You are on page 1of 20

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2009; 38:477–496


Published online 2 December 2008 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.864

Incremental dynamic analysis of woodframe buildings

Ioannis P. Christovasilis1, ‡ , Andre Filiatrault1, ∗, †, § , Michael C. Constantinou1, §


and Assawin Wanitkorkul2, ¶
1 Department of Civil, Structural and Environmental Engineering, State University of New York at Buffalo,
Buffalo, NY 14260, U.S.A.
2 Connell Wagner (Thailand), Bangkok 10110, Thailand

SUMMARY
The collapse of wood buildings was one of the main contributors to the heavy death toll and economic
losses during the 1995 Hyogo-ken Nanbu (Kobe) earthquake in Japan. In California, half of the property
loss from the 1994 Northridge earthquake was attributed to wood construction. Based on damage observed
in recent earthquakes, the seismic vulnerability of existing wood buildings under maximum credible
seismic events is uncertain. The main objective of this study is to quantify the seismic collapse fragilities
and collapse mechanisms of a two-story townhouse and three-story woodframe apartment building through
numerical analyses. Three construction quality variants (poor, typical and superior) were considered for
each building in order to assess the effects of construction qualities on seismic collapse fragilities. The
buildings were also re-designed according to the 2006 edition of the International Building Code to
quantify the seismic fragilities of modern woodframe construction. The results obtained suggest that the
construction quality, excitation direction and wall finish materials can influence significantly the collapse
fragilities of woodframe buildings. Copyright q 2008 John Wiley & Sons, Ltd.

Received 6 November 2007; Revised 8 September 2008; Accepted 8 September 2008

KEY WORDS: collapse; incremental dynamic analysis; seismic; wood structure

1. INTRODUCTION

Observations from significant earthquakes in the last two decades, such at the 1989 Loma Prieta and
the 1994 Northridge earthquakes in California and the 1995 Hyogo-ken Nanbu (Kobe) earthquake

∗ Correspondence to: Andre Filiatrault, Department of Civil, Structural and Environmental Engineering, State University
of New York at Buffalo, Buffalo, NY 14260, U.S.A.

E-mail: af36@buffalo.edu, filiatrault@mceermail.buffalo.edu

Graduate Student Researcher.
§ Professor.
¶ Senior Structural Engineer.

Contract/grant sponsor: Department of Homeland Security’s Federal Management Agency (FEMA)

Copyright q 2008 John Wiley & Sons, Ltd.


478 I. P. CHRISTOVASILIS ET AL.

in Japan, have demonstrated that seismic hazards pose a credible threat to residential woodframe
buildings. The 1994 Northridge earthquake in California demonstrated that both existing and new
woodframe buildings are vulnerable to strong ground shaking. More than half of the $40 billion
property loss that occurred in the Northridge earthquake was attributed to wood construction [1].
In Japan, the collapse of residential wood buildings was one of the main contributors to the heavy
death toll (more than 6400) and economic losses (more than $100 billion US) during the 1995
Kobe earthquake [2].
While seismic provisions included in current building codes [3, 4] govern the seismic design
of new engineered wood construction in the US, the performance of traditionally ‘nonengineered’
wall finish materials, such as gypsum wallboard or stucco, is not accounted for in the design
process. Additionally, many older existing woodframe buildings were poorly designed to resist
earthquake shaking. For example, thousands of woodframe, multi-unit residential buildings have
been constructed in California with tuck-under parking at the ground level. This type of wood-
frame construction usually includes a soft-story configuration that may lead to severe damage and
even collapse under strong earthquake shaking. Several buildings of this structural type, typically
constructed in the 1960s or 1970s, experienced ground story collapses during the 1994 Northridge
in California.
In order to develop consistent performance-based seismic design procedures for woodframe
buildings, their global system-level seismic fragilities under the complete range of seismic hazards
must be evaluated. This study attempts to shed some light on these issues by quantifying the seismic
collapse fragilities and collapse mechanisms of woodframe buildings using uni- and bi-directional
incremental inelastic time-history dynamic analyses.

2. DESCRIPTION OF ORIGINAL BUILDING MODELS

Two different woodframe residential building models, a townhouse and an apartment building,
are considered in this study. These buildings are part of a suite of prototype index buildings
developed under the recently completed FEMA-funded CUREE-Caltech Woodframe Project in
California for use in loss estimation and benefit-to-cost ratio analysis [5]. Detailed modeling on
these index buildings has been conducted by Isoda et al. [6]. Two design alternatives of these
two index buildings are considered: (1) the original buildings designed according to earlier code
requirements and (2) the re-designed buildings according to the requirements of the 2006 edition
of the International Building Code [3]. The construction details of the two original index building
models are summarized in Table I. Figure 1 shows architectural rendering and plan views of
the original index building models. The major structural components of the building models are
identified in Figure 1 and are described in detail in [7].

2.1. Original townhouse building


This two-story townhouse contains three units, each having approximately 170 m2 of living space
with an attached two-car garage, as shown in Figure 1. The building is assumed to be founded
on a level lot with a slab-on-grade and spread foundations. The original building is assumed to
have been built as a ‘production house’ in either the 1980s or 1990s, located in either Northern or
Southern California. The design is based on the 1988 edition of the Uniform Building Code [8]

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
INCREMENTAL DYNAMIC ANALYSIS OF WOODFRAME BUILDINGS 479

Table I. Construction details for original index woodframe buildings.


