See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/245408564

Beam–spring model for timber diaphragm and shear walls

Article  in  ICE Proceedings Structures and Buildings · January 2010

DOI: 10.1680/stbu.2010.163.4.227

CITATIONS READS

13 687

2 authors, including:

Weichiang Pang
Clemson University
116 PUBLICATIONS   1,257 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

NEES Soft View project

All content following this page was uploaded by Weichiang Pang on 20 September 2019.

The user has requested enhancement of the downloaded file.

 Proceedings of the Institution of
Civil Engineers
Structures and Buildings 163
August 2010 Issue SB4
Pages 227–244
doi: 10.1680/stbu.2010.163.4.227 Wei Chiang Pang David V. Rosowsky
Assistant Professor, Professor and Dean of
900078 Department of Civil Engineering, Rensselaer
Received 16/09/2009 Engineering, Clemson Polytechnic Institute, Troy,
Accepted 24/05/2010 University, Clemson, SC, NY, USA
USA
Keywords: diaphragm walls/seismic
engineering/timber structures

Beam–spring model for timber diaphragm and shear walls

W. C. Pang PhD and D. V. Rosowsky PhD

A roof or floor diaphragm in a wood building generally r1 , r2 , r3 , r4 modelling parameters for modified Stewart
spans across multiple lateral force-resisting elements hysteretic model
(shear walls). The diaphragm serves as a horizontal u1 , v1 , u2 , v2 nodal displacements of beam element
beam that distributes forces to the shear walls. For xi , xj ith or jth spring (shear wall) location of the
design purposes, it is common to assume the wood simplified beam–spring model
diaphragm is either completely flexible or completely Æ,
stiffness and strength degradation parameters
rigid. In North American low-rise light-frame wood ˜ shear wall deflection
construction, wood diaphragms are almost always ˜u deflection at maximum shear wall backbone
designed as flexible diaphragms and are modelled as force
simple beams spanning across two adjacent shear walls. [ª], [º] stiffness coefficient matrices for beam–spring
Strictly speaking, neither the flexible nor the rigid model
assumption is accurate for modelling the behaviour of a
wood diaphragm. Full-scale shake table tests of a two- 1. INTRODUCTION
storey wood-frame structure, conducted as part of the In North America, light-frame timber construction is the
NEESWood project, have confirmed that roof and floor dominant building method used in both low-rise single-family
diaphragms are semi-rigid. This paper examines the homes and mid-rise multi-storey residential structures. Light-
effect of diaphragm flexibility on shear wall deflections frame timber construction offers advantages over other
by considering the in-plane stiffness of the diaphragm to building methods (e.g. reinforced concrete and steel moment
be semi-rigid. A beam–spring analogue model is used to frames) in the form of lower construction costs and faster
represent the diaphragm–shear wall system where the overall construction time. However, light-frame timber
shear walls are modelled as springs and the diaphragm is buildings are extremely difficult to accurately model or analyse
modelled as an analogue beam that acts as a load due to the complex interactions between the interconnected
distribution mechanism. The resulting load sharing sub-assemblies and framing members.
among the shear walls is examined and possible
application of the beam–spring model to seismic design A light-frame timber building typically consists of dimension
of wood-frame structures is discussed. timber as framing members (wall studs, floor joists and roof
trusses) to which various sheathing materials such as plywood
NOTATION or oriented strand board (OSB) are attached. The framing
A cross-sectional area members and sheathing when oriented horizontally form the
a acceleration floor and roof diaphragms; when oriented vertically, they form
E modulus of elasticity the partition walls and shear walls. A typical diaphragm in a
Fi force intercept of pinched line (r4 K0 ˜ þ F1 ) timber building has two primary functions. The first is to carry
Fb shear wall backbone force gravity loads and transfer them to the load-bearing walls.
F0 force intercept of asymptotic line (r1 K0 ˜ + F0 ) Additionally, a diaphragm acts as a horizontal beam that
in modified Stewart hysteretic model collects lateral forces due to earthquakes or high wind events
Fu maximum shear wall backbone force and transfers them to the shear walls.
I moment of inertia
[I] identity matrix In a timber building, diaphragms and shear walls serve as the
[Kbeam ] stiffness matrix of beam element primary lateral-force-resisting systems. A diaphragm in a
Kd equivalent linear stiffness of diaphragm timber building generally spans across multiple shear walls.
backbone curve The distribution of lateral forces or load sharing among the
K0 initial shear wall stiffness shear walls depends on the flexibility of the diaphragm. For
Kp reloading stiffness design purposes, it is common to assume the timber diaphragm
k1 , k i , . . ., k n spring stiffness is either completely flexible or completely rigid. In low-rise
L length light-frame timber construction, timber diaphragms are almost
mi lumped mass at ith spring always designed as flexible diaphragms, and modelled as

Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky 227

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 simple beams spanning two adjacent shear walls. Strictly
speaking, neither the flexible nor the rigid assumption is
accurate for modelling the behaviour of a timber diaphragm. A
series of full-scale shake table tests, conducted as part of the
NEESWood project, has confirmed that roof and floor
diaphragms are indeed semi-rigid (Christovasilis et al., 2007).
In this paper, the effect of diaphragm flexibility on shear wall
deflections is examined by analysing the NEESWood test
building using two beam–spring analogue models (a finite-
element (FE) based model and a simplified beam–spring
model). In both cases, the shear walls are modelled as non-
linear springs and the semi-rigid behaviour of a diaphragm is
modelled using an analogue beam that acts as a load
distribution mechanism.
2. CLASSIFICATION OF DIAPHRAGM FLEXIBILITY
In the USA, design codes (ASCE, 2005; ICC, 2009) require that
diaphragms be classified as either flexible or rigid for the
purpose of distributing the design seismic forces to the shear
walls. A diaphragm that is designated as flexible is assumed to
distribute seismic forces based on tributary width or area, while
a rigid diaphragm is assumed to distribute seismic forces to the
shear walls in proportion to the walls’ stiffness.
The provisions for determining diaphragm flexibility can be
found in section 1613.6.1 of the current edition of the
International Building Code (IBC) (ICC, 2009) and in section
12.3.1.1 of ASCE/SEI 7-05: Minimum design loads for
buildings and other structures (ASCE, 2005). According to the
IBC, timber diaphragms are permitted to be idealised as flexible
if the following conditions are met.
(a) Constructed of timber structural panels with no more than
38 mm thickness of non-structural topping.
(b) Each line of vertical elements (shear walls) complies with
the allowable storey drift of Table 12.12-1 in the IBC (ICC,
2009).
(c) Shear walls are sheathed with timber structural panels
rated for shear resistance.
(d ) Cantilevered portions of timber diaphragms are designed in
accordance with section 4.2.5.2 of the American Forest &
Paper Association (AFPA) special design provisions (AFPA,
2008).
3. DESCRIPTION OF TEST STRUCTURE
As part of the NEESWood project, a two-storey light-frame
timber building (also known as the NEESWood benchmark
structure (Figure 1)) was built and a series of shake table tests
were performed to investigate the seismic performance of the
test structure at full scale. This benchmark structure is
representative of a typical townhouse building constructed in
the 1980s and located in southern California. The test structure
had approximately 170 m2 of living space and an attached
two-car garage. The storey height of the structure was 2.74 m.
Figure 2 shows the floor plans of the test building and the
layout of shear walls. All shear walls were constructed using
nominal 51 mm 3 102 mm Hem Fir studs apart from the shear
walls located on the west side of the first floor (garage walls
E1, E2, E16, E30 and E31), where 51 mm 3 152 mm studs were
used. In order to study the influence of wall finish materials on
the seismic response of the test building, multiple seismic tests
were conducted at various stages of construction (Table 1). In
this paper, only phase 1 and phase 3 test structures are
modelled.
4. FE BEAM–SPRING MODEL FOR TEST STRUCTURE
A numerical model for the test structure was constructed using
a specialised non-linear dynamic time–history analysis
program developed for light-frame timber structures, called
M-Saws (Matlab Seismic Analysis of Woodframe Structures). In
the M-Saws model, shear walls are modelled as non-linear
single-degree-of-freedom (SDOF) springs using the modified
Stewart hysteretic model (MSHM) (Folz and Filiatrault, 2001a)
and diaphragms are modelled using linear two-node beam
elements (Figure 3). The MSHM is essentially a non-linear
SDOF spring which includes hysteretic pinching, strength and
stiffness degradation. The modelling parameters for the MSHM
are shown graphically in Figure 4 and the backbone equation
for the hysteretic model is
Fb (˜) ¼
8

