Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Thin-Walled Structures 176 (2022) 109303

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Full length article

Seismic fragility assessment of non-structural drywall partitions with screw


and adhesive fixing through multi-axis cyclic testing
Hamidreza A. Yazdi a , Javad Hashemi b ,∗, Emad Gad b
a
Etex Australia (Siniat), Matraville, NSW 2036, Australia
b
Swinburne University of Technology, Hawthorn, VIC 3122, Australia

ARTICLE INFO ABSTRACT


Keywords: Drywall partitions are the most common type of non-structural walls used in modern multi-story buildings.
Drywall partitions Damage investigation of these walls during the past earthquakes has shown their vulnerability under small
Non-structural components to moderate inter-story drifts leading to significant economic losses. To propose innovative solutions to this
Multi-axis testing
problem, accurate damage assessment must be conducted using the data obtained from realistic experiments. In
Damage assessment
this study, the classic performance-based earthquake engineering framework is used as a method for performing
a probabilistic estimation of damage progression in the drywall partitions in a limited-ductility reinforced-
concrete building. The Multi-Axis Substructure Testing system at Swinburne University of Technology is used
to conduct large-scale three-dimensional cyclic tests on two assemblies of drywall partitions. The assemblies
included two types of fixings, namely, screw fixing as a common construction method, and adhesive fixing
recently developed for mitigating the creaking noise under serviceability wind loads. The results of the
experiments are used to define three levels of damage states to produce fragility curves of drywall systems
given the building inter-story drifts. Vulnerability curves are then developed to account for the uncertainties in
the building response due to record-to-record variability, and the factors controlling the progression of damage
in partition walls such as the variability in material and construction quality.

1. Introduction In recent years, light-weight steel drywalls constructed with cold-


formed steel stud and track framing system and sheathed with gypsum
The previous earthquakes such as San Fernando (1971), Northridge lining boards (plasterboard) have been widely used as non-structural
(1994), Christchurch (2011), and recently, the 2016 Central Italy earth- partition walls. In these components, the level of inter-story drifts that
quake have shown that in some cases the damage of non-structural causes damage is usually less than the prescribed design limits in the
components of the buildings have more significant effects on the occu- current standards. Under inter-story drifts imposed by the building side
pants safety, business interruption and the economic losses compared sway, the drywalls suffer moderate to intensive damages that often
to the damage of structural elements [1–4]. Among the non-structural
require repair or replacement of the walls. Furthermore, these damages
components, the interior and exterior partition walls are the most
may lead to the likely loss of fire, acoustic or thermal rating of internal
sensitive and vulnerable parts since during a seismic event these ele-
partitions walls. The loss of fire rating, in particular, can be potentially
ments are subjected to lateral drifts transferred from the surrounding
fatal to building occupants.
structural framing system. However, unlike the structural elements,
Further, in moderate seismicity regions, there is limited studies
non-structural walls are not normally designed for seismic actions due
to insufficient design and construction guidelines. Therefore, these conducted on the evaluation of the seismic resistant of cold-formed
elements are vulnerable even under low- to moderate inter-story drifts, steel drywalls. Kim and Shin [6] have recently used shake table tests
and consequently, take up most repair costs of the building after to assess the seismic resistance of drywalls built in regions with low-
an earthquake. Whitman et al. [5] investigated the economic losses to moderate seismic activities. The results of the large-scale shaking
of the buildings damaged after the San Fernando earthquake (1971) table tests showed that the seismic behavior of drywalls subjected to
and estimated that more than half of the building repair costs were lateral deformation can be improved by decreasing the stud spacing.
spent on the internal and external non-structural partition walls. Thus, Also, through statistical investigation, they reported that in earthquake
alleviation of the damage to non-structural walls could significantly with low intensities, the partition walls contribute the most towards
reduce economic losses in buildings after seismic events. losses due to repair or replacement.

