Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Load-Sharing and Redistribution in a One-Story

Woodframe Building
Phillip J. Paevere, A.M.ASCE1; Greg C. Foliente, A.M.ASCE2; and Bo Kasal, A.M.ASCE3

Abstract: This paper presents the results of a series of experiments on a full-scale one-story L-shaped woodframe house. The major
objective of the experiments was to collect data for use in the validation of analytical models, and to examine the load redistribution in
a nonsymmetrical light-frame structure under lateral loading. In these experiments, the distribution of the reaction forces underneath the
Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

walls, and the displaced shape of the house were measured in detail under static and static–cyclic loading. The experimental results have
provided the most detailed picture of the reaction forces beneath a nonsymmetrical light-frame structure under lateral loading ever
recorded. It was shown that there is potential for significant sharing and redistribution of applied lateral load, between the main shear walls
of a light-frame house under both pseudoelastic and inelastic response conditions. Commonly used techniques for lateral load distribution
in light-frame structures do not accommodate redistribution of the loading between the walls in the structure.
DOI: 10.1061/共ASCE兲0733-9445共2003兲129:9共1275兲
CE Database subject headings: Full-scale tests; Framed structures; Wooden structures; Load distribution; Data collection.

Introduction applied extensively, and forms the basis of the design philosophy
and methodology. Inevitably, societal and client expectations will
Many people live and work in light-frame buildings. They rely on drive a change toward a more technology-intensive approach to
these structures for protection against natural disasters such as the design and evaluation of light-frame structures, based on con-
hurricanes and earthquakes and have an inherent expectation of sistent and reasonably accurate prediction of their performance.
their safety. Although light-frame buildings have generally per- However, the development of improved performance prediction
formed well in natural disasters, many have also suffered exten- technologies, for light-frame structures, requires a detailed under-
sive damage 共Foliente 1998兲, causing financial loss and some- standing of the structural behavior of light-frame buildings, as
times injuries or death. In the 1994 Northridge Earthquake, for well as the environmental loadings to which they are subjected
example, 24 deaths 共including 16 deaths in the Northridge Mead- during their lifetime. Full-scale structural testing in the laboratory,
ows apartment building兲 and financial losses of more than $20 combined with analytical modelling, are essential in obtaining
billion occurred as a result of damage in light-frame construction. this understanding.
This represented most of the fatalities, and more than half of the This paper presents the results of experiments conducted on a
total damage bill 共Office of Emergency Services 1995; Kircher full-scale one-story L-shaped woodframe house. The major objec-
et al. 1997兲. tives of the experiments were to: 共1兲 Collect data for use in the
Until quite recently, modern technology has played only a validation of analytical models of light-frame structures, includ-
small role in the design and construction of light-frame buildings, ing a three-dimensional finite-element 共FE兲 model; and 共2兲 exam-
particularly those made from wood. The construction techniques ine the load distribution and load-sharing mechanisms in a non-
have been developed over long periods of time, based mainly on symmetrical light-frame structure. Complete specifications of the
tradition and experience. In other, relatively modern forms of test house and full details of the experimental program and results
construction, such as prestressed concrete, technology has been are given in Paevere 共2002兲. Common methods of seismic perfor-
mance evaluation—quasi-static cyclic testing, pseudodynamic
testing, and shaketable testing—have complementary strengths
1
Research Scientist, CSIRO Manufacturing and Infrastructure Tech- and weaknesses 共Foliente 1996; Leon and Deierlein 1996兲. Thus,
nology, P. O. Box 56, Highett, Victoria 3190, Australia 共corresponding each has a role to play in helping us better understand the seismic
author兲. E-mail: phillip.paevere@csiro.au
2 behavior of timber systems 共Foliente 1996兲. For the primary pur-
Team Leader and Principal Research Scientist, CSIRO Manufactur-
ing and Infrastructure Technology, P. O. Box 56, Highett, Victoria 3190, pose of obtaining detailed response data needed for the validation
Australia. E-mail: greg.foliente@csiro.au of three-dimensional models of light-frame buildings, quasi-static
3 testing has been employed. This allowed measurements of the
Associate Professor, Dept. of Wood and Paper Science, North
Carolina State Univ., Raleigh, NC 27695. E-mail: bo – kasal@ncsu.edu distribution of the reaction forces underneath the walls, and the
Note. Associate Editor: David V. Rosowsky. Discussion open until displaced shape of the test house in a degree of detail that has not
February 1, 2004. Separate discussions must be submitted for individual been previously obtained. This set of data has a unique value
papers. To extend the closing date by one month, a written request must
because it provides a check of internal forces within whole struc-
be filed with the ASCE Managing Editor. The manuscript for this paper
was submitted for review and possible publication on October 30, 2001; ture models of light-frame buildings. Whole structure model vali-
approved on February 8, 2002. This paper is part of the Journal of dation, based on global response data alone, may lead to false
Structural Engineering, Vol. 129, No. 9, September 1, 2003. ©ASCE, confidence in the capability of these models and even lead to
ISSN 0733-9445/2003/9-1275–1284/$18.00. erroneous conclusions when used in analysis.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1275