Components Construction details
Exterior walls Stucco (22 mm thick) over 11 mm oriented strand board (OSB) on outside
Gypsum wallboard 12 mm thick on inside
Furring nails (9 mm head) spaced at 150 mm on center along vertical studs used to
attach wire mesh of stucco finish to wood framing
Eight-penny common nails (diameter = 3.3 mm, length = 76.2 mm) spaced at 150,
100 or 75 mm along edges and 300 mm on field used to attach OSB panels to
framing
Interior walls Gypsum wallboard (12 mm thick) on both sides
Drywall nails (38 mm long) spaced at 175 mm on center along vertical studs
(spaced at 400 mm on center) used to attach gypsum walls to framing
Gypsum wallboard panels positioned vertically

for engineered construction. The height of the townhouse building from the first floor slab to the
roof eaves is 5.5 m and its total dead weight is 980 kN.
Seismically relevant characteristics that were intentionally featured in this townhouse include the
integral garage and for the end units, the imbalance in plan stiffness between the solid longitudinal
wall with gypsum wallboard at the common wall side versus the perforated walls with stucco or
oriented strand boards (OSB) on the exterior wall side.

2.2. Original apartment building


This index building represents a three-story, rectangular apartment with ten units (each with 85 m2
of living space) and space for mechanical and common areas, as shown in Figure 1. All walls
and elevated floors are light woodframe. It has parking on the ground floor. Each unit has two
bedrooms and one assigned parking stall. The building is assumed to have been constructed prior
to 1970 in Northern or Southern California, designed according to the 1964 edition of the Uniform
Building Code [9] and ‘engineered’ to a minimal extent. The height of the apartment building
from the first floor slab to the roof eaves is 8.2 m and its total dead weight is 1550 kN.

2.3. Construction variants


Three deterministic construction variants are considered for each index building in order to assess
the effects of construction qualities on the seismic fragilities and collapse mechanisms. The vari-
ants are representative of poor-quality, typical-quality and superior-quality construction and were
developed as part of the CUREE-Caltech Woodframe Project [5]. The managers of the CUREE-
Caltech Woodframe Project, Element 3 (Building Codes and Standards), assisted by a group of
structural engineers familiar with the seismic design of light-frame wood buildings in California,
selected four key characteristics that contribute most strongly to repair cost, and defined, based
on their experience, these characteristics for a poor-quality, typical-quality and superior-quality
variant. The characteristics of each construction variant are summarized in Table II.
The superior quality construction variant has good nailing of shear walls good connections
between structural elements, good quality stucco and good nailing of gypsum wallboard (all
components are assumed to exhibit strength and stiffness comparable with high-quality laboratory
test specimens). The typical quality variant has average nailing of shear walls and diaphragms

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
480 I. P. CHRISTOVASILIS ET AL.

Figure 1. Index woodframe building models: (a) architectural rendering; (b) plan views of townhouse
building; (c) architectural rendering; and (d) plan views of apartment building, after [7].

(5% increase in average nail spacing), average connections between structural elements (10%
reduction in shear wall stiffness and strength), average quality stucco (10% reduction in stiffness
and strength) and average nailing of interior gypsum wallboard (15% reduction in stiffness and
strength). The poor-quality variant has poor nailing of shear walls (20% increase in average nail
spacing) and poor vapor barrier installation (5% supplemental reduction in stiffness and strength),
poor connections between structural elements (20% reduction in shear wall stiffness), poor quality

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
INCREMENTAL DYNAMIC ANALYSIS OF WOODFRAME BUILDINGS 481

Table II. Summary of construction variants for original woodframe index buildings, after [6].
Components Superior quality Typical quality Poor quality
Shear walls Good nailing of shear Average nailing of shear Poor nailing of shear
walls, stiffness and walls, nail spacing 5% walls, nail spacing 20%
strength equal to that greater than that of greater than that of
obtained in high-quality superior quality superior quality plus 5%
laboratory tests supplemental reduction
in stiffness and strength
due to water damage
Connections Good connections Typical connections Poor connections
between structural between structural between structural
elements, stiffness and elements, 90% of elements, 80% of
strength equal to that stiffness and strength stiffness and strength
obtained in high-quality obtained in high-quality obtained in high-quality
laboratory tests laboratory tests laboratory tests
Exterior wall finish Good quality stucco, Average quality stucco, Poor quality stucco, 70%
stiffness and strength 90% of stiffness and of stiffness and strength
equal to that obtained in strength obtained in obtained in high-quality
high-quality laboratory high-quality laboratory laboratory tests
tests tests
Interior wall finish Superior nailing of Good nailing of interior Poor nailing of interior
interior gypsum gypsum wallboard, 85% gypsum wallboard, 75%
wallboard, stiffness and of stiffness and strength of stiffness and strength
strength equal to that obtained in high-quality obtained in high-quality
obtained in high-quality laboratory tests laboratory tests
laboratory tests

stucco (15% reduction in stiffness and strength) and poor nailing of interior gypsum wallboard
(15% reduction in stiffness and strength). Further details on how the constructions variants were
incorporated in the numerical model can be found in [6].

3. DESCRIPTION OF RE-DESIGNED PROTOTYPE BUILDING MODELS

Both original index woodframe buildings described above were designed according to earlier
building code requirements and may not be completely representative of current seismic design
practices in California. In order to evaluate the seismic response and collapse mechanisms of
woodframe buildings designed according to current building code requirements, both original index
buildings were re-designed according to the seismic requirements contained in the 2006 edition of
the International Building Code [3]. Figure 2 shows plan views of the re-designed index building
models.
In the re-design of the index woodframe buildings, the architectural layouts of the buildings
were maintained but only wood sheathed shear walls were considered for the lateral load-resisting
system in each principal direction of each building. The lateral stiffness and strength of exterior
and interior nonstructural wall finish materials were not considered in the re-design of the index
woodframe buildings. The analyses, however, are conducted with and without consideration of

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
482 I. P. CHRISTOVASILIS ET AL.