∗ Corresponding author.
E-mail address: jhashemi@swin.edu.au (J. Hashemi).

https://doi.org/10.1016/j.tws.2022.109303
Received 8 October 2021; Received in revised form 2 March 2022; Accepted 5 April 2022
Available online 27 April 2022
0263-8231/© 2022 Elsevier Ltd. All rights reserved.
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

The main factors contributing to the poor performance of drywalls noise [20]. It is critical to investigate the seismic behavior of the two
are the lack of standard assembly during construction, and the lack of selected assemblies to find how different fixing methods can affect the
research on improving construction detailing and the development of performance of the drywall system under seismic actions.
innovative technologies for damage mitigation of drywalls. Tasligedik The specimens are tested using the Multi-Axis Substructure Testing
et al. [3] highlighted that after Christchurch earthquake in 2011, (MAST) system at Swinburne University of Technology [21] that is
the drywall partitions were mostly damaged at the corners and their capable of simulating complex 6 degrees-of-freedom (6-DOF) boundary
joint boundaries due to inappropriate finishing, creating moderate to forces and deformations. The data obtained from the experimental tests
extensive crack and crush at these interfaces. Further, the loss as- and the numerical simulation are then used to develop vulnerability
sessment from the Canterbury earthquakes conducted by Arifin et al. curves that present the probability of the partition walls being in a
[7] on a 22-story building indicated that the annual losses can be specific damage state given a specified level of earthquake intensity.
reduced by improving the seismic resistance of building contents and These curves can be used to assess the risk of failure and verify the
non-structural elements such as the drywall partitions as one of the implementation of new approaches for minimizing the damage and
most vulnerable elements. Comparison of the cost–benefit analyses of improve the resilience of drywall partitions with screw and adhesive
buildings with and without well designed drywall indicated that the fixing methods during seismic events.
added costs for improving the performance of drywall partitions during It should be noted that in construction of these specimens, the
the construction stage are likely to be recovered due to the lowered surrounding boundary elements of the structures are modeled only
repair costs in a short period of time, i.e. ∼10–20 years for a 22-storey at the top with connections to the MAST crosshead and at the base
building. with the strong floor to represent the detailing in actual construction.
Several experimental research studies have been conducted dur- This also allows to simulate the transfer of the inter-story drifts from
ing the past years to investigate and propose new methods to en- the floors to the drywalls. However, the interaction of the vertical
hance the strength and deformation capacities of the drywall partitions surrounding structural elements such as columns or shear walls to
[8–17]. Araya-Letelier et al. [17] proposed a friction/sliding connection the drywall partitions are not considered since the focus of this study
for drywall partitions. This connection introduced a sliding interface is to assess the effect of partitional wall-to-wall and ceiling-to-wall
between the steel top track of the wall and the supporting structural connectivity and their mutual interactions. More information on the
element, which is the concrete slab. The results of quasi-static cyclic importance of the simulation of structural boundary elements and the
tests carried out on three specimens indicated that after using the pro- interactions between structural and non-structural components can be
posed connection, the building can experience the story drift of up to
found in [11], and [18].
1.0% without causing any damage in drywall partitions. Furthermore,
Fiorino et al. [18] investigated the unidirectional and bidirectional
2. Creaking noise assessment of drywall partitions and test speci-
seismic performance of drywall partitions and ceiling systems using
mens
shaking table tests considering the fixed and sliding connection con-
figurations at the wall and ceiling edges. The results of this study
Prior to conducting the seismic assessment study, the authors re-
show an improved seismic resistant of the test specimen using the
cently investigated wind-induced structure-borne noise originating from