J. Struct. Eng. 2003.129:1275-1284.


Overview of Light-Frame Testing nowa 共1998兲. These and several other dynamic tests are reviewed
in detail in Fischer et al. 共2001兲. In more recently reported dy-
Resistance to lateral loads in a light-framed structure is com- namic testing, Fischer et al. 共2001兲 conducted an extensive pro-
monly provided by shear wall or diagonal bracing systems, con- gram of shake table tests at the University of California, San
nected through horizontal diaphragms such as the roof and floors. Diego, on a two-story woodframe house as part of the Consortium
Foliente and Zacher 共1994兲, Barton 共1997兲, and Gad 共1997兲 have of Universitites for Research in Earthquake Engineering
reviewed literature on experimental testing of light-frame wood 共CUREE兲-Caltech Woodframe project 共Hall 2000兲. Their objec-
and steel structural systems. Much of this work has focused on tive was to determine seismic performance under different levels
the behavior under reversed-cyclic loading, due to the compli- of seismic shaking and for different structural configurations.
cated characteristics of the hysteresis of such systems. A common Their results showed that a fully engineered woodframe house has
observation from these tests is that the hysteretic behavior of a better seismic performance than a conventionally constructed
subsystem or subassembly is determined by the hysteretic char- house, and that the nonstructural wall finishes considerably stiff-
acteristics of its primary connection 共Dowrick 1986兲. For ex- ened the structure and changed the distribution of vertical reac-
ample, Stewart 共1987兲 and Dolan 共1989兲 reported that the behav- tions in the hold downs, and generally reduced the response level.
Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

ior of shear walls is dominated by the nailed sheathing Apart from the test by Phillips 共1990兲 on a light-frame struc-
connection. As a result of this observation, the vast majority of ture with a regular plan and elevation, no detailed conclusions
cyclic tests have been conducted on connections and subassem- about the load sharing and distribution in light-frame structures,
blies such as shear walls and diaphragms. Recommendations aris- under lateral loading can be drawn from previous full-scale house
ing from these tests typically ignore system effects, and the fact tests. In the work presented in this paper, the complete distribu-
that the actual forces that the subsystem experiences in a real tion of the reaction forces under a nonsymmetrical single-story
structure depends primarily on the geometric and structural char- woodframe house and the displaced shape are determined under
acteristics of the whole building. Boughton’s 共1988兲 summary of static and static-cyclic loading. Without the detailed measure-
a full-scale house and isolated wall testing showed that boundary ments of the distribution of the reaction forces underneath the
conditions in the wall test ‘‘can influence the stiffness, ultimate frames and the walls, it would be extremely difficult to validate
load, and failure characteristics of components under test.’’ He whole structure models of light-frame buildings because of inher-
also concluded that the commonly accepted assumptions regard- ent structural redundancy and indeterminacy in these systems.
ing load sharing and ductility were invalid, and in some cases Just because a whole building model matches the global response
unconservative, but that structural redundancies which were usu- parameters of a test house does not necessarily mean that this is
ally ignored compensated for this. Full-scale whole house testing an appropriate model because internal forces 共in both value and
is needed to properly understand the system behavior and prove distribution兲 could be wrongly matched. The set of data obtained
the validity of isolated component test results and their interpre- in the experiments presented herein provides a check of internal
tations 共Boughton 1988; Foliente and Zacher 1994兲. forces within whole structure models of light-frame buildings.
Testing of full-scale light-frame structures dates back to the
1950’s in Canada, where one of the first reported experiments was Description of Test House
conducted by Dorey and Schriever 共1957兲, on a single-story stud-
framed house. Since then, static testing has been conducted by The design and layout of the woodframe test house, which is the
Tuomi and McCutcheon 共1974兲, Reardon 共1986, 1990兲, Stewart subject of this presentation, were chosen to represent a typical
et al. 共1988兲, Reardon and Mahendran 共1988兲, Phillips 共1990兲, example of a North American single-story stud-framed house.
and Reardon and Henderson 共1996兲, among others. Many of the The specifications were jointly developed by the National Asso-
studies have pointed toward the importance of the interactions ciation of Home Builders Research Center 共NAHBRC兲 in the
between the subassemblies of the light-frame building and the USA and researchers at North Carolina State University. Slight
load-sharing mechanisms. The effect of the boundary conditions, modifications were made to the construction materials used 共Aus-
the contribution of the nonstructural components, and the loading tralian equivalents to US materials were used兲 and the floor plan
mechanisms, on the behavior of individual components in the 共the size of the rooms were adjusted slightly to match the strong
whole building are also highlighted in some of the studies. Phil- floor anchorage layout in the full-scale testing facility兲. For the
lips 共1990兲 and Phillips et al. 共1993兲 recorded the reaction forces purposes of the experiments described herein, the house was as-
beneath individual walls under static–cyclic loading, to address sumed to be built on a concrete slab foundation and to consist
the problem of load-sharing within a nonlinear system. The main only of the ‘‘structural’’ elements, and the gypsum board lining.
conclusions from this work were: 共1兲 the wooden roof behaved as The following ‘‘nonstructural’’ elements were not installed in the
a rigid diaphragm and contributed significantly to lateral load test house: Tapes and joint compound in the interior joints be-
sharing among the shear walls; 共2兲 up to 20% of the lateral load tween intersecting walls; tapes and joint compound or cornices
was carried by the transverse walls at small load levels; 共3兲 the between the wall and ceiling lining; Windows and doors and their
transverse walls did not contribute to load sharing among the frames; interior and exterior door and window trims; and exterior
shear walls; and 共4兲 the stiffness contributions provided by indi- wall and roof cladding. The experiment was designed in this way
vidual layers of sheathing were directly additive. These experi- so that it best satisfied the implicit assumptions which are the
ments offered important insight into load sharing in a full-scale basis of the FE modeling strategy, and so that the behavior of the
building with regular geometry. main structural elements 共which is the focus of engineering de-
Dynamic testing of full-size light-frame structures has been far sign兲 would be the basis for any conclusions. Although finishes
less common than static testing, presumably due to the high costs were not included, the house shell was constructed such that it
involved. However, since the 1995 Hyogo-Kanbu earthquake, could be clad externally with vinyl or timber siding on the walls
several Japanese-style woodframe structures have been shake- and vinyl, slate or asphalt tiles on the roof.
table tested in Japan. These include tests by Kohara and The house walls were constructed from Australian radiata pine
Miyazawa 共1998兲, Ohashi et al. 共1998兲, and Yamaguchi and Mi- with an average density 共at 12% moisture content兲 of 550 kg/m3