Figure 2. Plan views of re-designed index woodframe building models: (a) townhouse
building and (b) apartment building.

these nonstructural wall finishes. In addition, only the typical quality construction variant (see
Table II) is considered in the analyses of the re-designed index woodframe buildings.
The re-design of the townhouse building to the 2006 IBC resulted in two important differences:
(1) the shear capacity of the narrow wall piers along the first-level garage line had to be reduced
by a factor equal to 2bs / h, where bs is the width of each wall pier (0.9 m) and h is the height
(2.4 m), which caused a reduction of the edge nail spacing for these walls (this clause did not
exist in the 1988 edition of the UBC) and (2) new double wood shear walls between units in the
east–west direction were introduced to satisfy fire resistance requirements of the 2006 IBC. For
the re-designed apartment building, the long east–west interior wall on the first-level-incorporated
new wood sheathing, which also caused a significant increase of the lateral strength of the building
along that direction.

4. NUMERICAL MODELING

In this study, the numerical prediction of the seismic collapse of the index woodframe buildings
was based on bi-directional (horizontal) nonlinear time-history dynamic analyses; the vertical
acceleration component was not captured. For this purpose, a pancake structural model was
adopted [10]. This modeling approach simulates the three-dimensional seismic response of a
building through a degenerated two-dimensional planar analysis. The computer program SAWS—
Seismic Analysis of Woodframe Structures [10, 11], developed within the recently completed
CUREE-Caltech Woodframe Project in California, was used to analyze the prototype buildings.
In the SAWS model, the building structure is composed of two primary components: rigid
horizontal diaphragms and nonlinear lateral load-resisting shear wall elements, as illustrated

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
INCREMENTAL DYNAMIC ANALYSIS OF WOODFRAME BUILDINGS 483

Partition Wall, typ.


A B C y
A B C
3
Diaphragm (Rigid) 3

Generic Point on the Rigid


Diaphragm: (x , y )
2

f
y p
Shear Element-to-Diaphragm
1 Attachment Point
Shear Wall, typ. f
V
Θ
1
U x x
Foundation SDOF Shear Element
Origin: (0,0) Global Degrees-of-Freedom (Zero-Height), typ.
For Rigid Diaphragm
(a) (b) Foundation, typ.

Figure 3. SAWS model: (a) components of woodframe buildings considered in the SAWS program and
(b) structural model of woodframe buildings, after [10].

Capping Strength
Force
(δu,Fu) In-Cycle
K0 1 r K Degrading Strength
Nonlinear A 2 0
Backbone Curve 1 G r2Ik0 Degrading
F0 Capping Strength
Pinching and
Re-Loading Stiffness Kp
F Failure
FI 1 Displacement
E B
O r4K0 δf Displacement
1 C1 120
r3K0

Figure 4. Hysteretic behavior of shear spring element included in the SAWS Program, after [10, 11].

in Figure 3(a). The actual three-dimensional building is degenerated into a two-dimensional


planar model using zero-height shear wall spring elements connected between the diaphragms
and the foundation, as shown in Figure 3(b). The hysteretic behavior of the walls can be
characterized from cyclic test results on full-scale wall units or by an associated numerical
model [12] that predicts the walls’ load–displacement response under general quasi-static cyclic
loading. In the SAWS model, the hysteretic behavior of each wall panel (wood, interior gypsum
or exterior siding) can be represented by an equivalent nonlinear shear spring element. The
hysteretic behavior of this shear spring element includes pinching, stiffness and strength degra-
dation and is governed, in total, by ten physically identifiable parameters [10, 11], as shown in
Figure 4.
The monotonic racking response (backbone curve) of the wall model, shown in Figure 4, is
expressed in terms of the top-of-wall force F and corresponding top-of-wall horizontal displacement

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
484 I. P. CHRISTOVASILIS ET AL.

 by the following nonlinear relationship:



⎪ sgn()·(F0 +r1 K 0 ||)·[1−exp(−K 0 ||/F0 )], |||u |

F = sgn()· Fu +r2 K 0 [−sgn()·u ], |u | < |||F | (1)


0, ||>|F |

This force–deformation model for the backbone curve is characterized by six parameters:
F0 , K 0 ,r1 ,r2 , u and F , each identified in Figure 4. With this representation, the displacement u
is associated with the ultimate load Fu (capping strength), while failure of the wall occurs at the
displacement F .
The basic unloading and reloading rules defining the hysteretic wall model are also shown
in Figure 4. The force–deformation paths OA and CD follow the monotonic envelope curve as
expressed by Equation (1). All other paths are assumed to exhibit a linear relationship between
force and deformation. Unloading of the envelope curve follows a path such as AB with stiffness
r3 K 0 . Here, the wall unloads elastically. Under continued unloading, the response moves onto path
BC that is characterized by a reduced stiffness r4 K 0 . The very low stiffness along this path typifies
the pinched hysteretic response displayed by wood shear walls under cyclic loading. For wood
shear walls, this behavior is the result of previous crushing of the framing members and sheathing
panels around the connectors (in this case as the wall followed the path OA). First time loading
in the opposite direction forces the response onto the envelope curve CD. Unloading of this curve
is assumed elastic along path DE, followed by a pinched response along path EF, which passes
through the zero-displacement intercept FI , with slope r4 K 0 . Continued re-loading follows path
FG with degrading stiffness K p , as given by
 
