You are on page 1of 7

NONLINEAR SHEAR-WALL ANALYSIS

By Maurice W. White· and J. Daniel Dolan 2

ABSTRACT: A finite-element program, WALSEIZ, capable of perform~ng nonlinear analysis of a timber shear
wall subjected to monotonic or dynamic loads is presented. Eac~ wall IS ~ompose~ of four elements: a b~am
element to model the framing, a plate element to model the sheathmg, nonhnear spnngs to model the sheat~mg­
to-framing connectors (the 10ad-di~~lacemen~ properties of whi~h vary dependl.ng on wheth~r monotonic or
cyclic loads are applied), and a blhnear spnng to model beanng between adJ~cent sheathmg panels. The
program can compute displacements at each of the nodes, forces, and stresses In each of the elements as a
function of applied load for monotonic analysis and as a f~nction of time for dynamic analys!s. R~sults fro!U
the program are compared with experimental data to vahdate the program. The program IS bemg ~sed In
investigations on the response of shear walls subjected to dynamic loads such as earthquakes and hurrtcanes.
Downloaded from ascelibrary.org by Columbia University on 02/02/15. Copyright ASCE. For personal use only; all rights reserved.

INTRODUCTION resist lateral loads, such as those resulting from earthquakes


and wind. They also provide some resistance to vertical load-
The destruction that occurred during the earthquake in Los ing and act as partitions in a building. A typical timber shear
Angeles demonstrates the importa~ce ~f the proper d~s.ign wall, shown in Fig. 1, has three main components: the fram-
of structures located in regions wIth hIgh seismIC actIvIty. ing, sheathing, and sheathing-to-framing connectors. In North
Cases of structures being separated from foundations due to America, the framing usually consists of 38 mm X 89 mm (2
insufficient anchorage connections, collapse of entire wall sec- x 4 in nominal) lumber. It is covered with a sheathing ma-
tions and excessive architectural damage (i.e., cracks) oc- terial, which is usually made of plywood, oriented strand-
curr~d because of the earthquake. Unfortunately, this is not board (OSB), or another structural panel product. The
the first case of widespread structural damage resulting from sheathing is attached to the framing with connectors. Tra-
an earthquake. Similar cases of damage occurred in San Fran- ditionally, these connectors have been dowel-type fast~ners
cisco in 1989, Mexico City in 1985, and Anchorage, Alaska such as nails and staples. Recently, however, elastomenc ad-
in 1964. hesives have been used as a connector material with some
Timber shear walls usually provide the lateral resistance success and some potential problems. Nails are used along
for low-rise timber structures subjected to dynamic loads. with the adhesive, but they only hold the sheathing in place
Earthquake disasters indicate that more ~etaile? in~ormation until the adhesive cures.
on timber shear walls subjected to dynamic loadmg IS needed. The typical deflection configuration for a shear wa~1 su:b-
However, the volume of research on the dynamic behavior jected to monotonic loading follows the pattern shown m FIg.
of shear walls is minimal, primarily because of the expense 2, with the framing distorting into a parallelogram and !he
involved with dynamic experimentation. Due to the lack of sheathing rotating and remaining rigid. The framing resIsts
information on the dynamic response of timber shear walls, most of the out-of-plane bending and some vertical load,
provisions in the design codes are often based on static data while the sheathing resists most of the in-plane shear. The
and engineering judgement (which does not necessarily mean connectors play an important role in determining the shape
that every building will fail when subjected to dynamic load- of the load-deflection curve of the wall and in resisting the
ing). external load by dissipating energy as the nails distort (Tuomi
In order to gain more insight into the performance char- and McCutcheon 1978). The in-plane strength, which is a
acteristics of timber shear walls, a computer program, WAL- measure of the magnitude of load that a wall can resist before
SEIZ, has been developed. The program is capable of per- a limit state occurs, and stiffness, which is the relative de-
forming both static and dynamic analysis of a shear-wall model formability of a material under load (Davis et al. 1982), are
and of calculating the resulting displacements, forces, and the most important characteristics in the determination of the
stresses. The benefits of the program are that it provides a adequacy of a wall subjected to monotonic lo~ding. .
more economical means of performing dynamic analysis on A wall subjected to dynamic loads has a typIcal defleCtion
shear walls and a more efficient means of analyzing different pattern similar to that of a wall subjected to monotonic lo~~s,
wall sizes and configurations than experimental testing does, but inertia effects are introduced into the system and ductlhty
and it also allows for detailed information on load paths, along with stiffness become factors in the wall response. Duc-
internal forces, and anchorage requirements to be deter- tility is desirable in walls subjected to dynamic loads because
mined. it prevents the loads from becoming excessively high. A ma-
BACKGROUND
Shear walls are one of the major components of a low-rise
timber structure. The primary purpose of these walls is to
'Asst. Prof., Dept. of Wood Sci. and Forest Products, Virginia Poly-
technic Inst. and State Univ., Blacksburg, VA 24061-0323; formerly,
Grad. Asst.. Dept. of Civ. Engrg., Virginia Polytechnic Inst. and State
Univ.. Blacksburg, VA.
'Asst. Prof.. Dept. of Wood Sci. and Forest ProdUClS, Virginia Poly-
technic Inst. and State Univ., Blacksburg, VA.
Note. Associate Editor: Steven M. Cramer. Discussion open until
April I. 1996. To extend the closing date one monlh, a written request
must be filed with the ASCE Manager of Journals. The manuscript for
this paper was submitted for review and possible publication on June
20, 1994. This paper is part of the Journal o/Structural Engineering, Vol.
IZI, No. 11, November, 1995. ©ASCE, ISSN 0733-9445/95/0011-1629-
1635/$2.00 + $.25 per page. Paper No. 8695. FIG. 1. Primary Components of Shear Wall

JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1995/1629

J. Struct. Eng. 1995.121:1629-1635.


".. .
)."..........................
[.:, =::=~~ ..;,.:.:.:.:.;:...;.;..,;;..:..:..:;.: would reduce the cost of investigating various aspects of wall
performance.
:: ii
•• if
II
:: ii
II
:: if
••
:: . :. 11 There have been some attempts to model the response of
""
••

t. ",.
••
.
II

..
"
"
shear walls. Foschi (1977), Falk and Itani (1989), and Kasal
II • "
and Leichti (1992) developed finite-element models to sim-

I I I I :/
ulate the response of shear walls subjected to monotonic loads.
The model developed by Foschi (1977) contained four ele-
ments (cover, framing, frame connection, and cover-to-frame
: :: i ;i elements) and nonlinear connector behavior. Falk and ltani
! Ii . Ii !! (1989) developed a nonlinear finite-element model that ac-
.; Ii ! Ii Ii counts for the distribution and stiffness of fasteners connect-
I; ii : :: !i
:' :: J ii :: ing the sheathing to the framing. Kasal and Leichti (1992)
:
i if
:: :
; Ii
::
ii
.t
developed a nonlinear single degree-of-freedom finite-ele-
! iF i if if ment model of a shear wall for use as a substructure element
! Ii ; i; ii in a three-dimensional analysis model. Stewart (1987), Filia-
Downloaded from ascelibrary.org by Columbia University on 02/02/15. Copyright ASCE. For personal use only; all rights reserved.