1276 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003

J. Struct. Eng. 2003.129:1275-1284.


Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Diagram of the test house

共southern yellow pine could be considered a US equivalent兲. Wall Fig. 2. Floor plan, coordinate system and wall numbering notation of
frames were assembled from 35⫻90 mm studs at 400 mm cen- the test house 共dimensions in m兲.
ters, without blocking, double 35⫻90 mm top plate and a 45⫻90
mm bottom plate. Studs were endnailed to top and bottom plates
using 3.05⫻75 mm machine-driven nails 共two per end兲. Hold- Experimental Program
down restraints 共i.e., on end studs兲 were not used in any of the
walls. The bottom plate of all walls was anchored to a grid of load All testing described in this paper was carried out at the Com-
cells using 12.5 mm bolts and plate washers at approximately 1 m monwealth Scientific and Industrial Research Organization
spacings. The internal lining was a 13 mm gypsum board, ori- 共CSIRO兲 Division of Manufacturing and Infrastructure Technol-
ented horizontally, and attached by self-drilling screws at 300 mm ogy at their full-scale testing facility in Melbourne, Australia.
spacings. The gypsum lining was not connected with tape and Two different types of experiments were conducted on the test
joint compound in the corners between the ceiling and the roof house: ‘‘Low level’’ and destructive. In the low-level tests, small
panels, or between intersecting walls. However, in-plane joints in point loads were applied laterally at the ceiling level to determine
the gypsum lining within the wall and ceiling panels were joined the extent and the distribution of the reaction forces within the
with tape and joint compound. Wall bracing was provided by 9.5 structure. To avoid irrecoverable damage, the total house dis-
mm external plywood sheathing, attached with 2.87⫻50 mm placement in these tests was targeted to be less than 2 mm. These
machine-driven nails at 150/300 mm spacings 共i.e., 150 mm on tests were load controlled at load values roughly representative of
perimeter, 300 mm internally兲. The roof was constructed of pre- wind design values. In the destructive test, a lateral cyclic load up
to ⫾120 mm, was applied statically to the eastern side of the
fabricated trusses 共pine兲 laid out without blocking on 600 mm
house, in line with the north–south direction 共see Fig. 2兲.
spacings, sheathed with 12.5 mm plywood, and fastened using
2.87⫻50 mm machine-driven nails at 150/300 mm spacings.
A schematic diagram of the house is given in Fig. 1. The plan
Load and Displacement Response Measurement
dimensions, the coordinate system, and the wall notation used are
shown in Fig. 2, and the wall configurations and sheathing are
Reaction forces were measured underneath the bottom plate of the
shown in Fig. 3. Individual wall lengths can be derived from Fig. house throughout the static and destructive experiments, through
2, and are drawn to the same scale in Fig. 3. Note that the wall a grid of 60 load-cell units spaced approximately 1 m apart. Each
height is 2.4 m and the height of the roof ridge is 4.0 m. The four of these load-cell units is comprised of three identical unidirec-
north–south oriented walls in the house, walls W1–W4, were tional load cells coupled together to measure forces in three prin-
chosen to represent a range of different but typical wall configu- cipal directions, X, Y, and Z 共directions are shown in Fig. 2兲. The
rations used in a woodframe house 共see Figs. 2 and 3兲. The non- load-cell units are held in place and attached at the base of the
symmetrical L-shaped plan of the house was chosen for two rea- strong floor of the laboratory via a heavy metal frame. The tops of
sons: 共1兲 to extend and complement previous work done by the load cells are connected to a steel channel, which is intended
Phillips 共1990兲 on a symmetrical plan house; and 共2兲 to ensure to simulate a concrete slab foundation. The sill plates of the walls
some torsional component to the response and, hence, engage- are connected to the load cells via this steel channel, using 12.5
ment of all the structural subsystems and load-sharing mecha- mm bolts and plate washers spaced at approximately 1 m inter-
nisms, including the roof. vals. Full details of the load-cell system and connection to the test

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1277

J. Struct. Eng. 2003.129:1275-1284.


Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Wall configurations and sheathing arrangement of the test