< 1
exp
K 0 ˜ ð r1 K 0 ˜ þ F0 Þ for
>
˜ < ˜u
1 F0
>
:
Fu þ r2 K 0 (˜
˜u ) for ˜ . ˜u

N
The above criteria are applicable to most regular one- and two-
storey light-frame timber buildings, so flexible diagrams can be
assumed for most single-family dwellings. However, a two-
storey single-family house tested as part of the CUREE
(Consortium of Universities for Research in Earthquake
Engineering) Caltech timber-frame project (Fischer et al., 2001)
clearly exhibited rigid diaphragm behaviour. On the other 2·74 m
2
hand, shake table results obtained from the aforementioned
NEESWood test programme seem to suggest that the roof and 2·74 m
4
floor diaphragms are semi-rigid (Christovasilis et al., 2007).
D 5
Since the size of the CUREE test structure is significantly
smaller than the average size of single-family homes in the
USA, it was not used in the numerical study presented in this A 6

paper. Rather, the larger two-storey NEESWood structure

(described in Section 3) was selected as the benchmark for
Figure 1. Two-storey NEESWood benchmark structure
evaluating the effect of diaphragm flexibility on shear wall (Figures 2 and 3 refer)
deflections.

228 Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 (a)
2·95 m 2·93 m 3·35 m 3·43 m 2·44 m 0·99 m

E1 ⬍150⬎ E2 E35 ⬍150⬎ E4 ⬍150⬎ E5

D
H Stairway
Shear wall
H Partition wall Kitchen Nook
E6

UP
I27
Hold-down I1 ⬍100⬎ ⬍150⬎
⬍150⬎ N
E16 H
⬍100⬎
6·86 m

Garage
B H
E12 H
Shear wall ID ⬍100⬎
H

3·43 m
E13 E11 Dining area Living room E8
⬍150⬎ Atrium ⬍100⬎ ⬍75⬎
H Perimeter nail
y H H
spacing: mm
H H H
A
x E36 ⬍150⬎ E37 E10 ⬍150⬎ E9

6 5 4 2
5·87 m 3·35 m 6·86 m

(b) 0·99 m

E17 ⬍150⬎ E18 ⬍150⬎ E19 ⬍150⬎ E20

D
Stairway
E27
⬍150⬎
Bedroom #2 Laundry Closet Master bath E21
⬍150⬎
DN

E22
6·86 m

2·29 m
B
E28
E34 E33
⬍150⬎
⬍150⬎

4·65 m
Master bedroom
3·43 m

Bedroom #1 Open to below E23

2·52 m

E32 ⬍100⬎
y ⬍150⬎
E29 E26
⬍150⬎ ⬍150⬎
A
x E30 ⬍150⬎ E31 E25 ⬍150⬎ E24

6 5 4 2
5·87 m 3·35 m

Figure 2. Shear walls layout of the benchmark structure: (a) first-storey plan view; (b) second-storey plan view

Phase 1 Timber structural elements only

Phase 2 Phase 1 structure with passive viscous fluid dampers installed in selected shear walls

Phase 3 Phase 1 structure with 12.7 mm thick gypsum wall boards (GWBs) installed with No. 6 31.8 mm long dry wall
screws on load-bearing walls (also serve as shear walls)

Phase 4 Phase 1 structure with 12.7 mm thick GWBs installed with No. 6 31.8 mm long dry wall screws on all walls
and ceilings

Phase 5 Phase 4 structure with 22.2 mm thick stucco installed on exterior surface of the structure

Table 1. Summary of test phases

Further details on the MSHM are described by Folz and master nodes were placed on the floor and roof diaphragms of
Filiatrault (2001a). the test structure along the transverse wall lines 2, 4, 5 and 6
and these wall lines were connected together through beam
In order to consider the in-plane deformation of diaphragms, elements (Figure 3). Each master node has one rotational and

Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky 229

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 Beam element (semi-rigid diaphragm)
Master node
6
5 B4 B5 7 8
B6
Rigid link

Slave node 1 2
3 4

B1 B2 B3

Non-linear
SDOF spring

Y D

N
A
X
6 5 4 2

Figure 3. Depiction of numerical model for test structure. For clarity, shear walls (springs) parallel to the longitudinal direction
(x-axis) are not shown

(∆, Fu) sheathing-to-frame nail spacing of each shear wall. All shear
Force Unloading displacement
è walls had the same interior nail spacing (305 mm). Also shown
æ F0/K0
α
r1K0 r2K0 of previous loop
Kp ⫽ K0 ç
ç
æ ∆un in Figure 2 are the locations of hold-down devices. These hold-
è β∆un F0 down devices were installed in selected first-storey shear walls
Kp
K0 only; as a result, significant wall uplift (separation between end
r4K0
studs and bottom plate) was recorded at some shear wall
Displacement
locations during the seismic tests. This indicates that the shear
r4K0
walls in the test building were not fully anchored to the
r3K0 Fi
foundation.