0 
Kp = K0 (2)
max

with 0 = (F0 /K 0 ) and  a hysteretic model parameter, which determines the degree of stiffness
degradation. Note from Equation (2) that K p is a function of the previous loading history through the
last unloading displacement un of the envelope curve (corresponding to point A in Figure 4), so that

max = un (3)

where  is another hysteretic model parameter. A consequence of this stiffness degradation is that
it also produces strength degradation in the response. If on another cycle the shear wall is displaced
to un , then the corresponding force will be less than Fun , which was previously achieved. This
strength degradation is shown in Figure 4 by comparing the force levels obtained at points A and G.
In addition, with this model under continued cycling to the same displacement level, the force and
energy dissipated per cycle are assumed to stabilize beyond the second loading cycle.
With the simple approach adopted by the SAWS model, the response of the building is defined
in terms of only three-degrees-of-freedom per floor level. The capabilities of the SAWS computer
program have been investigated previously by comparing its response predictions with the shake
table tests of a large-scale, two-story, woodframe house under the CUREE-Caltech Woodframe
Project [11]. During this investigation, the test structure experienced maximum top-of-roof drift
levels of the order of 2%. At this response level, the SAWS model provided good agreement
with the test results. The capability of the SAWS program to predict the global seismic collapse

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
INCREMENTAL DYNAMIC ANALYSIS OF WOODFRAME BUILDINGS 485

intensities and the associated collapse mechanisms of wood buildings has not been investigated
yet and is the focus of this study.
In undertaking this collapse study, a number of limitations in the formulation of the SAWS
model must be recognized at the outset. First, the degenerated planar model utilized by SAWS
suppresses the vertical dimension of the building. Consequently, SAWS has no capability to capture
the vertical excitation of the test structure. Second, this degenerated planar model does not take
account of P– effects, which are not expected to become significant until near global instability
for low-rise short period wood buildings supporting light gravity loads.

5. ANALYSIS PROCEDURES

5.1. Incremental dynamic analyses


A collapse analysis procedure that is based on incremental dynamic analysis (IDA) [13] is used
in this study to predict the seismic response and collapse mechanisms of the woodframe index
buildings. Only global side-sway collapse mechanisms of light-frame wood buildings, caused by
excessive lateral wall deformations, are considered in this study, as illustrated in Figure 5.
In the IDA approach, nonlinear time-history dynamic analyses are performed for an ensemble
of earthquake ground motions scaled to a given intensity level. The scaling of the ground motions
intensity can take several forms. Because the index woodframe buildings considered in this study
exhibit short fundamental periods, the scaling is based on the median spectral acceleration of the
ensemble of ground motion records considered for the analyses at a period of 0.2 s. This means
that a single amplitude scaling factor is used for all records such that the median spectral value
taken across all records at a period of 0.2 s is equal to a target value.
From the results of each dynamic analysis, the peak inter-story drift experienced by any wall
line in the woodframe building model is retained. For each ground motion record, the analyses
are repeated for increasing intensities. The intensity of the ground motion causing collapse of the
woodframe building is defined as the point on the intensity-drift IDA exceeding a peak inter-story
drift of 7% in any of the wall line of the structure. This arbitrarily selected collapse drift value
does not affect the collapse intensities since (as it will be shown later) the IDA plots are essentially
horizontal for this drift value. The process is then repeated for all ground motions and a series of

Figure 5. Examples of global side-sway collapse mechanism of light-frame


wood buildings (USGS photos).

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
486 I. P. CHRISTOVASILIS ET AL.

intensity-displacement (IDA) plots can be obtained. The probability of collapse for a given intensity
level can then be estimated by counting the number of ground motion records causing collapse
and dividing this value by the total number of records considered in IDAs. In turn, a fragility
curve giving the relationship between the probability of collapse and the intensity measure can be
obtained and a lognormal cumulative probability distribution can be fitted to the data. In this study,
the fragility curves are computed in terms of empirical and lognormal cumulative distribution
functions (CDF) as a function of the median spectral acceleration across the earthquake ground
motion ensemble at a period of 0.2 s. The lognormal CDF of a random variable y is defined by
the median (50%) value of y m̂ y and by the dispersion parameter  expressed as the standard
deviation of the log of the values of y.

5.2. Earthquake ground motions


The suite of 22 pairs of scaled bi-directional historical ground motion records (44 records total)
considered for the ATC-63 Project [14] were used as an input to IDAs conducted in this study. The
selection of these records was based on representing extreme ground motions. Minimum limits were
imposed on the earthquake magnitude (M>6.5) and on the peak ground velocity and acceleration
(PGA >0.2 g or PGV>15 cm/s). These limits were chosen to select large motions, while ensuring
that enough motions will meet the selection criteria. Another limitation of the ground motion
ensemble is that all records were recorded at a significant distance from the fault (distance >10 km)
and on firm soils (shear wave velocities >180 m/s). Therefore, dominant near-field and soft soil
effects are not considered by this suite of records.
Each earthquake record pair was normalized in order to maintain a fairly constant variability in
spectral values across a wide period range. Table III lists the main characteristics of the ground
motion records along with the amplitude scale factor applied to each record, while Figure 6
presents 5% damped absolute acceleration response spectra for these 44 scaled records along with
the median spectral value. The design basis earthquake (DBE) and maximum credible earthquake
(MCE) intensity levels associated with these earthquake records correspond to a scaled median
spectral acceleration at a period of 0.2 s of 1.0g and 1.5g, respectively.

5.3. Directional effects


Uni-directional IDAs were carried out for each principal direction of each index building (north–
south and east–west) as well as bi-directional IDAs. For the bi-directional IDAs, each pair of ground
motions was used twice by rotating their components 90 degrees with respect to the principal
directions of each index woodframe building, thereby generating 44 analysis results per defined
spectral intensity. An increment of the 0.2 s spectral acceleration intensity measure of 0.10g was
used in the IDA. Approximately 30 000 nonlinear time-history dynamic analyses were performed
in this study.

6. ANALYSIS RESULTS

6.1. Pushover analysis results


Figure 7 presents the uni-directional pushover curves for both principal directions of each original
index woodframe building predicted by the SAWS program based on inverse triangular lateral load

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
INCREMENTAL DYNAMIC ANALYSIS OF WOODFRAME BUILDINGS 487

Table III. Characteristics of strong ground motions, after [14].