house

house are given in Paevere 共2002兲. The entire load-cell grid has a
stiffness of 350 kN/mm in the X direction, 325 kN/mm in the Y
direction, and 465 kN/mm in the Z direction. This means that the
entire load measurement system was about ten times stiffer than
the initial stiffness of the whole house in the direction of loading Fig. 4. Example of displaced shape and distribution of reaction
共see section entitled ‘‘Destructive Test’’兲. It should be noted that forces from low-level tests
the load measurement system is not as stiff as a concrete founda-
tion would be, and that this may have some effect on the load
distribution among the shear walls, when compared to a ‘‘real’’
house. The magnitude of this effect can be quantified using the
calibrated FE model of the house by including the structural char- buckles and threaded bars with load cells installed. The complete
acteristics of the load measurement system in the FE model. distribution of the reaction forces underneath the bottom plate and
Loads in the actuators 共i.e., the applied loads兲 and displace- the displaced shape of the house were measured for each loading
ments were also measured throughout the tests. Displacements configuration. The applied loads were chosen at a suitably low
were measured in the three principal directions, at the intersection level 共all less than 15 kN兲, so as not to induce any significant
of all walls at both the top and bottom plate level, and also at the inelastic response in the house which may have caused damage or
extreme corners of the roof. Wall W4 was more extensively in- permanent displacements and compromised further experiments
strumented than the other walls during the destructive test, with on the house. The results from one of the low-level tests are
diagonal displacements measured across all of the openings. shown diagrammatically in Fig. 4. In this example, a single point
load of 4.8 kN is applied to wall W2 only, at the top plate level,
in the X direction. The displaced shape and applied loading is
Low-Level Testing shown at the top of Fig. 4 and the distribution of the X-, Y-, and Z-
direction reaction forces are given underneath. It can be seen from
Before the low-level tests were conducted, the dead load distribu- the distribution of the X-direction reaction forces, that a signifi-
tion of the house was measured, without any applied loading. The cant amount 共50%兲 of the applied load is resisted by the parallel
total sum of the vertical reaction forces due to self-weight was nonloaded walls. The vertical reactions 共Z direction兲 are also dis-
50.775 kN. Next, a series of small point loads were applied at the tributed to the nonloaded walls. The applied load is distributed to
ceiling level at various places throughout the house. In these static the nonloaded walls mainly through the roof and ceiling dia-
tests, the loads were applied with simple apparatus such as turn- phragm, since the out-of-plane reactions in the transverse walls 共X

1278 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003

J. Struct. Eng. 2003.129:1275-1284.


Table 1. Load-Sharing in Low-Level Tests Under Single Point Load
Load resisted by each wall 共N兲/percentage of applied load
Test Loaded wall Applied load 共N兲 Direction W1 W2 W3 W4 W5 W678 W9
1 W1 2861 N–S 610 1046 674 461 19 23 29
21% 37% 24% 16% 1% 1% 1%
2 W2 4676 N–S 333 2303 957 955 12 64 52
7% 49% 20% 20% 0% 1% 1%
3 W3 4886 N–S 53 305 3751 490 86 56 144
1% 6% 77% 10% 2% 1% 3%
4 W4 4835 N–S 67 438 1195 2995 16 73 51
1% 9% 25% 62% 0% 2% 1%
5 W8 5004 E–W 52 45 219 278 711 2496 1203
1% 1% 4% 6% 14% 50% 24%
6 W5 3795 E–W 21 61 137 50 2784 731 12
Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

1% 2% 4% 1% 73% 19% 0%
Note: N—north, S—south, E—east, and W—west.

direction兲 are comparatively small. The in-plane reactions in the would be likely under the severe loads and displacements which
external cross walls 共Y direction兲, indicate the torsional nature of were to be imposed during the destructive test. Such localized
the response. failure mechanisms would render the destructive experiment
The results of all of the low-level tests, in which a single meaningless.
concentrated load was applied to one of the walls, are summa- The loading mechanism and protocol are not specifically rep-
rized in Table 1. The amount of load resisted by each of the four resentative of either seismic or wind loading. This is because the
main walls, when loaded individually, is compared to the amount major objectives of this experiment were to examine the load
of load resisted by the rest of the structure. Table 1 shows that distribution and load-sharing mechanisms in a nonsymmetrical
between 19 and 78% of the applied load can be shared between light-frame structure, and to collect data for use in the validation
the walls in the house under simple concentrated loading. As for of analytical models of light-frame structures, including a FE
the example shown in Fig. 4, the applied load in these tests is model which incorporates three-dimensional system behavior,
generally distributed to the nonloaded walls through the roof and and is capable of predicting load sharing 共Kasal 1992兲. So, an
ceiling diaphragm. The amount of load sharing depends on the important consideration in determining the loading was to ensure
stiffness of the loaded wall, and the configuration and stiffness of the full engagement of all of the load-sharing mechanisms, espe-
the surrounding structure. The amount of load redistribution ob- cially the roof and ceiling diaphragm. Hence, loads were applied
served in these low-level experiments, as given in Table 1, repre- to walls W3 and W4 only, at the top plate level, under displace-
sents a lower bound. This is because the applied loads are small ment control. Under this loading regime, the three-dimensional or
and hence all of the structural subsystems may not be fully en- torsional nature of the response in the structure is maximized, and
gaged. It is shown in the section on destructive testing, that the the roof system is fully engaged in distributing the applied loads
load redistribution increases significantly once the house is from the eastern, to the western side of the house. Although this is
pushed to higher displacement levels. It should be noted that the an unrealistic environmental loading scenario, it was desirable for
response of the test house under all of the low-level tests is non- the purposes of validating analytical models which can predict
linear elastic, as indicated by the difference between the tangent load sharing within the structure. Once a model is validated, it can
and 1.0 mm secant stiffness given in Table 2. These stiffness then be used to investigate the behavior of a range of configura-
values were derived from the initial stages of the destructive test-
tions under more realistic environmental loading.
ing, which is described in the following section.
The reversed-cyclic displacement-based protocol used in the
experiment is shown in Fig. 5, and is conceptually based on the
International Standards Organization 共ISO兲 standard for joints
Destructive Test
共Foliente et al. 1998; ISO 2000兲. Displacement control was con-
sidered preferable because of the difficulties associated with a
Experiment Description load-controlled experiment. This would have required the design
After the low-level tests were conducted, the displaced shape and of a multiactuator hydraulic control system, or an impractical
the distribution of reaction forces throughout the structure were spreader beam structure. Cycling of the applied load is particu-
measured under reversed-cyclic lateral loading. Identical static- larly important for capturing the behavior of light-framed struc-
cyclic displacements of up to ⫾120 mm were applied at the ceil- tures, given that they are prone to degradation of strength and
ing level of walls W3 and W4 in the north–south direction using stiffness, and pinching of the hysteresis under cyclic loads. In
servocontrolled screw jacks. The load was distributed along the addition to this, the irregular plan may result in differing re-
top plate of these two walls via a 3 mm metal strap, which was sponses depending on the direction of loading, hence, the reversal
embedded in the top plate. The additional strength provided by of the loading direction is also important.
the steel strap may have had a minor effect on the behavior of the Three different loading rates were used for different levels of
top plate during the destructive test. The strap was considered applied displacement, so that an appropriate density of data could
necessary, however, to provide a loading mechanism which would be obtained within one working day:
allow applied loads to be distributed along the length of the walls • For displacements up to 10 mm a rate of 2 mm/min,
in both the ‘‘pull’’ and ‘‘push’’ directions, and to avoid localized • For displacements between 11–50 mm a rate of 15 mm/min,
failure mechanisms from occurring at the loading points, which and