i Ii ; Ii Ii trault (1990), and Dolan (1989) developed finite-element


i~ !i :, :: ::
;.~ :.: l.: :.t......•......•:-i models to simulate the response of shear walls subjected to
to........ ....•.•........ .......•....•
- dynamic loads. However, the models developed by Stewart
(1987) and Filiatrault (1990) were single-degree-of-freedom
••••••• Fnming models and could only examine the overall wall response.
-Sheathing The model developed by Dolan (1989) simulated the plate,
framing, and connector elements but the analysis time was
FIG. 2. Typical Deformation Configuration of Shear Wall extensive and the program lacked the capability to calculate
forces or stresses.
tenal or system that is truly linear-elastic will experience higher PROGRAM DEVELOPMENT
loads up to the point at which failure occurs, while a system
with substantial ductility will experience higher loads only up The program WALSEIZ is a culmination of the efforts
to a yield load. Therefore, the loads experienced by the duc- these programmers. The program is a modification of the
tile system will be lower than those of the elastic system after program developed by Dolan (1989) and utilizes the finite-
the yield point is reached if the displacements and initial element method for analysis. The modifications include: a
stiffnesses are equal. These characteristics of ductility and reduced number of degrees of freedom in the plate, sheath-
lower yielding are incorporated in the seismic design. How- ing-bearing connector, and sheathing-to-framing connector
ever, walls with excessive ductility are subject to extensive elements to reduce the analysis time; the capability for mon-
displacements and costly architectural damage during earth- otonic analysis (load control); the capability of calculating
quakes of high magnitude. Stiffness prevents excessive dis- forces and stresses; and expansion of the program to accom-
placements from occurring in the wall and thereby reduces modate the analysis of larger walls with and without openings.
the extent of architectural damage. Therefore, a balance be- Assumptions in the model include: hinged connections be-
tween stiffness and ductility is desirable in walls subjected to tween the top and bottom plates and the studs; framing ele-
dynamic loads. ments that are homogeneous and isotropic; sheathing thick-
The connectors are also important in the response of the ness that is sufficient so that out-of-plane buckling is not
wall subjected to dynamic loading because they dissipate en- significant; and sheathing elements that are orthotropic.
ergy through hysteretic damping. Therefore, connectors that Four elements are used to model timber shear walls in the
produce large hysteresis loops can provide substantial energy- program. They are a framing element, which models the lum-
reducing mechanisms. Friction damping at the connectors, ber framing of the shear wall; a sheathing element, which
which occurs between the two members joined together, is models the panel sheathing; a sheathing-to-framing connector
also helpful in dissipating some of the energy imparted to the element, which models the dowel-type or adhesive fasteners
system. used to attach the sheathing to the framing; and a sheathing-
The factors important in determining the performance of bearing element, which prevents adjacent sheathing elements
shear walls subjected to dynamic loads, a combination of from numerically overlapping one another. Each element was
ductility and stiffness, are different from factors important derived so that compatibility and equilibrium are attained.
for walls subjected to static loads. However, much of the The framing element is a common two-dimensional, line-
current seismic design code is not based on data determined arly elastic, six-degree-of-freedom beam element with a linear
from dynamic testing. It is, however, based on data from displacement field in the axial direction and a cubic displace-
static tests of shear walls and by relating the performance of ment field for the deflections perpendicular to the beam axis.
buildings constructed with other materials to wood when sub- The formulation of the equation of motion for the framing
jected to natural hazard loads such as earthquakes and hur- element is found in many finite-element texts, including that
ricanes. This is done because of the lack of structural com- of Cook et al. (1989).
ponent test data for walls subjected to cyclic or dynamic loads. The sheathing element is an eight-degree-of-freedom, or-
The only dynamic tests on timber shear walls that the writers thotropic, linear-elastic, rectangular, plane stress element with
are aware of are seismic shear-wall tests performed by Stewart translations in the global X- and Y-directions at each of the
(1987), Dolan (1989), and Foschi and Filiatrault (1990), and four corner nodes. It has a linear displacement field in each
the cyclic shear wall tests performed by Stewart (1987), Gray direction and follows Kirchhoff's plate theory. The formu-
and Zacher (1988), Dolan (1989), and Hanson (1990). There- lation of the equation of motion for the framing element is
fore, it is important that further research be performed in found in many finite-element texts, including that of Cook et
this area. Unfortunately, this often becomes prohibitive be- al. (1989).
cause of the high cost of performing experimental dynamic The sheathing-to-framing connector element, which con-
tests. However, development of a computer program that can sists of two independent nonlinear springs, models the resis-
adequately simulate shear walls subjected to dynamic loads tance that a connector would provide when displacement oc-
1630 I JOURNAL OF STRUCTURAL ENGINEERING I NOVEMBER 1995