Seismic event Records
Pair Epicentral Amplitude
No. Magnitude Year Name Recording station distance (km) Scale factor
1 6.7 1994 Northridge Beverly Hills–14145 Mulhol 13.3 0.755
2 6.7 1994 Northridge Canyon Country-W Lost 26.5 0.832
3 7.1 1999 Duzce, Turkey Bolu 41.3 0.629
4 7.1 1999 Hector Mine Hector 26.5 1.092
5 6.5 1979 Imperial Valley Delta 33.7 1.311
6 6.5 1979 Imperial Valley El Centro Array #11 29.4 1.014
7 6.9 1995 Kobe, Japan Nichi-Akashi 8.7 1.718
8 6.9 1995 Kobe, Japan Shin-Osaka 46.0 1.099
9 7.5 1999 Kocaeli, Turkey Duzce 98.2 0.688
10 7.5 1999 Kocaeli, Turkey Arcelik 53.7 1.360
11 7.3 1992 Landers Yermo Fire Station 86.0 0.987
12 7.3 1992 Landers Coolwater 82.1 1.073
13 6.9 1989 Loma Prieta Capitola 9.8 0.822
14 6.9 1989 Loma Prieta Gilroy Array #3 31.4 0.880
15 7.4 1990 Manjil, Iran Abbar 40.4 0.787
16 6.5 1987 Superstition Hills El Centro Imp. Co. Cent 35.8 0.870
17 6.5 1987 Superstition Hills Poe Road (temp) 11.2 1.362
18 7.0 1992 Cape Mendocino Rio Dell Overpass-FF 22.7 1.516
19 7.6 1999 Chi-Chi, Taiwan CHY101 32.0 0.636
20 7.6 1999 Chi-Chi, Taiwan TCU045 77.5 0.563
21 6.6 1971 San Fernando LA-Hollywood Stor FF 39.5 2.096
22 6.5 1976 Driuli, Italy Tolmezza 20.2 1.440

distributions. The results are presented for each construction quality of each building. The vertical
axis of each plot is expressed by a seismic coefficient defined as the total lateral load applied (base
shear) divided by the total dead weight of each building. The design seismic coefficient for each
building is also indicated. The horizontal axis of each plot is expressed by the roof drift ratio,
defined as the lateral displacement at the center of the building roof divided by the height from
the first floor (ground) slab to the roof eaves of each building.
For both original index buildings, the peak lateral loads are reached at small roof drift ratios
(0.3–0.5%). This is the result of the high lateral stiffness and limited ductility of the wall finish
materials, particularly exterior stucco, included in the models. These stiffer and brittle finish
materials fail before the underlying wood shear wall reaches its capacity. Note, however, that the
lateral deformed shape of both buildings is influenced by soft-story mechanisms, for which a large
portion of the roof drift ratio is mobilized in the first story (as it will be seen later). Compounded
by the significant plan eccentricity of both buildings, the drift ratios of the first-level wall lines
(first story drift) experiencing the maximum deformations are 2 to 3 times larger than the roof
central drift ratios shown in Figure 7. For both buildings, the lateral strength in the (long) east–
west direction is higher than in the (short) north–south direction. The construction quality has a
substantial influence in both the stiffness and strength of both buildings but has only a minor effect
in their displacement capacities. Increasing the nail density along wood panel edges increases the
lateral stiffness and strength but has little effect on the displacement at failure that is governed by
the global kinematics of the wall assembly.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
488 I. P. CHRISTOVASILIS ET AL.

7.0
Median Spectrum
6.0

Spectral Acceleration, Sa (g)


5.0 5% Damping

4.0

3.0

2.0

1.0

0.0
0 0.51 1.5 2 2.5 3
Period (sec)

Figure 6. Acceleration response spectra at 5% damping of scaled ground motion records.

1 1
East-West North-South
0.8 0.8
Seismic Coefficient
Seismic Coefficient

0.6 0.6

0.4 0.4
Typical Quality Typical Quality
0.2 Superior Quality 0.2 Superior Quality
Design Seismic Coefficient = 0.183 Design Seismic Coefficient = 0.183
Poor Quality Poor Quality
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(a) Roof Drift Ratio (%) Roof Drift Ratio (%)

1 1
East-West North-South
0.8 0.8
Seismic Coefficient
Seismic Coefficient

Typical Quality Typical Quality


Superior Quality Superior Quality
0.6 0.6
Poor Quality Poor Quality

0.4 0.4

0.2 0.2
Design Seismic Coefficient = 0.133 Design Seismic Coefficient = 0.133

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(b) Roof Drift Ratio (%) Roof Drift Ratio (%)

Figure 7. Pushover curves of original index woodframe buildings: (a) townhouse


building and (b) apartment building.

Figure 8 presents the uni-directional pushover curves for both principal directions of the typical
quality construction variant of each index woodframe building re-designed to 2006 IBC. The
results are presented with and without consideration of the wall finish materials and are also

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
INCREMENTAL DYNAMIC ANALYSIS OF WOODFRAME BUILDINGS 489

1 1
East-West Re-Designed Typical Quality with Wall Finishes North-South Re-Designed Typical Quality with Wall Finishes
Original Typical Quality with wall Finishes Original Typical Quality with Wall Finishes
0.8 0.8
Seismic Coefficient

Seismic Coefficient
Re-Designed Typical Quality without Wall Finishes Re-Designed Typical Quality without Wall Finishes

0.6 0.6

0.4 0.4

1988 UBC Seismic Design Coefficient = 0.183 1988 UBC Seismic Design Coefficient = 0.183
0.2 0.2
2006 IBC Seismic Design Coefficient = 0.154 2006 IBC Seismic Design Coefficient = 0.154