Past research (Dean and Shenton, 2005; Lebeda et al., 2005)

∆
Figure 4. Modified Stewart hysteretic model
two in-plane translational degrees-of-freedom. The beam
elements permit relative movements between the master nodes
in the transverse direction (N–S) and therefore can be used to
model the semi-rigid behaviour of diaphragms.
5. SEISMIC WEIGHTS
Since the amount of finish materials applied to the test
structure varied at each test phase, supplemental weights were
installed in the test structure to maintain similar total weight
throughout the test phases. Table 2 shows the effective seismic
weights at several key locations of the test structure determined
experimentally through white noise tests (Christovasilis et al.,
2007). To model the weight distribution, lumped seismic
weights were assigned to slave nodes at the key locations listed
in Table 2. These slave nodes were connected to the master
nodes through rigid axial elements.
6. SHEAR WALL MODEL (NON-LINEAR SPRING)
The shear walls in the test building were sheathed with 11 mm
thick OSB connected to the framing members using 8d
common nails (63.5 mm long, 3.3 mm diameter). The locations
of shear walls are shown in Figure 2 along with the perimeter
has shown that hold-down devices have a significant effect on
the lateral stiffness and strength of shear walls. In a fully
anchored shear wall, hold-down devices installed in the end
posts create complete load paths to transfer the uplift forces
into the foundation. These continuous vertical load paths force
the wall to undergo a pure shear or racking deformation. In the
case where hold-down devices are not installed in the shear
wall, uplift forces are carried by the nails connecting the
sheathing to the bottom plate. Since these sheathing nails
generally have lower shear-slip capacity than that of the hold-
down devices, unzipping of the nails at the bottom plate and
rocking deformation will occur as the wall drifts horizontally
(Figure 4).
In order to model the rocking response of the partially
anchored shear walls, a specialised shear wall analysis
program, called M-Cashew2 (Matlab Cyclic Analysis of wood
SHEar Walls version 2), was developed (Pang and Hassanzadeh,
2010). The M-Cashew2 program can be used to predict the
non-linear force–displacement response at the top of the wall
by explicitly modelling the relative movements of the wall
components (nails, sheathing panels and framing members).
M-Cashew2 was developed from its predecessor, a Fortran
version of the Cashew program (Folz and Filiatrault, 2001a). In
the original Cashew program, the framing members are
modelled as pin-ended rigid elements that do not provide any
lateral stiffness and the in-plane lateral resistance of the shear

230 Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 Location Weight: kN Connectivity
(master node)
Test phase 1 Test phase 3

First floor diaphragm

Line A&6 29.4 23.1 1

Line A&5 29.7 23.6 2
Line A&4 2.5 4.0 3
Line A&2 35.9 28.0 4
Line D&6 22.8 24.0 1
Line D&5 40.7 41.8 2
Line D&4 36.5 37.8 3
Line D&2 38.9 36.1 4
Total 236.3 218.4

Roof diaphragm

Line A&6 3.8 4.5 5

Line A&5 4.3 5.1 6
Line A&4 4.6 5.5 7
Line A&2 4.1 4.8 8
Line D&6 3.6 4.3 5
Line D&5 3.2 3.8 6
Line D&4 3.7 4.4 7
Line D&2 5.3 6.4 8
Centre of west roof 20.5 20.5 0.5–5, 0.5–6
Centre of structure 30.8 30.8 0.5–7, 0.5–8
Centre of east roof 20.5 20.5 0.5–6, 0.5–7
Total 104.6 110.8

Table 2. Effective seismic weights for test phases 1 and 3

wall is solely provided by the sheathing-to-framing
connections. These modelling assumptions limit the use of the
original Cashew program to modelling fully anchored shear
walls only (i.e. shear walls with hold-down devices and no end
stud uplift). Unlike its predecessor, M-Cashew2 can be used to
create detailed shear wall models that explicitly account for
(a) bending and axial elongation of framing members
(b) shear deformation of the sheathing panels
(c) the non-linear force–slip response of sheathing-to-frame
connections (i.e. 8d common nails in this study)
(d ) the anchorage effect of hold-down devices
(e) the withdrawal of end-nail connections
( f ) the effect of gravity loading
(g) the separation/contact effect between the framing
members.
The M-Cashew2 program was used to predict the cyclic
response of each shear wall in the test building. Figure 5 shows
an example deformed shape obtained from M-Cashew2 for wall
E11 located in the first storey parallel to wall line 4. In order to
obtain the non-linear force–displacement response of wall E11,
a reversed cyclic simulation was performed using a
displacement controlled cyclic loading protocol applied at the
top of the wall. Note that a uniform gravity load was applied
on the double-top plate of the shear wall. Assuming that the
total gravity load (,340 kN) is carried by the load-bearing
walls in the first storey, the average uniform gravity load on
the top of the first storey shear walls is determined to be
6.38 kN/m (computed as the total seismic weight divided by the
total length of the first-storey-bearing walls, 53.3 m). Using the
same approach, the uniform gravity load for shear walls on the
second floors is 1.97 kN/m (105 kN/53.3 m). The top of wall
force–displacement response for wall E11 predicted by M-
Cashew2 is shown in Figure 6(a) and the associated SDOF
spring fitted to the ten-parameter MSHM is provided in Figure
6(b). This modelling process was repeated for each shear wall
in the test building and the equivalent SDOF parameters are
listed in Table 3. The connectivity of each shear wall to the
master nodes at the bottom and top of the wall (Nbot , Ntop ) is
also provided in Table 3. It should be noted that the SDOF
parameters for the shear walls in the test building were not
calibrated from wall test data. However, the ability of the M-
Cashew2 program to predict the timber shear wall response has
been verified in a separate study (Pang and Hassanzadeh,
2010).
The last row of Table 3 shows the unit-width parameters (per
metre of full-height wall panel) for walls sheathed with only
12.7 mm thick gypsum wall board (GWB) (i.e. no OSB)
connected by 31.75 mm long bugle head drywall screws at
406 mm on-centre. In phase 3, a layer of GWB was attached to
the exterior surface of all shear walls except for the interior
shear walls (I1 and I27) where two layers of GWBs were
attached, one to each side of the walls. To construct the phase
3 model, the GWB springs were superimposed on the phase 1
numerical model and the unit-width GWB parameters K0 , F0
and Fi were adjusted for the length of the full-height wall piers
while other parameters (r1 , r2 , r3 , r4 , ˜, Æ and
) were
unchanged. This modelling approach has been used in other
studies (Folz and Filiatrault, 2001b; Pang et al., 2010; van de
Lindt et al., 2010).

Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky 231

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 Gravity load
Force displacement

Contact element (to model

separation between framing
members)

Hold-down element

End-nail connection Sheathing-to-frame

(16d common nail) connection (8d common nail)

Figure 5. Deformed shape of M-Cashew2 shear wall model for wall E11

25 25

20 20

15 15

10 10

5 5
Force: kN

Force: kN

0 0 K0 ⫽ 1·85 kN/mm
r1 ⫽ 0·071
⫺5 ⫺5 r2 ⫽ ⫺0·027
r3 ⫽ 1·010
⫺10 ⫺10 r4 ⫽ 0·034
F0 ⫽ 12·67 kN
⫺15 ⫺15 Fi ⫽ 1·00 kN
∆ ⫽ 88·3 mm
⫺20 ⫺20 α ⫽ 0·91
β ⫽ 1·21
⫺25 ⫺25
⫺150 ⫺100 ⫺50 0 50 100 150 ⫺150 ⫺100 ⫺50 0 50 100 150
Displacement: mm Displacement: mm
(a) (b)

Figure 6. Top of the wall force–displacement response for wall E11: (a) M-Cashew2 prediction; (b) equivalent SDOF hysteretic
model

7. SEMI-RIGID DIAPHRAGM MODEL (BEAM
ELEMENT)
The floor diaphragm was constructed of nominal
51 mm 3 305 mm Douglas Fir joists spaced at 406 mm on-
centre and sheathed with 19 mm thick OSB connected to the
joists using 10d common nails (76.2 mm long, 3.8 mm
diameter). The 10d nails were spaced at 152 mm along the
edges of the panel and at 254 mm in the field of the panel. The
roof system consisted of trusses spaced at 610 mm on-centre.
In this study, only the bottom chords of the roof trusses are
considered as part of the roof ceiling diaphragm.
The in-plane behaviour of the floor and roof ceiling
diaphragms is very similar to that of shear walls. Consider the
floor diaphragm as an example: it is essentially a ‘shear wall’
oriented horizontally. Therefore, M-Cashew2 also was used to
estimate the in-plane force–displacement response of the floor
diaphragm. In the M-Saws model, the floor diaphragm was
divided into three segments and each segment was analysed
separately (Figure 7(a)). The diaphragm model employed in this
study is a one-way flexible diaphragm model. Axial elongation
or compression along the longitudinal direction of the
diaphragm is ignored (Figure 7(b)). The relative movement
between the wall lines parallel to the transverse direction is
modelled using a two-node beam element (Figure 7(c)) and the
stiffness matrix of the beam element is

232 Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 Wall ID K0 : kN/mm r1 r2 r3 r4 F0 : kN Fi : kN ˜: mm Æ
Nbot * Ntop Fu : kN

First storey (parallel to the longitudinal direction, x-axis)

E1 2.57 0.044
0.056 1.010 0.055 18.14 2.00 57.3 0.95 1.20 0 1 24.63
E2 2.59 0.049
0.051 1.010 0.055 16.75 1.63 54.9 0.85 1.20 0 2 23.75
E35 2.88 0.038
0.036 1.010 0.060 23.74 2.58 57.1 0.99 1.19 0 2 29.98
E4 2.55 0.038
0.054 1.010 0.051 22.51 1.88 72.6 0.99 1.21 0 3 29.59
E5 1.81 0.048
0.058 1.010 0.078 15.76 1.54 64.3 0.95 1.15 0 4 21.41
E9 2.66 0.036
0.063 1.010 0.046 28.13 2.29 70.5 0.99 1.18 0 4 34.85
E10 2.42 0.027
0.073 1.010 0.074 29.26 2.38 64.4 0.99 1.53 0 3 33.39
E12 0.78 0.000
0.082 0.819 0.126 21.39 1.78 102.4 0.99 1.15 0 2 20.87
E37 0.83 0.024
0.029 0.739 0.040 4.44 0.09 42.9 0.99 1.23 0 2 5.31
E36 4.04 0.033
0.040 1.010 0.050 37.01 3.34 64.7 0.99 1.19 0 1 45.59

Second storey (parallel to the longitudinal direction, x-axis)

E17 2.73 0.026
0.098 1.010 0.051 29.31 3.50 82.5 0.99 1.37 1 5 35.26
E18 1.83 0.017
0.040 1.010 0.049 19.37 1.20 66.3 0.99 1.26 2 6 21.44
E19 3.68 0.000
0.059 1.010 0.040 42.46 2.13 42.8 0.99 1.25 3 7 41.41
E20 3.71 0.000
0.059 1.010 0.057 42.90 3.93 42.8 0.95 1.15 4 8 41.85
E24 2.22 0.035
0.079 1.010 0.075 23.66 2.45 51.2 0.99 1.20 4 8 27.37
E25 1.00 0.027
0.030 1.010 0.045 13.16 1.42 95.1 0.95 1.15 3 7 15.71
E33 1.40 0.081
0.038 1.010 0.080 11.07 1.33 37.3 0.95 1.15 2 6 15.19
E31 0.61 0.039
0.046 1.010 0.024 3.10 0.27 35.3 1.00 1.29 2 6 3.93
E30 2.84 0.000
0.047 1.010 0.055 34.23 3.19 53.7 0.80 1.20 1 5 33.83
E22 0.60 0.020
0.034 1.010 0.030 2.81 0.35 41.8 0.99 1.24 4 8 3.30

First storey (parallel to the transverse direction, y-axis)

E6 1.07 0.026
0.055 0.455 0.062 15.70 1.71 48.6 0.95 1.15 0 4 16.44
E8 2.60 0.030
0.041 0.834 0.077 45.39 2.07 106.9 0.95 1.25 0 4 53.52
E11 1.85 0.071
0.027 0.699 0.034 12.67 1.00 88.3 0.91 1.21 0 3 24.29
E13 2.18 0.031
0.050 1.400 0.055 23.08 1.93 62.4 0.95 1.20 0 2 27.15
E16 1.36 0.100
0.063 0.861 0.053 15.40 1.76 78.9 0.85 1.25 0 1 26.17
I1 2.26 0.048
0.005 0.320 0.094 22.08 2.39 43.1 0.95 1.20 0 2 26.43
I27 1.96 0.046
0.038 1.010 0.044 19.66 1.00 88.1 0.94 1.15 0 3 27.52

Second storey (parallel to the transverse direction, y-axis)

E21 0.51 0.005
0.042 0.875 0.075 5.42 0.43 39.3 0.95 1.25 4 8 10.76
E23 1.54 0.011
0.020 1.010 0.062 23.48 1.24 39.8 0.95 1.20 4 8 22.39
E26 1.85 0.017
0.051 0.477 0.078 22.23 2.50 37.9 0.95 1.35 3 7 22.39
E32 1.93 0.062
0.058 0.718 0.061 12.31 1.19 34.1 0.89 1.30 2 6 16.33
E29 1.27 0.014
0.022 1.010 0.013 2.96 0.24 38.6 0.95 1.20 1 5 3.62
E28 1.11 0.043
0.005 0.524 0.029 6.83 0.61 70.3 0.98 1.25 1 5 10.19
E27 1.27 0.014
0.022 1.010 0.013 2.96 0.24 38.6 0.95 1.20 1 5 3.62

