Seismic performance of nailed wood-frame shear walls
Yamaguchi,N.1 Karacabeyli,E.2 Minowa,C.3 Kawai,N.4 Watanabe,K.5 Nakamura,I.6
ABSTRACT
Although earthquakes induce dynamic demands on structures, for simplicity, most seismic design procedures are based on
static analysis. The purpose of this paper is to present data on nailed wood-frame shear walls subjected to static and
dynamic testing schedules. Results from static and dynamic loading tests are compared. Test results indicate that forcedeformation curves are affected by the rate of displacement. The first envelope curve from a rapid Quasi-static reversedcyclic loading test was found to be the closest to the curve from the shaking table test. Slow loading tests using
monotonic, reversed-cyclic protocols and pseudo-dynamic response displacements resulted in smaller maximum strength
values due to time-dependent characteristics of nailed shear walls.
INTRODUCTION
For the purpose of refining seismic design provisions for timber structures, shaking table tests using full size specimens
subjected to previously recorded earthquake motions were performed. Responses from shaking table tests were compared
to those from static or cyclic loading tests with several rates of displacement. This research project was executed under
the collaborating research project between National Research Institute of Earth Science and Disaster Prevention Japan
(NIED), Forintek Canada Corp. and Building Research Institute Japan (BRI).
73
Stud 2326*38*89
2,440
2,845
HD-2A
Plywood
@150
1220*2440*9.5
@300
HD-2A
178
89*89
Specimens were nailed plywood shear walls for
wood-frame construction. Figure-1 shows a nailed
plywood shear wall specimen that consists of two
plywood-panels. Top and bottom of both side-studs
of panels are fastened to sills and top plates by
commercially available hold-down connectors.
Both end studs are composed of double studs.
Plywood is 2440* 1220*9.5 mm size of Canadian
Softwood Plywood. Studs and plates were 38*89
mm size of Canadian S.P.F. These materials are
fastened by nails (L=63mm,D=3mm) with spacing
of 150mm and 300mm for perimeter and
intermediate locations, respectively. Vertical loads
were not applied for specimens.
225 73
SPECIMENS
2,000
2,000
Figure-1: Specimens of Nailed Wood-frame Shear Walls
1
Senior Research Engineer, Department of Structural Engineering, Building Research Institute, MOC, 1 Tachihara,
Tsukuba, Ibaraki, Japan 305-0802
2
Wood Engineering Scientist, Forintek Canada Corp., 2665 East Mall, Vancouver, B.C., Canada V6T 1W5
3
Cooperative Research Officer, National Research Institute of Earth Science and Disaster Prevention, STA 3-1, Tennodai, Tsukuba, Ibaraki, Japan 305-0006
4
Head, Evaluation System Division, Building Research Institute, MOC
5
Research Director for Earthquake Disaster Reduction, Building Research Institute, MOC
6
Researcher, National Research Institute of Earth Science and Disaster Prevention, STA
TEST METHOD
The structural performance of nailed shear walls was determined by six tests shown as follows.
1)
2)
3)
4)
5)
6)
Slow monotonic loading test at BRI
Moderate monotonic loading test at Forintek
Slow reversed-cyclic loading test at BRI
Fast reversed-cyclic loading test at Forintek
Pseudo-dynamic loading test at BRI
Shaking table test at NIED
Details of these tests are shown in Table-1.
Table-1: Test Methods
No.
Name
1
2
3
4
5
6
Slow monotonic
Moderate monotonic
Slow reversed-cyclic
Fast reversed-cyclic
Pseudo-dynamic
Shaking table test
Number Number of
of stories specimens
1
1
1
1
1or 2
1or 2
1
2
1
6
1
1
Motion of specimen
monotonic
monotonic
reversed-cyclic
reversed-cyclic
pseudo-dynamic response
dynamic response
Rate of
Laboratory
displacement
slow
moderate
slow
fast
slow
real rate
BRI
Forintek
BRI
Forintek
BRI
NIED
Loading system at BRI
Loading system in BRI is controlled by a computer. Series of target displacements and positions for taking measurements
are pre-stored in the computer. Actuator moves step by step to the measuring positions, in order to prevent over-shooting
to the positions. Because the data acquisition system requires a few minutes to measure data at each points, the actuator
must stop at every measuring position. While measuring, specimens are in a stationary state for a few minutes. This
condition is called static in Japan. Consequently these tests require long hours to complete, and apparent rate of
displacement of the actuator is very slow. This loading and measuring system was used in monotonic and reversed-cyclic
loading tests and pseudo-dynamic tests at BRI.
Loading system at Forintek
In Forintek loading tests, a much greater rate of displacement was used and measuring were taken without stopping the
test. Rate of displacement was 0.0125 cm/s and 2 cm/s for moderate monotonic and fast reversed-cyclic loading tests
respectively.