sliding connections in comparison to fixed connections, and the damage
a typical light-weight wall and ceiling system due to movement in tall
states defined for the specimen with sliding connections were in higher
structures under the serviceability wind loads [20]. The investigation
inter-story drift ratios (IDR).
was conducted at Swinburne University Technology consisting of large-
Moreover, limited studies are available on the seismic vulnerability
scale 3D experiments on an assembly of drywall system composed
assessment of drywall partitions that are validated based on large-
of cold-formed metal framing and plasterboard (drywall system). The
scale testing of the three-dimensional (3D) assemblies of partition walls
MAST system was used in these experiments which is a unique facil-
and suspended ceiling systems. The application of multi-axis testing
ity that enables full-scale 3D testing of structural and non-structural
allows to simulate more realistically the complex load transfer at the
boundaries of the drywalls, and to capture more accurately the in-plane components. In these experiments, the specimen was subjected to slow
and out-of-plane response of these elements and their mutual inter- lateral cyclic displacements from the top. The test simulated the motion
actions. Hasani and Ryan [19] investigated the effect of out-of-plane experienced by plasterboard wall and ceiling systems in a typical
drift on in-plane damage assessment of two different configurations high-rise building exposed to lateral wind loads.
of partition walls, i.e. two straight partition walls and two C-shaped Fig. 1 shows a typical standard assembly and junction details of
wall assemblies, subjected to quasi-static bidirectional loading. The a drywall system with screw-only fixing method, which is normally
effectiveness of the innovative construction details on damage states used in high-rise building. For the first round of the experiments, a
of the drywalls was discussed, and it was shown that the distributed small room (2.5×2.5×3.0 m high) according to this assembly guide was
gap detailing delayed the damage to about 1% inter-story drift ratio. constructed incorporating cold-formed steel sections and plasterboard
The focus of this study is to conduct a large-scale experimental lining systems fixed by using screws. The displacement protocol for
program on the vulnerability assessment of 3D assemblies of non- each test consisted of constant amplitude sinusoidal cycling in displace-
structural drywall partitions and suspended ceiling systems under ment control mode up to H/500 inter-story drift limit, for serviceability
multi-axis cyclic loading. For this purpose, the classic performance- consideration. The displacement of the MAST crosshead with respect to
based earthquake engineering (PBEE) framework is implemented on the strong floor induced inter-story drifts in the rigidly attached room,
the model of a typical 5-story limited ductile reinforced-concrete (RC) thus emitting noises.
framed building. Incremental dynamic analysis (IDA) is performed by The location and magnitude of creaking noises in the test specimen
subjecting the building model to a set of earthquake ground motion were then identified and recorded with sensitive and sophisticated
records with increasing intensities. The damage analysis is then con- acoustic cameras and microphones, and external acoustic engineers
ducted on two large-scale rooms constructed by drywall partitions to interpreted the results. The outcomes of the analysis of the acoustic
obtain realistically the local and global responses of drywall partitions measurements and the test observations showed several locations with
when subjected to the previously defined inter-story drift demands. The the peak sound pressure level in the test specimen including at the
first selected configuration of drywall system is the typical construction height of the wall head track, at the height of the ceiling, around
method with screw fixing. The second specimen is a room constructed the door opening, at the floor height, etc. This indicated that noise
with adhesive fixing method to secure the plasterboard to the metal emissions were due to the interaction of the plasterboard with the
framing that has been found very effective in combating friction gen- framing as well as at the framing junctions. Within the test routine,
erated noises caused by building movement to mitigate the creaking few changes to the frame, plasterboards, connections, etc. were made to