0 0
0 1 2 3 4 0 1 2 3 4
(a) Roof Drift Ratio (%) Roof Drift Ratio (%)

1 1
East-West Re-Designed Typical Quality with Wall Finishes North-South Re-Designed Typical Quality with Wall Finishes
Original Typical Quality with wall Finishes Original Typical Quality with Wall Finishes
0.8
Seismic Coefficient

0.8

Seismic Coefficient
Re-Designed Typical Quality without Wall Finishes Re-Designed Typical Quality without Wall Finishes

0.6 0.6

0.4 0.4

0.2 2006 IBC Seismic Design Coefficient = 0.154 0.2 2006 IBC Seismic Design Coefficient = 0.154

1964 UBC Seismic Design Coefficient = 0.133 1964 UBC Seismic Design Coefficient = 0.133

0 0
0 1 2 3 4 0 1 2 3 4
(b) Roof Drift Ratio (%) Roof Drift Ratio (%)

Figure 8. Pushover curves of typical quality original and re-designed index woodframe buildings with and
without wall finishes: (a) townhouse building and (b) apartment building.

compared with the pushover curves obtained for the typical quality construction of the original
index woodframe buildings. The re-design of both buildings increases their lateral strength in
the east–west direction but has very little effect in the north–south direction. For both original
buildings, wood shear walls were already present along the wall lines in the north–south direction.
For the townhouse building, the re-design involved new double wood shear walls between units in
the east–west direction, which causes a significant increase in the lateral strength of the building
in that direction. For the re-designed apartment building, the long east–west interior wall on the
first-level-incorporated new wood sheathing, which causes also a significant increase of the lateral
strength of the building along that direction.
Note that when the wall finishes are not considered, the wood structure exhibits lower strength
values with peak lateral loads being reached at much larger roof drift ratios (1.5–2.5%) than the
ratios obtained for the re-designed structures incorporating the wall finishes (0.3–0.5%). This is
as a result of the stiffness and strength incompatibility between the wall finish materials (stucco
and gypsum wallboard) and the wood sheathed walls.

6.2. Free vibration analysis results


Table IV presents the computed first two initial natural periods (under elastic conditions, that is,
each shear spring element in the SAWS model had initial stiffness) of each construction variant

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
490 I. P. CHRISTOVASILIS ET AL.

Table IV. Initial natural periods of original and re-designed index buildings.
Natural period (sec)
Building Mode/direction Poor quality Typical quality Superior quality
Original 1/north–south 0.189 0.173 0.161
Townhouse 2/east–west 0.169 0.155 0.144
Original 1/north–south 0.276 0.249 0.233
Apartment 2/east–west 0.245 0.222 0.222
With wall finishes Without wall finishes
Re-designed 1/north–south 0.176 0.429
Townhouse 2/east–west 0.151 0.347
Typical quality
Re-designed 1/north–south 0.245 0.514
Apartment 2/east–west 0.219 0.486
Typical quality

for both original and re-designed index woodframe buildings considered in this study. The natural
periods of each original building are increased by approximately 10% for each lower quality
construction (i.e. increment of 10% from superior quality to typical quality and another increment
of 10% from typical quality to poor quality). This increase in natural periods is approximately the
same for each mode of vibration.
The fundamental periods of the re-designed buildings are essentially identical as that of the
original typical quality construction. On the other hand, the fundamental period for each re-designed
building is more than doubled when the wall finishes are removed from the structural model. This
result clearly indicates the important contribution of the wall finishes to the lateral stiffness and
dynamic characteristics of the index woodframe buildings.

6.3. IDA results


Figure 9 illustrates typical IDA results for the typical quality variant of the original townhouse
building subjected to bi-directional excitations. The IDA plots shown in Figure 9(a) are monotonic
and essentially bi-linear, which greatly simplifies the evaluation of the collapse level for individual
ground motion. The empirical collapse fragility curve constructed from these IDA plots is shown
in Figure 9(b) along with the best-fit lognormal CDF. The median collapse value and the dispersion
factor  are also shown. The collapse probabilities at the DBE and MCE levels are also identified
from the fragility curve.
Fragility information extracted from all IDA results is presented in Tables V and VI for the
original and re-designed index woodframe buildings, respectively. Listed in these tables are the
median and  values for all the computed lognormal CDFs along with the probabilities of collapse
associated with the DBE and MCE levels.
The construction quality has a significant influence on the computed collapse fragility of both
original buildings (Table V). For the bi-directional analyses on the original townhouse building, the
median collapse intensity (expressed in terms of the median spectral acceleration of the ensemble
of ground motion records considered for the analyses at a period of 0.2 s) increases by 15% when
the construction quality is upgraded from poor to typical and by 9% from typical to superior. The
corresponding increases in median collapse intensities are 13 and 10% for the original apartment

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
INCREMENTAL DYNAMIC ANALYSIS OF WOODFRAME BUILDINGS 491

7.0 1
Bi-Directional Bi-Directional
6.0
Median Sa at 0.2 sec (g)

Collapse 0.8

Probability of collapse
Limit-State
5.0

4.0 0.6
Empirical CDF
3.0 Lognormal CDF
0.4
Median =1.81 g
2.0 P[Collapse/MCE] = 0.31 β = 0.38
0.2
1.0
P[Collapse/DBE] = 0.06
0.0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7
(a) Maximum Interstory Drift (%) (b) Median Sa at 0.2 sec (g)

Figure 9. Results of bi-directional IDAs for typical quality original townhouse building: (a) IDA plots
and (b) collapse fragility curves.

Table V. Fragility information for original index woodframe buildings.