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1279

J. Struct. Eng. 2003.129:1275-1284.


Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Displacement-based loading protocol used in destructive test Fig. 6. Global hysteresis response of whole house during destructive
test

• For displacements between 51–120 mm, a rate of 25 mm/min.


The displacements were applied using two servocontrolled motor- vere loads and displacements applied to the building in this ex-
ized screw jacks, attached to the top plate of walls W3 and W4. periment and the global response of the structure, this type of
Load and displacement readings were taken once every 2 s, which single-story building should easily resist any lateral design load
was sufficiently fast given the slow loading rates used. without danger of collapse 共the design wind load is approximately
30 kN and the total mass of the structure is around 50 kN兲. How-
ever, this exact response cannot necessarily be extrapolated to
Global Hysteresis Response
other situations based on the results of this experiment alone.
In Fig. 6, the global hysteresis response of the whole building, in
the direction of loading 共X direction兲 is presented. The value on
Wall Hysteresis Responses
the ‘‘Load’’ axis was calculated by summing all of the reaction
forces in the X direction in the load cells under the bottom plate, Because of the extensive instrumentation used in the experiments,
at a given point in time 共same as the applied loading兲, and then it was possible to determine the behavior of the different struc-
repeating this for all points in time during the experiment. The tural subassemblies within the house. In Fig. 7, the in-plane hys-
value on the ‘‘displacement’’ axis is the centroidal displacement teresis response of each of the different walls of the house is
of the house in the direction of loading. Note that the global shown separately. The out-of-plane cross-wall contributions of
hysteresis response in Fig. 6 contains no information about the walls W5 to W9 共X direction兲 are also shown, and are lumped into
out-of-plane behavior of the house. The hysteresis is character- a single subsystem, as they are effectively acting together, and are
ized by ductile behavior and a large displacement capacity. Hys- negligible when plotted separately. The in-plane contributions of
teretic ‘‘pinching’’ at the origin, and degradation of the strength walls W6, W7, and W8 共Y direction兲 are also lumped into a single
and stiffness under cyclic loading are also apparent. This type of subsystem, as they are effectively acting as a single unit in the Y
behavior is to be expected from a light-framed system connected direction due to the geometry of the wall layout, and are also
with dowel-type fasteners such as nails and screws. negligible when plotted separately. Fig. 7 shows that the hyster-
The initial tangent stiffness of the whole house, in the direc- etic characteristics of all of the walls are similar, and exhibit
tion of the applied loading, is approximately 36 kN/mm. The comparable behavior to the global hysteresis shown in Fig. 6,
secant stiffness at 1.0 mm displacement is 22.5 kN/mm 共see Table except for the doubly sided gypsum clad internal wall W3. Wall
2兲, indicating that the load–displacement curve is nonlinear from W3 is initially stiffer than the other walls, but is more brittle and
the origin. As shown in Fig. 6, the total capacity of the house, in loses capacity at a much faster rate during the test. The in-plane
the direction of loading, is around 100 kN with the maximum initial stiffness and maximum load values for each of the sub-
capacity being reached at about 30 mm displacement. The house systems is given in Table 2. Note that the stiffness values for the
is still resisting 80 kN at 80 mm displacement and 70 kN at 110 internal cross–walls 共i.e., W6 –W8兲 in Table 2 were estimated
mm displacement, indicating that this type of construction is based on the perforated shear-wall method 共NAHBRC 2000兲, as
highly ductile under statically applied cyclic loads. Given the se- they could not be reliably determined from the experimental data.

Table 2. Initial In-Plane Stiffness and Capacity Characteristics of Test House and Individual Wall Systems
Structural system
Property Units House W1 W2 W3 W4 W5 W6 –W8 W9
Initial tangent stiffness 共kN/mm兲 36 0.67 4 21 9 12 12 b
16
Stiffness at 1.0 mm displacement 共kN/mm兲 22.5 0.55 3.5 11.5 6 8 9b 11
Maximum load 共kN兲 100 12a 30a 33 50 14a 5a 18a
Displacement at maximum load 共mm兲 32 58a 62a 10 35 18a 4a 11a
Maximum displacement 共mm兲 105 58a 62a 116 117 18a 4a 11a
Load at maximum displacement 共kN兲 70 12a 30a 14.5 15.4 14a 5a 18a
a
Wall
capacity may not have been reached.
b
Estimated value.