2
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Fig. 1. Configuration of a typical drywall assembly with screw fixing and connection details with screw and standard head track.

investigate the noise and performance of different room configurations, FE model of this building is developed and calibrated using the previous
which led to the development of several practical recommendations to experimental data. The building is subjected to IDA analysis using a
mitigate creaking noise. It was, however, noted that the creaking noise set of suitable ground motion records. Two large-scale assemblies of a
was significantly reduced when mechanical connections (for example 3D room with drywall partitions and a ceiling system are constructed
screw connections) were reduced. and tested using the Multi-Axis Substructure Testing (MAST) system to
According to all acoustic measurements and the experimental ob- replicate the conditions experienced in a typical multi-story RC build-
servations, the following recommendations were proposed to minimize ing. Fig. 3 shows a 3D schematic view of the RC framed building and
the creaking noise in drywall partition with ceiling system assembly in the first test specimen constructed under the MAST system. The drywall
high-rise buildings: partitions are constructed in-situ using cold-formed steel sections and
plasterboard lining systems following proprietary specifications showed
– Using adhesive material instead of mechanical (screw) connec-
in Figs. 1 and 2.
tions between plasterboard and wall studs as much as possible.
The experimental testing will provide data on the correlation be-
– Maximizing stud centers to 600 mm spacing using stronger pro-
tween drywalls damages and the structural response in terms of the
files to reduce the number of mechanical fixings (fewer mechan-
inter-story drifts that are imposed on the specimens using a bidirec-
ical fixings lead to fewer source points for creaking noise).
tional cloverleaf cyclic loading protocol. The results of the experiments
– Removing the fixing between the studs at the wall corners and
are then used to accurately capture and quantify the physical damages
intersections.
including the crack openings, bulging and separations of the plaster-
– Fix only two sides of the ceiling to the walls.
boards, yielding, deformation and buckling of the cold-formed steel
– Using of deflection head track instead of normal track at the top wall studs and tracks, and overall residual deformation of the partition
The follow up experiments indicated that the adhesive method of walls. The observed damages are used to define damage states (DSs)
securing the plasterboard to the metal framing combined with the associated to the inter-story drift of the building. Finally, by integrating
above recommendations can be very effective in combating friction the response of the building from IDA, and the damage assessment
generated noises caused by building movement. Fig. 2 shows the typical results of the tested drywall system, vulnerability curves are generated
configuration of the recommended drywall system with adhesive fixing that indicate the probability of the partition walls being in a DS given
method and its junction details. a specified level of earthquake intensity. The details of the numerical
While the application of the proposed solutions enhanced the ser- and experimental simulations are presented in the following sections.
viceability of drywalls, it is essential to investigate the seismic behavior
of the assemblies with the recommended improvements. In this study, 3.1. Description of prototype RC building
the two configurations of drywall system showed in Figs. 1 and 2,
respectively, representing the typical construction and a solution for The case study building consists of a symmetrical five-story (h1 =
mitigating the creaking noise, are further studied to investigate the 5 m, htyp = 4 m) five-by-five bay (b = 8.4 m) RC ordinary moment
seismic vulnerability under multi-axis cyclic loading. frame office building. All the columns in the building are identical
and have 500 mm × 500 mm cross-sections with a longitudinal rein-
3. Application of PBEE for seismic performance assessment of forcement ratio of 1.3%. A 3D FE model of the building is developed
drywall partitions in OpenSees [22]. The developed FE model follows the strong-beam
weak-column design that is widespread in regions of low- to moderate
The concept of performance-based earthquake engineering (PBEE) is seismicity. The structure beams are modeled stiff-elastic and columns
implemented on a case study RC building that houses the drywalls. The are modeled using beam-with-hinges elements, where the nonlinear

3
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Fig. 2. Configuration of specimen with adhesive fixing and the joints details with deflection head track.

Fig. 3. Schematic view of the prototype building and assembly of the drywall test setup under the MAST.

behavior is considered to occur within a finite-length at both ends moderate seismic regions. Therefore, these 22 far-field earthquake
based on the distributed-plasticity concept. The plastic hinges of the records were used in this study by undertaking ground motion scaling,
columns are calibrated to the experimental results of a 1:2 scale multi- where the pseudo-acceleration response spectrum of each ground mo-
axis hybrid simulation that was conducted by Hashemi et al. [23]. As tion was anchored to the same intensity level at the fundamental period
the structure shows a soft first-story collapse mechanism, the response of vibration of the building. The details of the selected ground motions
of the building is purely governed by the first-story columns, which are presented in Table 1 and their response spectra for 5% damping are
have been calibrated to pure experimental results. This will improve plotted in Fig. 4.
the reliability of predicting the demands on non-structural partition
walls since they are directly subjected to the movements of surrounding 3.3. Structural response analysis
structural framing system.
Structural response analysis was carried out to probabilistically
3.2. Seismic hazard analysis assess the response of the structure to the various ground motions
as the primary source of uncertainty, given a specific intensity level.
In the hazard analysis step a series of record from past earth- For each intensity level of earthquake hazard, nonlinear time history
quake events is selected to conduct the incremental dynamic analysis analyses were conducted to estimate the structural response in terms of
(IDA) to account for the uncertainty in the structural response due to a selected engineering demand parameters (EDP). There are a variety
record-to-record variability. In the absence of sufficient ground motion of parameters for local and global EDPs, the most commonly used
data from low-to moderate local earthquake records, two horizontal parameter is the inter-story drift ratio that serves as the performance
components of the 22 far-field earthquake records considered within metric for both structural and non-structural components, while the
FEMA P695 [24] were taken from the Pacific Earthquake Engineering spectral acceleration at the fundamental period of the structure (Sa (T1 ))
Research (PEER) center’s ground motion database. The selected records is used as the intensity measure (IM). The incremental dynamic analysis
are statistically enough in record number, independent of building (IDA) was then conducted to perform nonlinear dynamic analyses of
systems and suitable to use for different locations including low- to the modeled case study building under the previously selected suite of