Lognormal
CDF Probability of collapse
DBE MCE
Excitation Median level level
Building Variant direction (g)  (Sa−0.2 = 1.0g) (Sa−0.2 = 1.5g)
Original townhouse Typical East–west 2.26 0.39 0.02 0.15
North–south 2.26 0.44 0.03 0.17
Bi-directional 1.81 0.38 0.06 0.31
Poor East–west 1.92 0.37 0.04 0.25
North–south 1.92 0.42 0.06 0.28
Bi-directional 1.58 0.34 0.09 0.43
Superior East–west 2.53 0.39 0.01 0.09
North–south 2.44 0.45 0.02 0.14
Bi-directional 1.97 0.38 0.04 0.24
Original apartment Typical East–west 1.79 0.44 0.09 0.34
North–south 1.65 0.43 0.12 0.41
Bi-directional 1.40 0.33 0.15 0.58
Poor East–west 1.63 0.39 0.11 0.41
North–south 1.52 0.42 0.16 0.49
Bi-directional 1.24 0.33 0.25 0.71
Superior East–west 2.04 0.43 0.05 0.24
North–south 1.77 0.43 0.09 0.35
Bi-directional 1.54 0.32 0.09 0.47

building. The dispersion parameter  remains almost constant around 0.40 for all cases since only
the variation in ground motion characteristics is captured by IDAs. The excitation direction has also
a significant influence on the collapse fragility of both original buildings. The bi-directional analyses
reduce the median collapse levels obtained from the uni-directional analyses by approximately
20% for all three construction variants of both original buildings. This reduction can be attributed

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
492 I. P. CHRISTOVASILIS ET AL.

Table VI. Fragility information for typical quality construction of re-designed index woodframe buildings.
Lognormal
CDF Probability of collapse
Wall Excitation DBE level MCE level
Building finishes direction Median (g)  (Sa−0.2 = 1.0g) (Sa−0.2 = 1.5g)
Re-designed townhouse Included East–west 2.60 0.33 0.01 0.05
North–south 2.15 0.36 0.02 0.16
Bi-directional 2.00 0.40 0.04 0.24
Not included East–west 2.15 0.46 0.05 0.22
North–south 1.90 0.48 0.09 0.31
Bi-directional 1.70 0.43 0.11 0.39
Re-designed apartment Included East–west 2.38 0.36 0.01 0.10
North–south 1.65 0.44 0.13 0.41
Bi-directional 1.47 0.36 0.14 0.52
Not included East–west 2.15 0.46 0.05 0.22
North–south 1.88 0.44 0.08 0.31
Bi-directional 1.52 0.32 0.12 0.49

to the fact that, when a pair of bi-directional motions is rotated along each principal direction of
a building, the larger component is applied in both analyses along one of the directions of the
building, which increases its likelihood of collapse compared with uni-axial IDAs, where the large
component is applied only once. In addition, construction quality and excitation direction influence
significantly the resulting collapse probabilities at the MCE level.
The seismic weakness of tuck-under parking structures is evident from the fragility information
presented in Table V. The probability of collapse of the poor quality original apartment building
is more than 70% under the bi-directional MCE level ground motions. Even the superior quality
variant has a corresponding collapse probability near 50%.
The collapse fragility of the typical quality re-designed townhouse with wall finishes (Table VI)
is reduced substantially in the east–west direction compared with the collapse fragility of the
typical quality original townhouse building (see Table V). This is the direct result of the re-designed
parallel wood shear walls between units in the east–west direction. The same observation can
be made between the typical quality re-designed and original apartment buildings in the east–
west direction as a result of the incorporation of wood sheathing in the long interior wall of the
re-designed apartment. When wall finishes are excluded from the structural models, the collapse
fragility of both re-design buildings is increased in both directions compared with the typical
quality original buildings.

6.4. Collapse mechanisms


Figures 10 and 11 illustrate the collapse mechanisms associated with the uni- and bi-directional
IDAs on the original index woodframe townhouse and apartment building, respectively. Each figure
shows the percentage of collapse cases initiated in a given wall line (i.e. having maximum inter-
story drift at collapse) in the building. Each percentage value shown in the figures was obtained
by the number of earthquake records causing a particular wall line to reach its collapse limit state
first in the building divided by 44 records; therefore, the sum of the percentages shown for a given
building is equal to 100%. All collapse mechanisms are associated with wall lines located on the

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
INCREMENTAL DYNAMIC ANALYSIS OF WOODFRAME BUILDINGS 493

Figure 10. Distribution of collapse mechanisms, original townhouse: (a) uni-directional


IDAs and (b) bi-directional IDAs.

Figure 11. Distribution of collapse mechanisms, original apartment: (a) uni-directional


IDAs and (b) bi-directional IDAs.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
494 I. P. CHRISTOVASILIS ET AL.

Figure 12. Distribution of collapse mechanisms, typical construction quality re-designed index buildings
with wall finishes: (a) townhouse uni-directional IDAs; (b) townhouse bi-directional IDAs; (c) apartment
uni-directional IDAs; and (d) apartment bi-directional IDAs.

first story of each original building (weak story collapse). For both original buildings, the collapse
mechanism associated with uni-directional IDAs (Figures 10(a) and 11(a)) corresponds to the same
wall line for all ground motions (100% collapse) and is independent of the construction quality.
For bi-directional IDAs, a variety of collapse mechanisms occur. This distribution of collapse
mechanisms is also influenced by the construction quality although the majority of collapses occur
along the same first story wall lines (north for the townhouse, east for the apartment).
Figures 12 and 13 illustrate the collapse mechanisms associated with the uni-directional and
bi-directional IDAs on the typical construction quality re-designed index woodframe buildings
with and without wall finishes, respectively. For both re-designed buildings with wall finishes
(Figure 12), all collapse mechanisms are associated with wall lines located on the first story
(weak story collapse). For the re-designed townhouse with and without wall finishes, all collapse
mechanisms associated with uni-directional IDAs in the north–south direction occur along the west
wall line (Figures 12(a) and 13(a)) as opposed to the east wall for the original townhouse building
(see Figure 10(a)). This migration of the collapse mechanism from the east to the west wall is
due to the narrow wall piers along the east garage wall of the townhouse building that triggers a
special clause of the 2006 IBC leading to closely spaced nailing of the wood shear walls along
the east wall of the re-designed townhouse.
For both re-designed buildings without wall finishes, second-level collapse mechanisms are
observed (see Figure 13) indicating a more uniform distribution of lateral displacements among
the various floor levels.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
INCREMENTAL DYNAMIC ANALYSIS OF WOODFRAME BUILDINGS 495