J. Struct. Eng. 1995.121:1629-1635.


P-
I ~ P' - I" Sheal!llnl;
IF"I = (Po + Kzlamaxl) [1 - exp (-K~~maxl)]
Connector Framing
- K 3 (llil - lamaxl) for lal > lamaxl
~ ~
(lb)
where Ii = connector displacement; Po = Y-intercept of the
K z line; K o = slope of the curve at the origin; K 2 = slope
F of the curve at the maximum displacement; limax = displace-
}t ment at which the maximum load occurs (after which the load
capacity of the connector decreases); and K 3 = slope of the
/1 curve after the maximum displacement has been reached. The
/ : ~v hysteretic curve, used for cyclic loads, is shown in Fig. 4(b).
V/.I The curve, developed by Dolan (1991), is broken into four

~u
.f_\.e ...: sections, and the sections are governed by the following equa-
~u tions (the subscripts indicate the section number):
Downloaded from ascelibrary.org by Columbia University on 02/02/15. Copyright ASCE. For personal use only; all rights reserved.

FIG. 3. Schematic for Sheathing-Framing Connector: (a) before Flu = -PI + K 4 a + [exp(a11i) - 1],
Loading; (b) after Loading

p (2a,b)

I
I In( - PI - F z + K 4uZ + 1)
I az = IUzl (3a,b)
L.--'-- ~I--:--~--I~I
I~...""I F3" = PI + K 4a - [exp(a 3 Ial) - 1],

Load
In(P I - Fz + K 4 uZ + 1)
Pt1 - (u,.F.) \
a3 = IUzl (4a,b)
Load Envelope - - - "
\
F4 " = PI + K 4 a + [exp(a 41i) -1],
\
section 4 \\

..._::::::::f K• \ (5a,b)
\
\ where Ii connector displacement; PI load at zero de-
Oeftectlon
\
flection; K 4 = slope at zero deflection; U I _2 = maximum and
leCtion 1 minimum deflections; and F I _2 = maximum and minimum
\ " K. :t::::::>······ loads. The load-displacement relationships for both mono-
\
\
tonic and cyclic loads were determined by curve-fitting ex-
\ perimental data, and the data detail the relationship between
--
\
\
Hysteresis the bearing deformation of the main and side members, de-
... formation of the fastener material, and overall strength of
the connection. The equations for the hysteretic curve rep-
FIG. 4. Load-Displacement Curves for Connectors: (a) Mono- resent the load-displacement curve at maximum displace-
tonic; (b) Hysteretic ment. Small amplitude displacements are modeled using load-
displacement curves similar in shape to the virgin hysteresis
curs between concurrent points on the framing and sheathing. curve that blends into the curves of the hysteresis loop. Val-
Given the condition shown in Fig. 3(a), where points P on idation of the connector model involved monitoring a con-
the sheathing and P' on the framing are concurrent, an ap- nector and correlating the resulting load-displacement path
plied loading causes relative displacement between the points. with the theoretical load-displacement path. The result was
This displacement, shown in Fig. 3(b), represents the relative an almost one-to-one correlation between the two load-dis-
displacement of the connector, where liu is the x-component placement paths. Details of the validation are given by White
of the displacement and Ii v is the y-component of the dis- (1995).
placement. The connector displacement is determined by tak- The sheathing-bearing element is a bilinear spring with a
ing the difference between the framing and sheathing dis- high modulus of elasticity in compression and a low modulus
placements at that point. Once the displacement is determined, in tension (see Fig. 5), which is similar to a gap element. The
the corresponding connector load and stiffness are calculated high modulus in compression accounts for the effect of ad-
by using a load-displacement relationship for the connector. jacent sheathing elements bearing on each other, while the
The force-displacement relationship for a connector varies
according to whether a monotonic- or cyclic-load analysis is
performed. The curve for monotonic loading, shown in Fig.

1
6

····
4(a), is governed by the following equations developed by 6Y 6Y
Foschi (1977) and modified by Dolan (1989) for displacements
CD ®
beyond the maximum load:
·. 6X

IF"I = (Po + K2 1lil) [1 - exp ( -Kolal)]