4
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Table 1
Details of selected ground motions.
ID Name Recording station Year M PGAdir−1 (g) PGAdir−2 (g)
12011 Northridge Beverly Hills - Mulhol 1994 6.7 0.416 0.516
12012 Northridge Canyon Country-WLC 1994 6.7 0.410 0.482
12041 Duzce, Turkey Bolu 1999 7.1 0.728 0.822
12052 Hector Mine Hector 1999 7.1 0.266 0.337
12061 Imperial Valley Delta 1979 6.5 0.238 0.351
12062 Imperial Valley El Centro Array #11 1979 6.5 0.364 0.380
12071 Kobe, Japan Nishi-Akashi 1995 6.9 0.509 0.503
12072 Kobe, Japan Shin-Osaka 1995 6.9 0.243 0.212
12081 Kocaeli, Turkey Duzce 1999 7.5 0.312 0.358
12082 Kocaeli, Turkey Arcelik 1999 7.5 0.219 0.150
12091 Landers Yermo Fire Station 1992 7.3 0.245 0.152
12092 Landers Coolwater 1992 7.3 0.283 0.417
12101 Loma Prieta Capitola 1989 6.9 0.529 0.443
12102 Loma Prieta Gilroy Array #3 1989 6.9 0.555 0.367
12111 Manjil, Iran Abbar 1990 7.4 0.515 0.496
12121 Superstition Hills El Centro Imp. Co. 1987 6.5 0.358 0.258
12122 Superstition Hills Poe Road (temp) 1987 6.5 0.446 0.300
12132 Cape Mendocino Rio Dell Overpass 1992 7.0 0.385 0.549
12141 Chi-Chi, Taiwan CHY101 1999 7.6 0.353 0.440
12142 Chi-Chi, Taiwan TCU045 1999 7.6 0.474 0.512
12151 San Fernando LA - Hollywood Stor 1971 6.6 0.210 0.174
12171 Friuli, Italy Tolmezzo 1976 6.5 0.351 0.315

Fig. 4. Different response spectra of the selected ground motions for 5% damping: (a) Pseudo-acceleration; (b) Displacement; (c) Acceleration–displacement..

ground motion records that were scaled to incrementally increasing IM cyclic test was conducted on two 3D large-scale assemblies of drywall
levels. In IDA, the model was subjected to biaxial ground excitations partitions with suspended ceiling system. Then, three discrete DSs for
using two horizontal components of each 22 selected ground motions. the tested drywall specimens are considered. In the seismic damage
The structural response was restricted to side-sway only collapse with analysis, damage levels experienced by the damaged component can
a drift limit of 5%. The IDA results provided the distribution of EDP show variance, even for the same value of EDP. This is mainly due
(i.e. inter-story drifts) for a given intensity measure (IM) considering to differences in the displacement history of the response and conse-
the uncertainties in record-to-record variability. quently different path of achieving the same value of EDP. In addition,
The probability density function of EDPs, for a given IM is com- uncertainties associated with variability in construction quality, mate-
monly assumed to follow a lognormal distribution. The IDA results of rial strength, failure modes and other factors influencing the extent of
the building were used in Eqs. Eqs. (1a) and (1b) to obtain the mean damage to a component are considered in damage analysis. Note that,
and the logarithmic standard deviation of EDP as a function of IM in a the uncertainties are solely related to the onset of damage as a function
continuous form: of EDP, and is independent of uncertainty associated with the IM or the
𝜇EDP|IM = 𝛼1 (IM)𝛼2 (1a) prediction of EDP [25].