GWB per metre of full-height wall width

0.332 0.017
0.037 1.055 0.002 1.43 0.074 60.9 0.81 1.01 1.76

*Node 0 means connected to ground

Table 3. Hysteretic parameters for shear walls

2 38 9
AL2 0
AL2 0 >
> u1 >
6 < > = area, A, to restrain the elongation of diaphragm along the
E 0 12I 0
12I 7 v
2 Kbeam ¼ 36 7 1
longitudinal direction. E and I are the modulus of elasticity and
L 4
AL2 0 AL2 0 5 >
> u >
: 2> ; the equivalent moment of inertia. The modulus of elasticity of
0
12I 0 12I v2
Douglas Fir (12 410 MPa) was used as E for the beam elements.
In order to estimate the effective I value, a displacement-
where u and v are the displacements parallel to the controlled monotonic pushover analysis was conducted for
longitudinal and transverse directions, respectively. The length each diaphragm segment to obtain the force–displacement
of each element or diaphragm segment, L, is shown in Figure 2. backbone curve. Figure 8 shows the backbone curve for the
Since axial elongation of the element was not considered, a middle segment of the floor diaphragm obtained using M-
very large value was assigned to the element cross-sectional Cashew2. The effective I can be estimated using

Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky 233

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 (a)

Opening
(stairway)

Force–displacement
Opening

Force–displacement Force–displacement

6 West segment 5 Middle segment 4 East segment 2

Rigid link
(b)

v2
Roller
B3 4
B2 3
v1
u2
B1 2 N
1
u1

(c)
Y
Two-node beam element

Figure 7. (a) Deformed shapes of floor diaphragm segments: (b) semi-rigid diaphragm model; (c) two-node beam element

four tests were performed in phase 3 (Figure 9). The

K d L3
3 I¼
E
where Kd is the stiffness (slope) of the backbone curve. The
results of shake table tests showed that the peak inter-
diaphragm displacement under ground motion level 2
(,10 mm) was within the linear range of the backbone curve.
Therefore, the effective I can be estimated using the initial
stiffness of the backbone curve. From the middle segment of
the floor diaphragm, the effective I was 22 476 cm4 (0.0742
kN/mm 3 (3.35 m)3 4 12 410 MPa). This modelling process
was repeated for every diaphragm segment and the effective I
values are shown in Table 4.
8. DYNAMIC TIME–HISTORY ANALYSES
The 1994 Northridge, Canoga Park earthquake motions scaled
to peak ground accelerations (pga) equal to 0.05g (level 1),
0.22g (level 2) and 0.36g (level 3) were used as the input
table motions for the phase 1 and 3 tests (Christovasilis et
al., 2007). Eight seismic tests were conducted in phase 1 and
designations ‘3D’ and ‘2D’ mean triaxial and biaxial ground
motions, respectively. Similarly, ‘1X’ and ‘1Y’ mean the
applied ground motions were uniaxial ground motions
parallel to the longitudinal and transverse directions,
respectively. The horizontal components of the ground
accelerations are plotted in Figure 9; the vertical components
are not shown.
It should be noted that, in dynamic time–history analyses, only
the horizontal ground accelerations were considered. The
ground motions for phase 1 and 3 tests were combined into
two sets of continuous biaxial ground motions and were used
as the inputs motions in time–history analyses. This modelling
approach allows for consideration of cumulative damage that
occurred in the test structure. The Rayleigh damping ratios
assigned for the test structure were 0.01 (1% of critical
damping) and were associated to the first and third vibration
modes. Similar damping ratios have been used in other
numerical studies (Folz and Filiatrault, 2001b, Pang et al.,
2010; van de Lindt et al., 2010).

234 Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 35

30

25 Kd ⫽ 0·0742 kN/mm

20
Force: kN

Force–displacement

15

Kitchen Nook

UP
c
10 Garage D1
D3
D4 Dining Living D2
5
W1
Atrium
W2 W3

0
0 10 20 30 40 50 60 70 80 90 100
Displacement: mm

Figure 8. Backbone curve for middle segment of floor diaphragm

Element ID Location E: MPa I: cm4 L: m 0·6

acceleration: g

0·4
B1 Floor west 12 410 265 556 5.87
X-ground

0·2
B2 Floor middle 12 410 22 476 3.35
B3 Floor east 12 410 261 810 6.86 0
B4 Roof west 12 410 224 765 5.87 ⫺0·2
B5 Roof middle 12 410 16 649 3.35
⫺0·4 Level 1 Level 1 Level 1 Level 1 Level 2 Level 2 Level 2 Level 3
B6 Roof east 12 410 191 466 6.86 0·4 3D 2D 1X 1Y 3D 1X 2D 1X
acceleration: g

0·2
Y-ground

Table 4. Summary of the two-node (diaphragm) beam element

properties 0

⫺0·2

⫺0·4
9. COMPARISON BETWEEN SEMI-RIGID AND RIGID 0 50 100 150 200 250 300
Time: s
DIAPHRAGM MODELS (a)
The M-Saws program was used to obtain all numerical 0·4
predictions presented in this paper. In addition to the
acceleration: g

0·2
X-ground

previously discussed beam–spring model (for a semi-rigid

diaphragm), a rigid diaphragm model also was created for each 0

test phase. Table 5 compares the initial natural periods ⫺0·2

estimated through white noise tests (Christovasilis et al., 2007) ⫺0·4 Level 1 Level 1 Level 2 Level 2
0·4
and the M-Saws model predicted periods. In general, the 3D 2D 3D 2D
acceleration: g

numerical models overpredicted the initial periods. This may be 0·2

Y-ground

attributed to the fact that the M-Saws model considers only the 0
in-plane stiffness of shear walls while the additional out-of-
⫺0·2
plane wall stiffness and initial stiffness contribution from
friction are ignored. ⫺0·4
0 20 40 60 80 100 120 140 160
Time: s
(b)
Figures 10 and 11 present comparisons between the test and
simulated inter-storey drifts in the transverse direction for the
Figure 9. Ground motions for (a) phase 1 and (b) phase 3
phase 1 structure (structural panels only). For clarity, only the experiments
time-series drift responses of one test (level 2 biaxial test) are

Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky 235

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 Mode Natural period: s

Phase 1 Phase 3

Test Beam–spring Rigid diaphragm Test Beam–spring Rigid diaphragm

1 0.33 0.38 0.36 0.30 0.32 0.30

2 0.23 0.32 0.31 0.22 0.27 0.26
3 0.18 0.26 0.26 0.17 0.23 0.23

Table 5. Comparison of test and model-predicted initial natural periods

shown in Figures 10 and 11. Note that the drift responses in acceleration, a. A similar beam–spring model subjected to
the transverse direction are not presented as the values are very uniform load has been used by others to study the load-sharing
small. In general, both semi-rigid and rigid diaphragm models effect of floor joists (McCutcheon, 1984) and wall systems
were able to predict the first-storey drifts well (Figure 10). As (Bulleit et al., 2005). Following the modelling technique
observed in the test, both models predicted maximum first- employed in the uniformly loaded beam–spring model, the
storey drifts at the garage wall (wall line 6). At the second system of equations for this point-load beam–spring model is
storey, however, the beam–spring model was able to reproduce derived using the consistent deformation approach (Pang,
the test deformed shape while the rigid diaphragm model was 2001)
not (Figure 12). The rigid diaphragm model predicted peak
2 2 3 3
second-storey drift in the exterior wall (line 6) while the actual 0 0 0 0 0
recorded peak second-storey drift occurred in one of the 6 6 7 7
6 6 0 k2 0 0 07 7
interior wall lines (line 4). In the rigid diaphragm model, the 6 6 7 7
6 6 7 7
6 6 7 7
roof ceiling diaphragm deformed and rotated in a rigid body 6 ª i, j þ º i, j 6 0 0 . . . 0 077 þ ½ I 7f˜g
6 6 7
motion. Hence, maximum drift in the rigid diaphragm could 4 6 6 7 7
6 6 7 7
only occur in one of the exterior walls. This means a rigid 6 60 0 0 k n
1 07 7
4 4 5 5
diaphragm model may underpredict the drift demand in
0 0 0 0 0
interior wall lines (Figure 13).
 
¼ ª i, j þ º i, j f m i a i g
The test and model-predicted results for phase 3 (structural
panels + GWB) are presented in Figures 14 to 16. As also seen
in the phase 1 simulation results, the phase 3 rigid diaphragm where subscripts i and j are the matrix row and column
model also displayed linear deformed shapes in both the first numbers, respectively. I is the identity matrix, k is the shear
and second storeys (Figure 16). Compared with the phase 1 wall (spring) stiffness and ˜ is the shear wall deflection. The
results, the discrepancies between the test and the rigid matrices ª and º are given by
diaphragm model were even more apparent in phase 3. The
8
rigid diaphragm model greatly underestimated the drifts in the > x (L
x j )(x j 2
2Lx j
x i 2 )
> i
> xi , x j
second storey (by as much as 50%). This is attributed to the >
> 6EIL
>
>
fact that the rigid diaphragm model overestimated the load >
< 2 2
sharing between the shear walls. Underpredictions of drift ª i, j ¼ x i (L
x i ) xi ¼ x j
5 >
> 3EIL
response in the second storey were also reported in a separate >
>
>
>
study in which a rigid diaphragm assumption was utilised to > x j (L
x i )(x j 2
2Lx i þ x i 2 )
>
: xi . x j
model a two-storey timber building (van de Lindt et al., 2010). 6EIL
In contrast, the beam–spring model predicted drifts for all four
wall lines are in good agreement with the test results.  
1 (L
x j )(L
x i ) x i x j
6 º i, j ¼ þ
10. SIMPLIFIED BEAM–SPRING ANALOGUE MODEL L2 k1 kn
The interaction between shear walls and diaphragms is fairly
complicated. Although FE models such as the M-Saws program where EI is the in-plane bending stiffness of the diaphragm
can be used to analyse the semi-rigid diaphragm effect, these (beam), xi and xj are the shear wall locations measured from
FE models tend to be complex and computationally intensive. the left-hand end of the diaphragm and L is the length of the
For design purposes, the effect of diaphragm flexibility on diaphragm.
shear wall deflections can be approximated using a simplified
beam–spring analogue model. Consider the roof diaphragm 11. ILLUSTRATIVE EXAMPLE
and shear walls system of the test structure as an example To illustrate application of the beam–spring model to seismic
(Figure 17(a)). This roof diaphragm system can be analysed design, the load-sharing effect among the shear walls in the
using a simplified beam–spring model as shown in Figure second storey of the phase 1 test structure is now analysed. In
17(b). the beam–spring model, non-linear shear walls are
approximated using equivalent elastic springs. The elastic
In the simplified beam–spring model, each spring carries a stiffness is determined by connecting a straight line between
point load that is equal to the lumped mass m times the spring the origin and a point on the non-linear backbone curve at

236 Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 Line 6 Line 5 Line 4 Line 2
Structures and Buildings 163 Issue SB4

55
52 (⫹14%)
45
35 (⫺10%)
35 30 (⫺5%) 30 (⫺2%)
Inter-storey drift: mm
25
15
5
(a)
⫺5
⫺15
⫺25 ⫺16 (⫺46%) ⫺16 (⫺46%)

⫺35 ⫺28 (⫺22%)

⫺45
⫺44 (⫹1%)
⫺55

55
45 41 (⫺10%)
34 (⫺11%)
Beam–spring model for timber diaphragm and shear walls

35 32 (⫹0%) 30 (⫺5%)
Inter-storey drift: mm

25
15
5
(b)
⫺5
⫺15
⫺25 ⫺18 (⫺37%)
⫺24 (⫺17%)
⫺35 ⫺28 (⫺22%)
⫺35 (⫺20%)
⫺45
⫺55

55
45 46
39
35 32 31
Inter-storey drift: mm

25
15
5
(c)
⫺5
⫺15
⫺25
⫺29 ⫺29
⫺35
⫺35
Pang • Rosowsky

⫺45 ⫺43
⫺55
255 260 265 270 275 255 260 265 270 275 255 260 265 270 275 255 260 265 270 275
Time: s Time: s Time: s Time: s

Figure 10. Phase 1 structure, first-storey transverse inter-storey drifts: (a) semi-rigid diaphragm model; (b) rigid diaphragm model; (c) test
237

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 238
Structures and Buildings 163 Issue SB4

Line 6 Line 5 Line 4 Line 2

55
45
37 (⫹16%) 39 (⫹4%)
35
28 (⫹16%) 25 (⫺11%)
Inter-storey drift: mm
25
15
5
(a)
⫺5
⫺15
⫺15 (⫺29%)
⫺25 ⫺20 (⫺17%) ⫺22 (⫺25%)
⫺26 (⫺7%)
⫺35
⫺45
⫺55

55
Beam–spring model for timber diaphragm and shear walls

45
37 (⫹54%) 34 (⫹7%) 33 (⫺11%) 32 (⫹13%)
35
Inter-storey drift: mm

25
15
5
(b)
⫺5
⫺15
⫺25
⫺24 (⫺18%) ⫺22 (⫺21%) ⫺23 (⫹6%)
⫺35 ⫺28 (⫹17%)
⫺45
⫺55

55
45
37
35 32
29
24
Inter-storey drift: mm

25
15
5
(c)
⫺5
⫺15
⫺25 ⫺21
⫺24
Pang • Rosowsky

⫺35 ⫺29 ⫺28

⫺45
⫺55
255 260 265 270 275 255 260 265 270 275 255 260 265 270 275 255 260 265 270 275
Time: s Time: s Time: s Time: s

Figure 11. Phase 1 structure, second-storey transverse inter-storey drifts: (a) semi-rigid diaphragm model; (b) rigid diaphragm model; (c) test

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 60

Maximum inter-storey displacement: mm

50
2
40 4
5
6
30
Linear deformed shape
Test
20 Beam–spring
Rigid diaphragm
10 6 5 4 2