Po
Y

~x
i

"
j

,
:..J lIn J
for lal < lamaxl (la) FIG. 5. Sheathing-Bearing Element

JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1995/1631

J. Struct. Eng. 1995.121:1629-1635.


low modulus in tension allows for free movement of the pacity of the experimental walls and the wall models was
sheathing panels when they are not bearing on one another. calculated and is tabulated in Table 2. The largest percentage
This element functions in the following manner. The differ- difference was 2%. The difference could be because average
ence in the displacements of adjacent nodes on adjacent ele- nail parameters were used or because localized bearing failure
ments is calculated. The effect of bearing is included if the (crushing) occurred around the framing connectors during the
nodes overlap and is neglected if the nodes do not overlap. experiments. However, the results seem to indicate that the
Analysis is performed by merging the stiffness, damping, program adequately models walls subjected to monotonic
and mass matrices using conditions of compatibility (Holzer loading.
1985), and then solving for the displacements using the New- Two 2.4 m by 2.4 m (8 ft by 8 ft) 4-6 walls were modeled
ton-Raphson for monotonic analysis and the Newmark-beta in order to validate the dynamic capabilities of the model.
method for dynamic analysis. The models have geometries and mechanical properties sim-
ilar to those of the wall models used for the monotonic val-
PROGRAM VALIDATION idation tests. The walls were subjected to the acceleration
record of the S69E component of the 1953 Kern County,
Validation of the program consisted of the following: (1) California earthquake. The record, shown in Fig. 7 along with
A comparison of the peak monotonic load from the experi-
Downloaded from ascelibrary.org by Columbia University on 02/02/15. Copyright ASCE. For personal use only; all rights reserved.

the corresponding response spectrum shown in Fig. 8, con-


mental data and the program results to determine how ac-
curately the program predicts the load capacity of a wall; (2)
correlation between program results and experimental data
to determine the accuracy of the load-displacement curve for
monotonic analysis and the time-displacement history for dy-
40

35

30
~ r::::--
---
namic analysis; and (3) a correlation of the fast-Fourier trans-
~~
~
form (FFT) of predicted and experimental time-displacement
histories to determine the accuracy of the prediction of nat-
ural frequencies of the response for the wall.
~25
i
~2O
.. / ;:;i'

Two walls were modeled to validate the monotonic analysis 15 ~


capability of the program. Both walls were 2.4 m by 2.4 m
(8 ft by 8 ft), with 8d nails (diameter = 3 mm, length = 63.5 10 ~V
1/
I'
mm) spaced 10 cm (4 in.) along the perimeter of the sheating 5
and IS cm (6 in.) on the interior of the sheathing. The studs
were placed 61 cm (24 in.) on center. One wall was covered o
with two panels of waferboard sheathing and the other with
o 10 20 30 40 50 60 70 80
Displacement (mm)
two panels of plywood sheathing. The values used for the
sheathing and framing mechanical properties were based on
tabulated design values and the properties for the sheathing- 1- Experimenlal Data A. Program Results
to-framing connectors, found in Table 1, were based on ex-
FIG. 6. Comparison of Monotonic Plywood, Predicted and Ex-
perimental data from Dolan (1989). The sole plate of each perimental Data
wall was fixed, the ends of the studs pinned, and each wall
was subjected to a monotonic distributed load applied at the
TABLE 2. Results of Monotonic Program Verification
top sill plate, which increased at increments of 0.88 kN/m
(5.0 Ib/in.). The results from the program were compared to Nail Spacing Maximum Strength
Correlation
experimental data from Dolan (1989). The walls used in the Experi- Percent- coefficient
experimental study were 2.4 m by 2.4 m (8 ft by 8 ft) with mental Program age for
38 mm by 89 mm (2 in. by 4 in. nominal) spruce-pine-fir stud Sheathing Exterior Interior results results differ- monotonic
framing spaced at 61 cm (24 in.) on center. Four of the walls type (em) (em) (kN) (kN) ence curve
were sheathed with two panels of 9 mm (3/8 in.) waferboard (1 ) (2) (3) (4) (5) (6) (7)
and three of the walls were sheathed with two panels of 9 Waferboard 10 15 31.8 32.4 1.8 0.9965
mm (3/8 in.) three-ply Softwood plywood. The sheathing was Plywood 10 15 33.5 33.1 l.l 0.9938
attached with 8d common nails spaced at 10 cm (4 in.) on
the perimeter of the sheathing and 15 cm (6 in.) on the interior
of the sheathing. The load-displacement relationships from 0.2~--r---r---.---.----'----'
the computer models and the experimental data are plotted
in Fig. 6 for the plywood walls and model. The plot shows a 0.15+--+-:++--++--+--+---+----j
good correlation between the program results and the ex-
0.1
perimental data, and this is supported by the results of a
correlation ofthe load-displacement curves between each wall
model and the average of the corresponding experimental
data. The results of the correlation, shown in Table 2, reveal
a correlation coefficient of 0.9965 for the waferboard wall and
0.9938 for the plywood wall, which is considered to be ex-
cellent. The percentage difference between the ultimate ca-