𝜎ln(EDP)|IM = 𝛽1 (IM)𝛽2 + 𝛽3 (1b) 3.4.1. Drywall partition and test setup


where 𝜇EDP|IM is the mean of the EDP for a given IM, 𝜎ln(EDP)|IM is the The experimental test was carried out on the Smart Structural
lognormal standard deviation of EDP given IM, and 𝛼1 , 𝛼2 , 𝛽1 , 𝛽2 and 𝛽3 Laboratory at Swinburne University of Technology. Test setup details
are calculated factors according to the regression analysis. The 𝜇EDP|IM was designed to apply displacement-controlled bidirectional cyclic load
starts from zero, hence, it is fitted with one term power equation. to two 3D large-scale assemblies of drywall partitions with suspended
However, the standard deviation is a measure of dispersion, hence, it ceiling system. The drywall partition walls were made of light-weight
is fitted with two terms power equation. Fig. 5 shows the results of the cold-formed steel framing (typically composed of studs and tracks) and
IDA analysis and the results of regression analysis by the fitted curves sheathed by plasterboards. This assembly is considerably popular in
for the mean and logarithmic standard deviation. the construction of internal walls of buildings in Australia and around
the world. Typically, the installation procedure of the drywall partition
3.4. Seismic damage analysis includes the assembly of the cold-formed steel framing, fixing the
plasterboards to the studs, sealing the joints between the boards using
The damage analysis translates the EDP to the corresponding physi- joint compound, and finally assembly of the suspended ceiling system.
cal damage states (DSs) in drywall partitions. Therefore, a bidirectional The specimens are then fixed to the MAST crosshead from the top and

5
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Fig. 5. The results of IDA and curve fitting to obtain the mean and logarithmic standard deviation of the EDP as a function of IM in continuous form: (a) IDA results for case
study building; (b) one term power function for mean value; (c) two term power function for logarithmic standard deviation.

Fig. 6. Steel frame general layout and the used stud and track profiles.

secured to the strong floor from the base using plywood boards and all gypsum laid between two layers of heavy-duty recycled paper. The
thread rods. suspended ceiling system with more than 600 mm suspension depth
For the first test specimen, named Specimen-1, a full-scale room was hung from the top frame by a number of vertical steel rods. The
with screw only fixing method was constructed under the MAST system details of the suspended ceiling are shown in Fig. 7. The steel rods were
following the general construction details shown in Fig. 1. The test cut to the required length to slide into the hanger clip tabs. The system
specimen approximately occupies a 2.5 m square room surrounded
was supported to the surrounding walls using fixed wall track around
by 120 mm thick and 3.0 m high drywalls and 2.4 m high ceiling.
the perimeter of the ceiling and walls. It is worth noting that, for both
Fig. 6 shows the general layout of the steel frame in drywall and the
cold-formed steel stud and track sections that used in construction of test specimens, ceiling framing connection to the walls was screw fixed
the two test specimens. In Specimen-1, the stud profile 92 mm (BMT on two sides and the other two adjacent sides were able to move freely.
0.55 mm) appropriate to the wall height was selected. Then, standard Fig. 8 shows Specimen-1 including an assembly of drywall partitions
track 92 × 30 mm (BMT 0.50 mm) profile matching the size and gauge and a suspended ceiling system that was built for testing under the
of the stud was chosen as the base and head track. All sections were MAST system. The assembly was finished with joint compound and
selected from standard corrosion resistance products made from G300 50 mm paper tape for both the plasterboard joints and the corners
steel (minimum yield strength of 300 MPa). between the walls. It is noted that in this type of assembly, the studs
During the installation procedure of Specimen-1, the studs were were free to slide through the tracks, while the plasterboards were fixed
inserted into the fixed standard base and head tracks with 600 mm to the studs. Therefore, the assembly of plasterboards and studs can
spacing from centers and 100 mm spacing form each end. It is recom-
slide through the tracks that are fixed to the flooring system (top and
mended to face studs in the same direction to allow easier fastening of
base plywoods).
plasterboard. A twisting action was then used to friction fit the studs
into the tracks. Then, the studs were braced to the plasterboards on For the second test specimen (Specimen-2), the following changes
each face by using coarse threaded drywall screws. The used lining were made in comparison to Specimen-1 according to the recommen-
product was 13 mm thick standard plasterboard made from a core of dations explained before to mitigate the creaking noise (see Fig. 2):