0
(a)

50
Maximum inter-storey displacement: mm

Linear deformed shape

40

30

Test
20
Beam–spring
Rigid diaphragm
10 5
6 4 2

0
0 2 4 6 8 10 12 14 16 18
Distance from east end of structure: m
(b)

Figure 12. Comparison of maximum positive (a) first-storey and (b) second-storey inter-storey drifts of phase 1 structure in the
transverse direction

t ⫽ 263·4492 s

Base

Second floor

Roof

(a) (b) (c)

Figure 13. Example deformed shapes: (a) rigid diaphragm model; (b) semi-rigid diaphragm model; (c) phase 1 test

75% of the peak backbone force Fu (see Figure 4). Table 6
summarises the properties of the roof beam–spring model. The
spring stiffness k was determined using the wall parameters
listed in Table 3 and the shear wall backbone equation
(Equation 1). It should be noted that wall E21 is not directly in
line with wall E23 (Figure 17a) but, for simplicity, the stiffness
for spring No. 4 (wall line 3) is computed as the sum of the
elastic stiffnesses of walls E21 and E23.
In seismic design, one determines the spring acceleration a
based on the seismic hazard for the location of the structure.
The spring acceleration can be approximated using the design
spectral acceleration Sa specified in building codes (ASCE,
2005; ICC, 2009) or other seismic design guidelines (e.g. ASCE,
2006). In this example, the roof ceiling acceleration is taken as
1.5g. This magnitude of acceleration is in line with the code-
specified spectral acceleration at the maximum considered
earthquake level for timber structures located in southern
California (Pang et al., 2010). While an assumed spectral
acceleration is used in this example, further study is needed to

Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky 239

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 240
Structures and Buildings 163 Issue SB4
Line 6 Line 5 Line 4 Line 2

50
43 (⫹13%)
40 34 (⫹17%)
30 29 (⫺2%) 29 (⫹10%)
Inter-storey drift: mm
20
10
(a) 0
⫺10
⫺20 ⫺16 (⫺34%) ⫺17 (⫺29%)
⫺30 ⫺26 (⫹4%)
⫺40 ⫺37 (⫹9%)
⫺50

50
Beam–spring model for timber diaphragm and shear walls

49 (⫹27%) 39 (⫹35%)
40 35 (⫹20%)
29 (⫹11%)
30
Inter-storey drift: mm

20
10
(b) 0
⫺10
⫺20
⫺22 (⫺14%)
⫺30 ⫺21 (⫺16%) ⫺21 (⫺14%)
⫺30 (⫺11%)
⫺40
⫺50

50
40 38
29 29
30 27
Inter-storey drift: mm

20
10
(c) 0
⫺10
⫺20
Pang • Rosowsky

⫺30 ⫺25 ⫺25 ⫺24

⫺34
⫺40
⫺50
130 135 140 145 150 155 130 135 140 145 150 155 130 135 140 145 150 155 130 135 140 145 150 155
Time: s Time: s Time: s Time: s

Figure 14. Phase 3 structure, first-storey transverse inter-storey drifts: (a) semi-rigid diaphragm model; (b) rigid diaphragm model; (c) test

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 Line 5 Line 4
Structures and Buildings 163 Issue SB4

Line 6 Line 2
50
40 39 (⫺2%) 35 (⫺13%)
34 (⫹24%)
30 25 (⫺12%)
Inter-storey drift: mm

20
10
(a) 0
⫺10
⫺20 ⫺16 (⫺43%)
⫺30 ⫺23 (⫺30%) ⫺23 (⫹0%)

⫺40 ⫺31 (⫹2%)

⫺50

50
40
Beam–spring model for timber diaphragm and shear walls

30
21 (⫺23%) 20 (⫺49%) 20 (⫺50%) 20 (⫺31%)
Inter-storey drift: mm

20
10
(b) 0
⫺10
⫺20
⫺19 (⫺15%)
⫺30 ⫺24 (⫺26%) ⫺22 (⫺28%)
⫺29 (⫹3%)
⫺40
⫺50

50
39 40
40
28
30 27
Inter-storey drift: mm

20
10
(c) 0
⫺10
⫺20
⫺30 ⫺23
⫺28
⫺40 ⫺33 ⫺31
Pang • Rosowsky

⫺50
130 135 140 145 150 155 130 135 140 145 150 155 130 135 140 145 150 155 130 135 140 145 150 155
Time: s Time: s Time: s Time: s

Figure 15. Phase 3 structure, second-storey transverse inter-storey drifts: (a) semi-rigid diaphragm model; (b) rigid diaphragm model; (c) test
241

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 50
40
Inter-storey displacement: mm
30
20
10
6 2
0
50
40
Inter-storey displacement: mm
30
20
10
6 2
0
0 2
Figure 16. Comparison of maximum positive (a) first-storey and (b) second-storey inter-storey drifts of phase 3 structure in the
transverse direction
determine an appropriate procedure for estimating the design
acceleration for use with the beam–spring model.
Three diaphragm flexibility conditions are investigated: semi-
rigid, rigid and flexible. For the semi-rigid condition, EI is
equal to the bending stiffness of the beam element B4 shown
in Table 4 (2.78031010 kNmm2 ). EI cannot equal zero; thus, for
the flexible diaphragm condition, EI is set to a very small value
(1e
10 3EI). Similarly, for the rigid diaphragm condition, EI of
1000 times the stiffness of the semi-rigid diaphragm is used.
For all three conditions, the length of the beam–spring model
is equal to 16.21 m.
Substituting the properties listed in Table 6 into Equations 4–6
and solving for the deflections ˜ yields the results shown in
Figure 18. The deformed shape of the semi-rigid condition
predicted using the simplified beam–spring model is very
similar to that predicted by the more detailed FE model and the
actual test results (Figure 12(b)). As expected, the rigid
diaphragm model has a linear deformed shape. Compared with
the results of the semi-rigid and flexible models, the rigid
diaphragm analysis yields smaller displacements along the
interior wall lines 5 and 4 and larger displacements along the
exterior walls. This phenomenon was also observed in the
previous FE analysis (Figure 12(b)). The ability of the simplified
242 Structures and Buildings 163 Issue SB4
Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
Linear deformed shape
Test
2 Beam–spring
Rigid diaphragm
4
5
6 5 4
(a)
Linear deformed shape
Test
5 4 Beam–spring
Rigid diaphragm
4 6 8 10 12 14 16 18
Distance from east end of structure: m
(b)
beam–spring model to reproduce the deformed shape of the
actual test data makes it a suitable model for use in a
performance-based seismic design framework (Pang et al.,
2010) where accurate prediction of shear wall displacements is
needed.
12. SUMMARY AND CONCLUSIONS
Two beam–spring models were developed to model the
interaction between diaphragms and shear walls in light-frame
timber structures. In a beam–spring model, shear walls are
modelled as springs and the diaphragm is modelled as an
analogue beam that acts as a load distribution mechanism. The
first beam–spring model presented here was an FE model
which can be used in non-linear dynamic time–history
analyses of the complete light-frame timber structure. The
second simplified beam–spring model was developed
specifically for use in seismic design.
The FE beam–spring model was coded into a specialised time–
history analysis program developed for light-frame timber
structures (M-Saws). Using the M-Saws program, two
numerical models were constructed to predict the actual shear
wall displacements of a two-storey timber-frame structure,
namely the NEESWood benchmark structure tested in a series
of full-scale shake table experiments. The shear walls in the
Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky
 E23 E21
E27
E32 E26
E28