Monotonic loading tests at BRI and Forintek
Slow and moderate monotonic loading tests were conducted both at BRI and at Forintek. A load spread beam was
subjected to lateral displacements of an actuator.
Reversed-cyclic loading tests of BRI and Forintek
Slow and fast reversed-cyclic tests were conducted at BRI and at Forintek. These tests used the cyclic displacement
procedure proposed in the draft ISO standard (ISO 1998). Amplitudes of the initial single reversed-cycles were 1.25%,
2.5%, 5%, 7.5% and 10% of the ultimate displacement that was derived as displacement at 80% of the maximum strength.
These were followed by groups of 3 cycles with amplitudes of 20%, 40%, 60%, 80% 100% and increments of 20% of the
ultimate displacement. One specimen was tested at BRI, and 6 specimens were tested at Forintek.
Pseudo-dynamic test at BRI
Pseudo-dynamic test specimen was subjected to response displacements calculated by equation of vibration using input
motions. Whenever the specimen reached each target displacement for measuring, the motion of specimens was stopped,
and loads and displacements were measured. While the motion of specimens was at rest, next target displacement was
simulated by calculation. This calculation uses Newmark’s β methods with the stiffness derived from loads and
displacements of specimens. A pair of shear walls was used for one specimen. Input motion records were initial 15
seconds of the records measured on shaking table test, because pseudo-dynamic test require much longer time to complete
full records. Consequently, pseudo-dynamic test is slowest test among these tests. Damping factor was 2%. One specimen
was tested (Kawai 1998, Watanabe et al. 1998).
Shaking table test by NIED
Shaking Table test was executed using one-dimensional
earthquake simulator of NIED and moving frame (Yamaguchi
and Minowa 1998, Yamaguchi Minowa and Miyamura 2000).
Photo-1 shows a Shaking table test. Figure-2 shows a pair of
specimens and a moving frame on the shaking table. The
moving frame consists of a steel floor and a pair of crosswalls. Steel plates for masses weighted the steel floor. Steel
floor supported by a pair of cross-walls. A pair of shear walls
is set beside the moving frame. Upper beams of a pair of shear
walls and the steel floor are connected using some horizontal
bolts. Because steel floor has vertical long holes (slide joints)
as shown in Figure-2, the weights of steel floor and mass are
not transferred to the shear walls. The weights of steel floor
and mass are transferred to shaking table through the cross
walls. Consequently, the shear walls are subjected to
horizontal forces
of the mass, and the weights of steel floor and mass are not applied
Photo-1: Specimens & Moving Frame
to the shear walls. This new method was developed for this
shaking table test, in order to make same loading condition for shear walls as loading tests in BRI and Forintek. The
weight (W) of the mass was determined by Equation-1.
W=
Pal
C0
− − − − − 1)
Base shear coefficient (C0) for allowable strength is 0.16 in this test. Allowable shear strength (Pal) of this shear wall was
40% of the maximum shear strength of this wall. Input motion record was NS component of strong-motion record
observed in El Centro during 1940 Imperial Valley Earthquake. Scales of input motion records were arranged in
accordance with their maximum accelerations of 0.35G and 0.6G. After El Centro of 0.35G was applied to a specimen, El
Centro of 0.6G was applied next.
C O NC EPT O F SD F
W e ig h t
S D F U S IN G M O V IN G F R A M E
W e ig h t
In er tia F o r c e
In e rtia F or ce
s lid e j oi nts
M O V IN G F R A M E
S P E C IM E N
S P E C IM E N
c r o s s w a ll
S h a k i ng T a b le
S h a k i n g T a b le
S D F F O R S H A K IN G T A B L E T E S T
M O V IN G FR A M E
In e r tia F o r ce
In er tia F o r c e
s tee l fr a m e s u p p o r t
s te e l fra m e s u p p o rt
c ro s s w a ll
S PE C IM E N
h in ge s
c r o s s w a ll
S h a k i n g T a b le
Figure-2: Concept of Shaking Table Test using Moving Frame
hi n g es
Repeatability of Response Motion by Shaking Table
Tests
Two shaking table tests were performed in 1997 and
1998. Figure-3 shows force-deformation angle curves of
response motions obtained in shaking table test 1997 and
1998 using El Centro 0.35G. Forces generated in this
single degree of freedom system are inertia forces
calculated by m( x + y) . This inertia force is equivalent
to the sum of damping force and spring force as shown in
Equation-2. Spring force kx shown in Figure-3 was
calculated by Equation-3. Displacements were transferred
to shear deformation angles by dividing height of
plywood 2440mm.
