
timate strength of brick veneer walls is dependent not only
on the ultimate strength of the individual tie connections, For nailed tie connections, their strength was typi-
but also on their stiffness. (This means that brick veneer cally governed by nail pullout from the wood stud, while
walls undergoing significant differential (vertical) offset their stiffness was mostly a function of the amount of tie
displacements between the masonry and the wood backup eccentricity and the tie thickness. Nailed tie connections
could possibly experience a reduction in their out-of-plane not meeting current minimum installation requirements
capacity, although the tie connection subassembly test re- exhibited reductions in strength (from using short roofing
sults with initial offset displacement generally showed only nails) and in stiffness (from using thinner gage ties or short
small decreases in initial stiffness and ultimate strength.) roofing nails) of up to about 50% and 65%, respectively.
The brick veneer wall model behavior with reduced tie spac- On the other hand, tie connections with wood screws had
ing (Figure 9b) exhibits an increase in overall strength; the much higher strength, but similar stiffness, when compared
reduction in tie connection spacing (a 40% increase in the to nailed tie connections just meeting the minimum required
number of ties used) resulted in an overall strength increase installation criteria.
of roughly 30% for most wall models.
Tie connection strength and stiffness were both found
Overall, the wall models with tie connections fastened to be important toward determining the overall ultimate
using 11/2 in. roofing nails were the weakest ones. That wall strength of brick veneer walls. For wall panel models with
TMS Journal December 2005 117You can also read