TABLE 1. Values for Connector Data


Sheathing Ko Po K2 ~max K3 P,
type (kN/m) (N) (kN/m) (mm) (kN/m) (N)
(1 ) (2) (3) (4) (5) (6) (7)
10 20 30 40 50 60
TIme (sec)
Waferboard 831 800 35.4 12.7 500 334
Plywood 972 639 42.4 12.7 500 334 FIG. 7. Acceleration Record of 1953 Kern County, California
Earthquake

1632/ JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1995

J. Struct. Eng. 1995.121:1629-1635.


0.014~---.----"'---"'T""--'----""---' 3.5

3
/
0.012+----1---+--+--+---t----j
-:/
0.01+--++---1---1---1---t-----i
/
V
V
0.004 0.5
/
O.OO2+--1~-+-IH-
V 0.5 1.5 2 2.5 3 3.5
ExperImental Data
Downloaded from ascelibrary.org by Columbia University on 02/02/15. Copyright ASCE. For personal use only; all rights reserved.

00 2 4 6 12
Frequency (Hz)
FIG. 8. Response Spectrum of 1953 Kern County, California
I ... RegIllllSion data - 45 degree line

Earthquake FIG. 10. Comparison of Predicted and Experimental Natural Fre-


quencies of Plywood Wall from Fast-Fourier Transforms (FFT)
20

15
TABLE 4. Results of Force and Overturning Moment Validation
for Walls SUbjected to Static Load
10
Hori-
I 5
A~ zontal Per- Over- Percent-
rv

i ~ ~N
11\1 Applied reaction centage turning Resisting age

Iro ~\
Sheathing load force differ- moment moment differ-
~
~ type (kN) (kN) ence (kN·m) (kN'm) ence
V (1 ) (2) (3) (4) (5) (6) (7)
·10 Waferboard 19.2 18.5 3.3 46.8 47.1 0.6
Plywood 19.2 19.1 0.4 46.8 47.6 1.6
-15

-20
o 2 3 456 7 8 9 10
Tlme (sec) the framing connectors. This was, however, considered to be
a relatively good correlation.
1- ExperIrnentaI data - Program resulls
A FFT of the load-displacement curve was taken for the
plywood and waferboard displacement traces. The peaks of
the FFT curves from the experimental data and the program
FIG. 9. Comparison of Dynamic Plywood, Predicted and Experi-
mental Data (Initial 10 s) results were determined and compared to determine how closely
the program predicts natural frequencies. The correlation be-
tween the program results and the experimental results, shown
TABLE 3. Results of Dynamic Program Verification
in Table 3, was 0.9990 for waferboard and 0.9998 for plywood.
Nail Spacing Correlation This indicates that the program does an excellent job of pre-
coefficient for Correlation dicting the natural frequencies in the response. A plot show-
Sheathing Exterior Interior displacement coefficient for ing the relationship between the natural frequencies from a
type (em) (em) trace FFT FFT for the experimental data and the program results is
(1) (2) (3) (4) (5) shown in Fig. 10 for walls with plywood sheathing. All the
Waferboard 10 15 0.8345 0.9990 points would have been exactly on the 45° line had the cor-
Plywood 10 15 0.8456 0.9998 relation been perfect.
Validation of the force- and stress-calculating capabilities
of the program is impossible because of the lack of experi-
tained acceleration points for every 0.01 s and was 54.34 s mental data on forces and stresses in a shear wall. Therefore,
long. The output from the simulation runs for both walls was validation of the force and stress calculations consists only of
also compared with experimental data from Dolan (1989). a force balance for a wall model subjected to a static load of
The walls used in the seismic analysis segment of the exper- 19.2 kN (4,320 Ib), the results of which are shown in Table
imental study were similar in material properties and ge- 4, and traces of the axial force in a framing element and the
ometries to the walls used in the monotonic segment. A rep- shear stress in a sheathing element as a function of time,
resentative lO-s sample of the results of the computer models shown in Figs. 11 and 12, of a wall model subjected to the
and the maximum and minimum of the experimental data are Kern County earthquake. The results of the force balance
plotted in (Fig. 9) plywood walls and model. The figure shows show that the shears and moments at the base of the wall are
a close correlation between the experimental and program in equilibrium with the applied loads, while the force and
time-displacement time histories. The results of a correlation stress traces demonstrate the ability of the program to cal-
in the time-displacement histories between each wall model culate loads and stresses as a function of time.
and the average of the corresponding experimental data are The excellent predictions indicate that the reduction of the
shown in Table 3. Both walls had a correlation coefficient of degrees of freedom made to simplify Dolan's (1989) model
approximately 0.84. Derivations from a perfect correlation were acceptable. The sheathing thickness modeled was 9 mm
could be due to the fact that average nail parameters were (3/8 in.), which is the minimum thickness that Dolan's model
used or localized bearing failure (crushing) occurred around indicated the buckling of the sheathing should be ignored.
JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1995/1633