m( x + y) + cx + kx = 0
kx = −{m( x + y) + cx}
− − − − − 2)
− − − − − 3)
c = 2h km
− − − − − 4)
Force (kN)
RESULTS & DISCUSSION
Test of 1997
Test of 1998
Deformation Angle (rad)
Fugure-3: Force-deformation Angle Curves in 1997
& 1998 tests using El Centro 0.35G
Damping factor (h) was assumed to be 2%. Stiffness (k) was secant stiffness at allowable strength point on forcedeformation angle curves by slow reversed-cyclic test. Forces of force-deformation angle curves indicate spring forces
(kx) of one shear wall shown. All of force-deformation angle curves by shaking table test on this paper were calculated by
these procedures. The results obtained in 1997 and 1998 tests were found to be very close to each other suggesting
excellent repeatability.
Response in collapse
The weight of mass on floor was supported by cross-walls in these tests. When cross-walls were tilted with shear walls,
horizontal component of weight of the mass (P- ∆ forces) is generated on moving frame. Although the shear walls are not
subjected vertical load of the mass, this horizontal force on moving frame are transferred to shear walls. While the shear
walls and the moving frame tilted, P- ∆ forces are subjected to the shear walls. Figure-4 and -5 show force-deformation
angle curves of response in shaking table tests of 1997 and 1998 using El Centro 0.6G. P- ∆ forces (F) calculated in
Equation- 5 are shown by dotted line in Figure-4 and –5. m , g and θ of Equation-5 is mass, acceleration of gravity and
deformation angle of shear walls respectively.
F = mg tan θ
−−−−−5)
Figure-4 and –5 show their maximum strengths cause around 0.03- 0.04 radians. Maximum strength in positive and
negative sides of 1998 test are a little greater than those of 1997 test, and shapes of their maximum strengths points are
slightly different. They lost their horizontal strengths around 0.04 radians. Figure-4 of 1997 test shows the shear wall
came to near-collapse point, but was not collapsed after the shear wall lost its horizontal strength. Photo-2 shows failed
nailed-fasteners between plywood and studs in 1997 test.
Figure-5 of 1998 test shows that the shear wall collapsed after the shear wall lost its strength. Figure-6 shows time history
records of deformation angles for these tests. These two responses in Figure-4 and -5 are close until the response
deformation angles are less than 0.06 radians, a practical limit for all cyclic and monotonic tests. These are also close
before 8 seconds in Figure-6. These responses are very repeatable like as in El Centro 0.35G tests. But after 0.06 radians
and 8 seconds, the shear walls lost their horizontal strengths, and the shear walls start to make different responses. The
end of response of 1998 test in Figure-6 indicates the shear wall collapsed in reality and landed at the steel frame support.
Photo-3 shows collapsed shear wall in 1998 test using El Centro 0.6G.
1997
P-Delta
Photo-2: Failed Nailed-fasteners in 1997 Test
using El Centro 0.6G
Figure-4: Force-Deformation Angle Curves in
1997 Test using El Centro 0.6G
1998
P-Delta
Photo-3: Collapsed Shear Wall in 1998 Test
using El Centro 0.6G
Figure-6:
Figure-5: Force-Deformation Angle Curves
in 1998 Test using El Centro 0.6G
Responses in Case of Collapsed and Nearly but Not Collapsed
in 1997 and 1998 Test using El Centro 0.6G
Effect of Rate of Displacement
Figure-7 shows two kinds of time history records of response displacements obtained in pseudo-dynamic test using El
Centro 0.35G, and in shaking table test 1998. Figure 7 shows that response by pseudo-dynamic test has following
features.
l
Some of peak displacement were greater than those obtained in shaking table test. Responses between 2 to 4 seconds
are very different from those obtained in shaking table test, and positive maximum values during 2 to 4 seconds are
almost same values.
After 6 seconds, centered line of the response displacement with small amplitude has shifted from that obtained in
shaking table test.
Displacement (cm)
l
Time (sec)
Figure-7: Response Displacement by Pseudo-dynamic Test and Shaking Table Test
l
l
Relatively longer time of loading in pseudodynamic test resulted in greater strength
degradation than that in shaking table test. This
phenomenon suggests that longer time of loading
may degrade strength. This phenomenon may be
called strength-degradation due to load duration.
Relatively longer time of loading in pseudodynamic test resulted in greater deformations
than that in shaking table test. This phenomenon
suggests that longer time of loading may increase
deformations. This phenomenon might be called
deformation-gradation (creep deformation).
Force (kN)
Figure 8 shows force-deformation angle curves obtained in these two tests using El Centro 0.35G. Figure-7 and -8
indicate the number of cycles by both tests is same. Figure-8 shows amplitude of displacement in pseudo-dynamic test is
larger than that in shaking table test, but amplitude of force in shaking table test is larger than that in pseudo-dynamic test.