1280 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003

J. Struct. Eng. 2003.129:1275-1284.


Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Approximate hysteretic behavior of roof and ceiling


diaphragm during destructive test

only the main section of the roof covering walls W2, W3, and W4
was considered. The data on the load axis are the measured re-
storing force under wall W2. This can be assumed to be the ap-
proximate load if the roof and ceiling diaphragm is assumed to be
the main path for transferring the applied loads from the eastern
共i.e., loaded兲 side of the house, over to the western 共i.e., non-
loaded兲 side of the house. If the contribution of the cross walls to
the load sharing is ignored, then the only path for the applied load
to be distributed to wall W2 is the roof system and, as such, the
reaction under wall W2 can be considered as the approximate
Fig. 7. Hysteresis response of wall systems during destructive test in-plane shear reaction force for the main section of the roof and
ceiling diaphragm. The hysteresis plot in Fig. 8 indicates that the
roof and ceiling diaphragm is flexible but is relatively rigid com-
The damage to the house walls was typical of gypsum and pared to the walls, since the shear distortion is only around 10 to
plywood sheathed woodframe walls, with severe damage ob- 15 mm, compared to a differential displacement between the east-
served in the plywood and gypsum on walls W3 and W4 and mild ern and western walls of around 40 mm. Fig. 8 also indicates the
to moderate damage observed on the other walls. More details of inelastic behavior of the roof system under the applied loading.
the damage status of the different parts of the house can be found This is consistent with the damage observed during the experi-
in Paevere 共2002兲. ment, where there were signs of slight working of the fasteners
Under current practice, the testing and analysis of residential holding the plywood and gypsum on to the roof trusses on the
shear-wall structures are based on the results of isolated wall test- eastern side of the house. It should be noted that the deformation
ing. The results of this experiment go further than this, and give level observed in the diaphragm during the test is larger than they
us an insight into how the shear walls behave when they are part would be under a more uniformly distributed applied loading, and
of a whole structure. Currently, the four main shear-walls does not indicate the likely deformation under environmental
W1–W4 are being tested as isolated walls, the results of these loading.
tests will then be compared to the response when part of the
whole building.
Displaced Shapes
Roof and Ceiling Diaphragm Behavior The displaced shapes of the house, at three stages during the
It is evident from the hysteresis plots of the individual walls in destructive test, are shown in Fig. 9, the undeformed edges are
Fig. 7 that the displacements of the nonloaded walls W1 and W2 also shown for reference. Only the perimeter walls are shown,
are much smaller than the applied displacements on walls W3 and and the roof has been removed for clarity. These plots are snap-
W4. The displacement difference is due to two factors: 共1兲 the shots of the displaced shape, taken at the time of peak displace-
rigid-body rotation of the house under the applied loading; and ment on three selected cycles of displacement in the north direc-
共2兲 the in-plane shear distortion 共or racking兲 of the roof and ceil- tion. Note that the displacements of walls W3 and W4 are equal,
ing diaphragm. An approximation of the in-plane hysteresis char- since the actuators are attached to these two walls and are oper-
acteristics of the roof and ceiling diaphragm is given in Fig. 8. ated in displacement control mode. It can be seen in Fig. 9 that
The data on the displacement axis of Fig. 8 was derived by cal- the major component of the displacement is in the direction of the
culating the in-plane shear distortion, in the north–south direction loading, relatively little out-of-plane or vertical displacement is
of the main section of the roof and ceiling diaphragm covering apparent. At the maximum displacement of 120 mm, the maxi-
walls W2, W3, and W4. The values were derived from roof- mum out-of-plane displacement is 19 mm, which is 16% of the
mounted displacement gauges which measured the X and Y dis- in-plane displacement. The maximum vertical displacement is
placements of the test house at the corners of the roof. Note that around 10 mm.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1281

J. Struct. Eng. 2003.129:1275-1284.


Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Displaced shapes and distribution of reaction forces, at three displacement levels during destructive test

Load Distribution centage of the applied load, which is taken by the out-of-plane
walls is also plotted in Fig. 10. Initially, 4% of the total load is
The distribution of the reaction forces underneath the bottom
taken by the cross walls, but this increases to a maximum of 12%
plate of all the walls in the house, at various stages during the test,
at a displacement of 60 mm. This indicates that the contribution
for the X, Y, and Z direction are also shown in Fig. 9, underneath
of the cross walls to the load sharing increases under inelastic
the displaced shapes. These diagrams are graphical snapshots of
response but is still small compared to the load sharing through
the reaction force data, taken at the time of peak displacement in
the roof and ceiling diaphragm.
the north direction, and correspond to the same points during the
experiment as the displaced shapes plotted above them.
As was the case for the low-level tests, Fig. 9 shows that
Comments on Experiments
significant sharing of the in-plane 共X direction兲 and vertical 共Z
direction兲 loads occurs during the destructive test. Fig. 9 also The experiments and results described herein are based on a semi-
shows that as the amplitude of the applied loading is increased, finished house, which consisted only of the structural elements
the proportion of the applied load redistributed to the parallel
nonloaded walls, W1 and W2, also increases. The behavior for the
Z-direction reactions is similar in this manner. The load redistri-
bution becomes more pronounced as the house is pushed further
into inelastic response, and the load-carrying capacity of walls
W3 and W4 is reduced. A similar trend occurs for the peaks of the
loading cycles in the reverse direction to that shown in Fig. 9.
In Fig. 10, the percentage of the total in-plane load, which is
taken by each wall subsystem, is plotted at the peaks of the load-
ing cycles in the north direction. Here, it is highlighted that the
distribution of the load throughout the structure changes signifi-
cantly during the destructive experiment with the amount of load
resisted by each subsystem depending on the level of applied
displacement. Initially, the stiffest wall 共W3兲 resists the majority
of the load, but as the displacement increases, the more ductile
wall W4 takes over. Walls W1, W2, and the cross walls also take
Fig. 10. Percentage of applied load taken by each wall subsystem
more load as the applied displacement is increased. Similar be-
during destructive test
havior also occurs for loading in the reverse direction. The per-