J. Struct. Eng. 1995.121:1629-1635.


2tI CONCLUSIONS
20 A finite-element program, WALSEIZ, which is capable of
111
performing monotonic or dynamic analysis on timber shear
walls up to 4.9 m x 12 m (16 ft x 40 ft), was developed.
10 The program uses four elements to model a shear wall: a
framing element, sheathing element, sheathing-to-framing
~
A
& connector element, and a sheathing-bearing element. The
~ II \
I 0
IV IIy II IV
program is capable of calculating displacement, velocity, and
acceleration of the nodes, forces, and stresses in the elements
III as a function of time. Validation of the program consisted of
·10
analyzing wall models subjected to monotonic and dynamic
loading and comparing the results with existing experimental
·111 data. The program predicted the maximum strength of a wall
subjected to monotonic loading to within 2%, and correlated
·20
Downloaded from ascelibrary.org by Columbia University on 02/02/15. Copyright ASCE. For personal use only; all rights reserved.

o 2 3 468 7 8 10 well with dynamic test results, with correlation coefficients


nme(aec) of 0.835 and 0.846 for displacements.
FIG. 11. Axial Force History of End-Chord Element In Plywood
Shear Wall (Initial 10 s) ACKNOWLEDGMENT
The writer wishes to acknowledge the support of the National Research
400 Initiative USDA/CSRS competitive grants program received through
Project No. 90-38420-5252.
300