A
x
6 5 4
2

(a)

a1 a2 Point load (mi ⫻ ai )

ai an

m1 m2 mi mn

k1 k2 … ki … kn

xi or xj Linear or non-linear spring (shear wall)

Analogue beam (floor or roof diaphragm), EI

L

(b)

Figure 17. (a) Roof diaphragm and shear wall system of the test structure. (b) Simplified beam–spring analogue model

wall deflections was examined by considering the in-plane

Wall line
stiffness of the diaphragm to be semi-rigid and completely
6 5 4 2 rigid. Good agreement was observed, in terms of the magnitude
of the displacements and deformed shapes, between the
Spring no. 1 2 3 4 numerical predictions of the semi-rigid FE beam–spring model
m: kN/g 17.65 33.15 33.95 19.65 and the actual experimental results. On the other hand, the FE
x: m 0.00 5.94 9.30 16.21 analyses showed that the rigid diaphragm model may
k: kN/mm 1.401 0.908 1.040 1.186 underestimate the magnitude of the displacements observed in
the shake table experiments, especially of the roof diaphragm.
Table 6. Properties of roof beam–spring model for phase 1
structure
Finally, the formulation of a simplified beam–spring model has
been described and possible application of this model to
seismic design presented. Three diaphragm flexibility
60 conditions (semi-rigid, rigid and completely flexible) were
modelled using the proposed simplified beam–spring model.
50
Displacement: mm

The modelling results confirmed that the simplified beam–

40 spring model is capable of estimating the displacements
predicted using the more detailed FE model with reasonable
30
Semi-rigid accuracy. While this simplified beam–spring model has been
20 Rigid (EI ⫽ ¥) shown to be an attractive tool for use in performance-based
Flexible (EI ⬃ 0) seismic design, further study is still needed to determine an
10 appropriate method for estimating the acceleration demand.
0 2 4 6 8 10 12 14 16 18
Spring location: m
REFERENCES
Figure 18. Comparison of displacements predicted using AFPA (American Forest & Paper Association) (2008) Special
simplified beam–spring models for flexible, semi-rigid and
Design Provisions for Wind and Seismic (SDPWS). American
rigid diaphragm conditions
Wood Council, Washington, DC.
ASCE (American Society of Civil Engineers) (2005) ASCE/SEI
test structure were modelled using non-linear SDOF springs. A 7-05: Minimum design loads for buildings and other
separate shear wall analysis program, M-Cashew2, was structures. ASCE, Reston, VA.
developed to predict the non-linear shear wall responses (i.e. ASCE (2006) ASCE/SEI 41-06: Seismic rehabilitation of
SDOF springs). The effect of diaphragm flexibility on shear existing buildings. ASCE, Reston, VA.

Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky 243

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
 Bulleit WM, Pang WC and Rosowsky DV (2005) Modeling
wood walls under combined axial and transverse loads.
ASCE Journal of Structural Engineering 131(5): 781–793.
Christovasilis IP, Filiatrault A and Wanitkorkul A (2007)
Seismic Testing of a Full-scale Two-storey Wood Light-Frame
Building: NEESWood Benchmark Test. University at Buffalo,
Buffalo, NY, NEESWood Report NW-01.
Dean PK and Shenton HW (2005) Experimental investigation of
the effect of vertical load on the capacity of wood shear walls.
ASCE Journal of Structural Engineering 131(7): 1104–1113.
Fischer D, Filiatrault A, Folz B, Uang C-M and Seible F (2001)
Shake Table Tests of a Two-Storey Woodframe House.
Consortium of Universities for Research in Earthquake
Engineering, Richmond, CA, CUREE Report W-06, Task 1.1.1.
Folz B and Filiatrault A (2001a) Cyclic analysis of wood shear
walls. ASCE Journal of Structural Engineering 127(4): 433–
441.
Folz B and Filiatrault A (2001b) A Computer Program for
Seismic Analysis of Woodframe Structures. Consortium of
Universities for Research in Earthquake Engineering,
Richmond, CA, CUREE Report W-21, Task 1.5.1.
ICC (International Code Council) (2009) International Building
Code. Building Officials and Code, International Code
Council Inc., Country Club Hills, IL.
Lebeda D, Gupta R, Rosowsky D and Dolan JD (2005) The effect
of hold-down misplacement on the strength and stiffness of
wood shear walls. Practice Periodical on Structural Design
and Construction 10(2): 79–87.
McCutcheon WJ (1984) Deflections of Uniformly Loaded Floors:
A Beam–spring Analog. Forest Products Laboratory,
Madison, WI, United States Department of Agriculture
Research Paper FPL 449.
Pang WC (2001) Analysis of Light-frame Wood Walls. MS
thesis, Michigan Technological University, Houghton, MI.
Pang WC and Hassanzadeh M (2010) Next generation
numerical model for non-linear in-plane analysis of wood-
frame shear walls. Proceedings of the 11th World
Conference on Timber Engineering, Trentino, June 20–24.
See www.wcte2010.org for further details (accessed 13/07/
2010).
Pang WC, Rosowsky DV, Pei S and Van De Lindt JW (2010)
Simplified direct displacement design of six-storey
woodframe building and pre-test performance assessment.
ASCE Journal of Structural Engineering 136(7): 813–815.
van de Lindt JW, Pei S, Liu H and Filiatrault A (2010) Three-
dimensional seismic response of a full-scale light-frame
wood building: a numerical study. ASCE Journal of
Structural Engineering 136(1): 56–65.

What do you think?

To discuss this paper, please email up to 500 words to the editor at journals@ice.org.uk. Your contribution will be forwarded to the
author(s) for a reply and, if considered appropriate by the editorial panel, will be published as discussion in a future issue of the
journal.
Proceedings journals rely entirely on contributions sent in by civil engineering professionals, academics and students. Papers should be
2000–5000 words long (briefing papers should be 1000–2000 words long), with adequate illustrations and references. You can
submit your paper online via www.icevirtuallibrary.com/content/journals, where you will also find detailed author guidelines.

244 Structures and Buildings 163 Issue SB4 Beam–spring model for timber diaphragm and shear walls Pang • Rosowsky

Downloaded by [ R M COOPER LIBRARY ACQUIS UNIT] on [22/12/15]. Copyright © ICE Publishing, all rights reserved.
View publication stats