1282 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003

J. Struct. Eng. 2003.129:1275-1284.


and the gypsum board lining. Inclusion of the nonstructural ele- the house; and Michael Collins of North Carolina State University
ments in the experiment 共i.e., tapes in the corners of the interior for his assistance and advice on various aspects of the project.
panel materials, cornices or similar nonstructural links between
the roof and wall panels, windows, doors, and exterior wall finish兲 References
may alter the load paths at small displacements. The presence of
the nonstructural finishes may also increase the strength and stiff- Barton, A. D. 共1997兲. ‘‘Performance of steel framed domestic structures
ness of the house, and also increase the apparent damage status. subjected to earthquake loads.’’ PhD thesis, Dept. of Civil and Envi-
This could be addressed by testing a finished house to compare ronmental Engineering, Univ. of Melbourne, Australia.
and enhance the results presented here, and so that the results Boughton, G. N. 共1988兲. ‘‘Full-scale structural testing of houses under
cyclonic wind loads.’’ Proc., 1988 Int. Conf. on Timber Engineering,
could be compared against field observations. It would also be
Forest Products Research Society, Madison, Wis., 2-82– 88.
desirable to conduct similar experiments on a light-frame house Dolan, J. D. 共1989兲. ‘‘The dynamic response of timber shear walls.’’ PhD
with a very different configuration than the house described thesis, Dept. of Civil Engineering, Univ. of British Columbia, Vancou-
herein to provide additional data for model validation. ver, BC, Canada.
Dorey, D. B., and Schriever, W. R. 共1957兲. ‘‘Structural test of a house
Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

under simulated wind and snow loads.’’ ASTM Special Technical Rep.
Conclusions No. 210, ASTM, Philadelphia.
Dowrick, D. J. 共1986兲. ‘‘Hysteresis loops for timber structures.’’ Bull.
The experimental results presented herein provide the most de- New Zealand Natl. Soc. Earthquake Eng., 19共20兲, 143–52.
Fischer, D., Filiatrault, A., Folz, B., Uang, C.-M., and Seible, F. 共2001兲.
tailed picture of the reaction forces beneath a nonsymmetrical
‘‘Shake table tests of a two-story woodframe house.’’ CUREE Publi-
light-frame structure under lateral loading ever recorded. They cation No. W-06, CUREE, Richmond, Calif.
have shown that there is potential for significant load sharing and Foliente, G. C. 共1996兲. ‘‘Issues in seismic performance testing and evalu-
redistribution between the main shear resisting walls in the house ation of timber structural systems.’’ Proc., Int. Wood Engineering
under lateral loads, for both pseudoelastic and inelastic response Conf., V. K. A. Gopu, ed., New Orleans, 1-29–36.
conditions. Lateral loads are distributed mainly through the roof Foliente, G. C. 共1998兲. ‘‘Design of timber structures subject to extreme
and ceiling diaphragm system. Commonly used techniques for loads.’’ Progr. Struct. Eng. Mater., 1共3兲, 236 – 44.
lateral load distribution in light-frame structures 共NAHBRC Foliente, G. C., Karacabeyli, E., and Yasumura, M. 共1998兲. ‘‘International
2000兲 do not accommodate redistribution of the loading between test standards for joints in timber structures under earthquake and
the walls in the structure. wind loads.’’ Proc., Structural Engineers World Congress 关SEWC
’98-共CD-ROM兲兴, Elsevier Science, New York, T-222–226.
Under static–cyclic loading in the north–south direction on
Foliente, G. C., and Zacher, E. G. 共1994兲. ‘‘Performance tests of timber
one side of the house, the roof and ceiling diaphragm provided a structural systems under seismic loads.’’ Analysis Design and Testing
robust load–distribution path from the loaded to the nonloaded of Timber Structures Under Seismic Loads—Proc. Research Needs
walls. It effectively behaved as a rigid diaphragm under the ap- Workshop, Greg C. Foliente, ed., Univ. of California Forest Products
plied loading, in comparison with the flexibility of the walls. The Laboratory, Richmond, Calif., 21– 86.
amount of lateral load resisted by the out-of-plane walls was ini- Gad, E. F. 共1997兲. ‘‘Performance of brick-veneer steel-framed domestic
tially 4%, but rose to as much as 12% as the structure was loaded structures under earthquake loading.’’ PhD thesis, Dept. of Civil and
into the inelastic range. The total strength capacity of the house, Environmental Engineering, Univ. of Melbourne, Australia.
in the direction of loading, was around 100 kN with maximum Hall, J. F. 共2000兲. ‘‘The CUREE-Caltech woodframe project.’’ Proc.,
capacity being reached at about 30 mm displacement. The house World Conf., on Timber Engineering 共CD-ROM兲, Univ. of British
Columbia, Vancouver, BC, Canada.
was still able to resist 70 kN at 110 mm displacement, indicating
ISO. 共2000兲. ‘‘Timber structures—Joints made with mechanical
a highly ductile response under static–cyclic lateral loading. fasteners—Quasi-static reversed-cyclic test method.’’ ISO/WD
The set of data obtained in these experiments provides a check 16670.3, Working Draft of ISO TC 165/SC/WG7, International Orga-
of internal forces within whole structure models of light-frame nization for Standardization, Geneva, Switzerland.
buildings and is, thus, of great value in model validation. It is Kasal, B. 共1992兲. ‘‘A nonlinear three-dimensional finite-element model of
recommended that a three-dimensional FE model of the whole a light-frame structure.’’ PhD thesis, Oregon State Univ., Corvallis,
house be fully validated against the experimental results pre- Ore.
sented herein, and against the results of further full-scale experi- Kircher, C., Reitherman, R., Whitman, R., and Arnold, C. 共1997兲. ‘‘Esti-
ments on a house 共1兲 which is fully finished; and 共2兲 with a mation of earthquake losses to buildings.’’ Earthquake Spectra, 13共4兲,
different plan configuration. Experimentally validated models can 703–720.
Kohara, K., and Miyazawa, K. 共1998兲. ‘‘Full-scale shaking table tests of
then be used to conduct sensitivity studies for a wide range of
two-story wooden houses.’’ Proc., 5th World Conf. on Timber Engi-
practical house configurations. These are needed to develop the neering, Swiss Federal Institute of Technology, Lausanne, Switzer-
basis for general recommendations for lateral force distribution in land, Presses Polytechniques et Universitaires Romandes, 2-548 –555.
light-frame buildings. Leon, R. T., and Deierlein, G. G. 共1996兲. ‘‘Considerations for the use of
quasi-static testing.’’ Earthquake Spectra, 12共1兲, 87–109.
NAHB Research Center. 共2000兲. Residential Structural Design Guide:
Acknowledgments 2000 Edition—A State-of-the-Art Review and Application of Engi-
neering Information for Light-Frame Homes, Apartments and Town-
The work presented herein has been jointly funded by the CSIRO houses, U.S. Department of Housing and Urban Development, Wash-
ington, D.C.
and NAHB. The writers are grateful to the following people for
Office of Emergency Services and EQE International. 共1995兲. ‘‘The
their contributions to the work: Jay Crandell of NAHBRC, for Northridge Earthquake of January 17, 1994: Report of Data Collec-
providing the test house details and advice on various aspects of tion and Analysis.’’ Governor’s Office of Emergency Services, Sacra-
the experimental program; Lyndon Macindoe, Craig Seath, and mento, Calif.
Rod Banks of CSIRO, for the design of the experimental instru- Ohashi, Y., Sakamoto, I., and Kimura, M. 共1998兲. ‘‘Shaking table tests of
mentation systems and assistance in construction and testing of a real scale wooden house subjected to Kobe earthquake.’’ Proc., 5th