200 APPENDIX I. REFERENCES


A AI Cook, R. D., Malkus, D. S., and Plesha, M. E. (1989). Concepts and
I~
II
f\ IA
applications of finite element analysis, 3rd Ed., John Wiley & Sons,
Inc., New York, N.Y.
V \) Davis, H. E., Troxell, G. E., and Hauck, G. F. W. (1982). The testing
of engineering materials, 4th Ed., McGraw-Hill Book Co., Inc., New
v York, N.Y.
Dolan, J. D. (1989). "The dynamic response of timber shear walls,"
PhD thesis, Univ. of British Columbia, Vancouver, B.C., Canada.
Dolan, J. D. (1991). "A numerical model to predict the dynamic response
of timber shear walls." Proc., Int. Timber Engrg. Conf., Vol. 4,267-
274.
Falk, R. H., and Itani, R. Y. (1989). "Finite element modeling of wood
-500
o 2 3 468 7 8 10 diaphragms." J. Struct. Engrg., ASCE, 115(3), 543-559.
nme(aec) Filiatrault, A. (1990). "Static and dynamic analysis of timber shear walls."
Can. 1. Civ. Engrg., Ottawa, Canada, 17(4), 643-651.
FIG. 12. Shear-Stress History of Sheathing Element In Plywood Foschi, R. O. (1977). "Analysis of wood diaphragms and trusses. I:
Shear Wall (Initial 10 s) Diaphragms." Can. J. Civ. Engrg., 4(3), 345-352.
Foschi, R. 0., and Filiatrault, A. (1990). "Performance evaluation of
3M Scotch Grip Wood Adhesive 5230 for the static and dynamic design
Therefore, the limitation imposed on this model is that the of timber shear walls and diaphragms." Rep., Dept. of Civ. Engrg.,
sheathing modeled must have a bending stiffness sufficient to Univ. of British Columbia, Vancouver, B.C., Canada.
Gray, R. G., and Zacher, E. G. (1988). "Dynamic testing of wood shear
prevent buckling. If a thinner or more flexible sheathing is panels." Arch., 77(3),121-124.
to be used, Dolan's more detailed model must be used in the Hanson, D. (1990). "Shear wall and diaphragm cyclic load testing; cyclic
analysis. shear fastener testing; and panel durability performance testing of
Weyerhaeuser sturdi-wood oriented strand board." Rep.. Prepared for
Ofc. of State Archit. of California, Struct. Safety Sect.
FUTURE USES Holzer, S. (1985). Computer analysis of structures, Elsevier Publishing
Co., New York, N.Y.
The program is currently being used as a tool for further Kasal, B., and Leichti, R. J. (1992). "Nonlinear finite-element model
for light-frame stud walls." J. Struct. Engrg., ASCE, 118(11).3122-
investigations into the response characteristics of shear walls.
3135.
Openings are necessary in shear walls to accommodate doors McCutcheon, W. J. (1985). "Racking deformations in wood shear walls."
and windows, and have been shown to reduce the strength J. Struct. Engrg., ASCE, 111(2), 257-269.
and stiffness of a shear wall loaded monotonically (Pat- Patton-Mallory, M., Wolfe, R. W., Soltis, L. A., and Gutkowski, R.
ton-Mallory et al. 1985). However, no research has been per- M. (1985). "Light-frame shear wall length and opening effects." 1.
Struct. Engrg., ASCE, 111(10),2227-2239.
formed that examines the effect openings have on a shear Stewart, W. G. (1987). "The seismic design of plywood sheathed shear
wall subjected to dynamic loading. The effect of the aspect walls," PhD thesis, Univ. of Canterbury, Christchurch, New Zealand.
ratio is another area that has not been adequately examined. Tuomi, R. L., and McCutcheon, W. J. (1978). "Racking strength of
Tests have been performed to characterize the effect that light-frame nailed walls." J. Struet. Div., ASCE, 104(7), 1131-1140.
White, M. W. (1995). "Parametric study of timber shear walls,"
aspect ratio has on the response of a shear wall subjected to PhD dissertation, Virginia Polytechnic Inst. and State Univ.,
monotonic loading. However, research in the area of dynamic Blacksburg, Va.
analysis is lacking. WALSEIZ is currently being used to ex-
amine these phenomena so that shear-wall design can be per- APPENDIX 11. NOTATION
formed more accurately. In addition, the information pro-
vided by WALSEIZ can be used to develop performance The following symbols are used in this paper:
characteristics for a macroelement, which can be used to model
an entire structure. al~4 = parameter used to define hysteresis loop;
1634/ JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1995

J. Struct. Eng. 1995.121:1629-1635.


F" force experienced by connector subjected to mono- slope of monotonic connector curve after maximum
tonic loading; displacement is reached;
maximum and minimum force of the hysteresis slope of hysteresis curve at zero displacement;
loop; Y-intercept of the K 2 line on monotonic connector
force experienced by connector subjected to dynamic curve;
loading; Y-intercept of hysteresis loop;
K" slope of monotonic connector curve at zero displace- maximum and minimum displacement of hysteresis loop;
ment; connector displacement; and
slope of monotonic connector curve at maximum dis- maximum connector displacement for connector sub-
placement; jected to monotonic loading.
Downloaded from ascelibrary.org by Columbia University on 02/02/15. Copyright ASCE. For personal use only; all rights reserved.

JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1995/1635

J. Struct. Eng. 1995.121:1629-1635.

You might also like