Neglecting the difference between force and displacement, these two tests have almost same condition in terms of number
of cycles and amplitude. Major difference was distinguished when maximum strength points were compared. Maximum
strength in shaking table test was found to be 125% of
that in pseudo-dynamic test. The deformation of
maximum strength point in shaking table test was
found to be 70% of that in pseudo-dynamic test. These
results suggest following features.
Pseudo-dynamic Test
Shaking Table Test
Deformation Angle (rad)
Figure-8: Force-deformation Angle Curves
in Pseudo-dynamic Test and Shaking Table Test
Force-displacement Curves by Six Tests
Figure-9 shows backbone curves of force-displacement curves obtained in six tests. The curve in shaking table test was
obtained in 1998 test using El Centro 0.35G.
The difference between moderate monotonic and fast cyclic tests shows the effect of number of cycles and amplitude in
case of these loading tests. Number of cycles and amplitude caused more strength degradation after the maximum
strength was achieved. The difference between slow monotonic and slow cyclic tests also shows the effect of number of
cycles and amplitude in case of slow loading tests. Again, number of cycles and amplitude caused similar strength
degradation.
The difference between moderate and slow monotonic tests indicates the effect of rate of displacement in case of
monotonic tests. Moderate loadings give more strength than slow loadings. The difference between fast and slow
reversed-cyclic loading tests indicates the effect of rate of displacement in case of reversed-cyclic loading tests. Again,
fast loadings resulted in greater strength than slow loadings.
Although pseudo-dynamic test and shaking table test have almost same condition in term of number of cycles and
amplitudes, force-displacement curve in pseudo-dynamic test was the furthest from that of shaking table test. Fast
reversed-cyclic test gave the force-displacement curve closest to that of shaking table test. Fast reversed-cyclic test uses a
little longer time of loading than shaking table test. But pseudo-dynamic test uses much longer time of loading than the
shaking table test. These results indicate that the displacement rate should be suitable for evaluation of seismic
performance of wood structures.
The fast loading test resulted in smaller ductility than that of slow loading test. This suggested that evaluating
performance by slow loading test might cause over estimation of ductility. Effect of rate of displacement must not be
neglected in evaluation of seismic performance of these structures. Seismic performance should preferably be evaluated
using fast loading test. It is preferable that computer simulation use force-deformation hysteretic models based on fast
loading test results.
Moderate Mono
Slow Mono
Force (kN/m)
Shaking Table
Slow Cyclic
Fast Cyclic
Pseudo
Pseudo
Fast Cyclic
Slow Mono
Slow Cyclic
Shaking Table
Moderate Mono
-150
-100
-50
0
50
100
150
Displacement (mm)
Figure-9: Force-displacement Curves by Six Tests
CONCLUSIONS
In order to evaluate the seismic performance of nailed wooden-frame shear walls, shaking table tests, moderate and fast
monotonic tests, and slow and fast cyclic tests and a pseudo-dynamic test were performed. We made the following
observations;
l
l
l
Pseudo-dynamic test resulted in a smaller maximum strength and greater displacements than shaking table test. Slow
cyclic test followed a similar trend.
Fast reversed-cyclic test results were found to be closest to those obtained in shaking table test.
The force-deformation behaviors of nailed wood-frame shear walls were influenced by rate of displacement and the
number of cycles and their amplitudes.
ACKNOWLEDGEMENTS
The assistance and contribution of Japan 2 by 4 Home Builders Association are greatly appreciated.
REFERENCES
ISO 1998, “Timber structures-Joints made with mechanical fasteners- Quasi-static reversed-cyclic test method”, Draft
standard, ISO-TC165 Timber Structures Working Group 7, February 10, 1998
Kawai,N. 1998. “Pseudo-dynamic test on shear walls”, Proceedings of The Fifth World Conference on Timber
Engineering (5WCTE), EPF Lausanne, Montreux, Switzerland, August, 1998
Yamaguchi,N., Minowa,C. 1998. “Dynamic Performance of Wooden Bearing Walls by Shaking Table Test”, Proceedings
of Fifth World Conference on Timber Engineering (5WCTE), EPF Lausanne, Montreux, Switzerland, August, 1998
Yamaguchi,N., Minowa,C.,and Miyamura,M. 2000. “Seismic Performance of Wooden Shear Walls on Dynamic
Condition”, 12th World Conference on Earthquake Engineering (12WCEE), Auckland, New Zealand, Jan. 2000.
Watanabe,K., Kawai,N., Yamaguchi,N., Minowa,C. 1998. “Seismic Performance Testing and Prediction of Wood
Construction ”, Structural Engineers World Congress (SEWC), San Francisco, July 1998