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003 / 1283

J. Struct. Eng. 2003.129:1275-1284.


World Conf. on Timber Engineering, Swiss Federal Institute of Tech- two-story test house.’’ Proc., Int. Wood Engineering Conf., V. K. A.
nology, Lausanne, Switzerland, Presses Polytechniques et Universi- Gopu, ed., New Orleans, 4-313–319.
taires Romandes, 2-556 –563. Reardon, G. F., and Mahendran, M. 共1988兲. ‘‘Simulated cyclone wind
Paevere, P. J. 共2002兲. ‘‘Full-scale testing, modeling, and analysis of light- loading of a Melbourne style brick veneer house.’’ Rep. No. 34, James
frame structures under lateral loading.’’ PhD thesis, Dept. Civil and Cook Cyclone Structural Testing Station, QLD, Australia.
Environmental Engineering, The Univ. of Melbourne, Parkville, Vic- Stewart, A. H., Kliewer, A., Goodman, J. R., and Salsbury, E. M. 共1988兲.
toria, Australia. ‘‘Full-scale tests of manufactured houses under simulated wind load.’’
Phillips, T. L. 共1990兲. ‘‘Load sharing characteristics of three-dimensional Proc., 1988 Int. Conf. on Timber Engineering, Forest Products Re-
wood diaphragms.’’ MS thesis, Washington State Univ., Pullman, search Society, Madison, Wis., 2-97–111.
Wash. Stewart, W. G. 共1987兲. ‘‘The seismic design of plywood-sheathed shear
Phillips, T. L., Itani, R. Y., and McLean, D. I. 共1993兲. ‘‘Lateral load walls.’’ PhD thesis, Univ. of Canterbury, Christchurch, New Zealand.
sharing by diaphragms in wood-framed buildings.’’ J. Struct. Eng., Tuomi, R. L., and McCutcheon, W. J. 共1974兲. ‘‘Testing of a full-scale
119共5兲, 1556 –1571. house under simulated snow loads and wind loads.’’ Research Paper
Reardon, G. F. 共1986兲. ‘‘Simulated cyclone wind loading of a brick ve- FPL234, United States Department of Agriculture, Forest Products
neer house.’’ Technical Rep. No. 28, James Cook Cyclone Structural Laboratory, Madison, Wis.
Downloaded from ascelibrary.org by University of Exeter on 07/19/15. Copyright ASCE. For personal use only; all rights reserved.

Testing Station, QLD, Australia. Yamaguchi, N., and Minowa, C. 共1998兲. ‘‘Dynamic performance of
Reardon, G. F. 共1990兲. ‘‘Simulated cyclone wind loading of a nu-steel wooden bearing walls by shaking table test.’’ Proc., 5th World Conf.
house.’’ Technical Rep. No. 36, James Cook Cyclone Structural Test- on Timber Engineering, Swiss Federal Institute of Technology, Lau-
ing Station, QLD, Australia. sanne, Switzerland, Presses Polytechniques et Universitaires Ro-
Reardon, G. F., and Henderson, D. 共1996兲. ‘‘Simulated wind loading of a mandes.

1284 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2003

J. Struct. Eng. 2003.129:1275-1284.

You might also like