Figure 13. Distribution of collapse mechanisms, typical construction quality re-designed index buildings
without wall finishes: (a) townhouse uni-directional IDAs; (b) townhouse bi-directional IDAs; (c) apartment
uni-directional IDAs; and (d) apartment bi-directional IDAs.

7. CONCLUSIONS

This study conducted incremental dynamic analyses on two original and re-designed index wood-
frame buildings with and without wall finishes: a townhouse and an apartment building. Three
deterministic construction variants (poor, typical and superior) were considered for each original
index building in order to assess the effects of construction qualities on the seismic fragilities
and collapse mechanisms. The re-designed buildings were based on the seismic requirements of
the 2006 Edition of the International Building Code. Only the typical construction quality was
considered for these re-designed index buildings, but models with and without wall finishes were
included. Based on the results obtained, the following conclusions can be drawn:
• The construction quality had a substantial influence on the stiffness and strength of both original
buildings, but had only a minor effect in their displacement capacities.
• The construction quality had a significant influence on the collapse fragility of both original
buildings. For the bi-directional analyses on the townhouse building, the median collapse
intensity increased by 15% when the construction quality was upgraded from poor to typical
and by 9% from typical to superior. The corresponding increases were 13 and 10% for the
apartment building.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe
496 I. P. CHRISTOVASILIS ET AL.

• The collapse fragility of the typical quality re-designed townhouse with wall finishes was
reduced substantially in the east–west direction compared with the collapse fragility of the
typical quality original townhouse building. The same observation can be made between
the typical quality re-designed and original apartment buildings in the east–west direction.
When wall finishes were excluded from the structural models, the collapse fragility of both
re-design buildings was increased in both directions compared with the typical quality original
buildings.
• The excitation direction had a significant influence on the collapse fragility of both original and
re-designed buildings. The bi-directional analyses reduced the median collapse levels obtained
from the uni-directional analyses by approximately 20%.

ACKNOWLEDGEMENTS
The material described in this paper was developed as background material for the ATC-63 Project
‘Quantification of Building System Performance and Response Parameters’, which was funded by the
Department of Homeland Security’s Federal Management Agency (FEMA). The authors kindly acknowl-
edge Dr Charles Kircher, Chair of the ATC-63 Project Management Committee, Mr Christopher Rojahn,
ATC-63 Project Executive Director, and Mr John Heinz, ATC-63 Project Quality Coordinator, for their
guidance and support during the course of this project. The authors acknowledge also Kelly Cobeen from
Cobeen and Associates Structural Engineering, who re-designed the index buildings considered in this
study.

REFERENCES
1. Kircher CA, Reitherman RK, Whitman RV, Arnold C. Estimation of earthquake losses to buildings. Earthquake
Spectra 1997; 13(4):703–720.
2. Takashi N. Lesson learned from hurricane Katrina in 2005 and Kobe earthquake in 1995. Academic Emergency
Medicine 2006; 13(8):9102–9103.
3. ICC (2006). International Building Code-2006 Edition, International Code Council, Falls Church, VA.
4. ICBO. 1997 Uniform building code. International Conference of Building Officials, Whittier, CA, 1997.
5. Porter KA, Beck JL, Seligson HA, Scawthorn CR, Tobin LT, Young R, Boyd T. Improving loss estimation for
woodframe buildings. Report No. W-18, Consortium of Universities for Research in Earthquake Engineering,
Richmond, CA, 2001; 293.
6. Isoda H, Folz B, Filiatrault A. Seismic modeling of index woodframe buildings. Report No. W-12, Consortium
of Universities for Research in Earthquake Engineering, Richmond, CA, 2001; 144.
7. Reitherman R, Cobeen K, Serban K. Design documentation of woodframe project index buildings. Report No.
W-29, Consortium of Universities for Research in Earthquake Engineering, Richmond, CA, 2003; 258.
8. ICBO. 1988 Uniform building code. International Conference of Building Officials, Whittier, CA, 1988.
9. ICBO. 1964 Uniform Building Code. International Conference of Building Officials, Whittier, CA, 1964.
10. Folz B, Filiatrault A. Seismic analysis of woodframe structures I: model formulation. ASCE Journal of Structural
Engineering 2004; 130(8):1353–1360.
11. Folz B, Filiatrault A. Seismic analysis of woodframe structures II: model implementation and verification. ASCE
Journal of Structural Engineering 2004; 130(8):1361–1370.
12. Folz B, Filiatrault A. Cyclic analysis of wood shear walls. ASCE Journal of Structural Engineering 2001;
127(4):433–441.
13. Vamvatsikos D, Cornell AC. The incremental dynamic analysis and its application to performance-based earthquake
engineering. 12th European Conference on Earthquake Engineering, London, U.K., 2002. Paper No. 479, on
CD-ROM.
14. FEMA. Quantification of building seismic performance factors. ATC-63 Project Report—90% Draft, FEMA P695,
Applied Technology Council, Redwood City, CA, 2008.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:477–496
DOI: 10.1002/eqe

You might also like