6
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Fig. 7. Configuration of the suspended ceiling with two sides screw fixed and two sides floating for both test specimens.

Fig. 8. Assembly of test setup (Specimen-1) under the MAST system.

– The adhesive fixing method for securing the plasterboard to the protocols that represent the asymmetric lateral drift demands on the
metal framing was used to reduce the number of mechanical structure during sequential ground motions [23]. This is important
fixings. as the damage evolution is path-dependent and is a function of the
– The DH track 92 × 50 mm (BMT 0.55 mm) (Fig. 6) was used history of loading on the component. A few studies have focused on
instead of standard track for the head track. The DH track is a developing these load protocols for structural components in multi-
50 mm deep profile that provide 20 mm clearance to stud and story buildings, e.g. steel columns [26]. However, there is currently a
plasterboard as shown in Fig. 2 joint details. Therefore, the studs lack of such loading protocols for non-structural components. Hence, in
must be cut 20 mm shorter of the supporting structure because the absence of the shake tables, the conventional symmetric cyclic load
this allows for the structure to deflect downwards and there is no protocols are commonly used for seismic assessments. The application
interaction between the sliding stud and the track. of symmetric cyclic displacements with increasing amplitudes may lead
to overestimation of the damage, and therefore, can be considered as a
3.4.2. Loading protocol conservative approach.
Accurate damage assessment of structural and non-structural com- In this study, for the suitability of biaxial testing, the cyclic bidirec-
ponents using experimental tests requires seismic-consistent loading tional cloverleaf path which included the simultaneous function of two

7
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Fig. 9. Cloverleaf cyclic loading protocol.

Fig. 10. Observed residual displacement, crack propagation, and gap opening around the opening of Specimen-1.

sine waves in X and Y perpendicular directions, was considered [27]. and cracks on internal and external corners and a physical gap between
The test specimens were subjected to the cloverleaf cyclic loading the ceiling and walls.
protocol with the ultimate drift of 3.0% over the specimen wall height Pushing the specimens to the next cycle with 1.5% drift, it was
of 3.0 m (see Fig. 9). To collect data on the damages during the full observed that all previously developed damages, including cracking,
range of response, 9 sequential increasing levels were applied with gap openings and delamination were further extended. In Specimen-1,
two cycles at each amplitude. Note that the direction along which the the crack propagation started at 1.0% drift and developed through the
specimen is subjected to the maximum lateral displacement is in 45◦ boards during the next stage of the testing with 1.5% peak drift. How-
diagonal direction. At the end of each loading stage, the MAST system ever, for Specimen-2 the damage was different, and the plasterboards
was returned to the center and the residual physical damages observed and studs started to separate at some locations close to the top due to
from the test specimen were traced and documented.
the failure of the adhesive.
Once this stage was completed for each test specimen, the wall
3.4.3. Experimental results
assembly was carefully inspected to identify residual damages. This
This section describes the observed damages to each of the test
procedure was repeated for the larger intensity levels with the peak
specimens, which progressively included drop of gypsum dust, resid-
ual deformations, detached areas between walls and ceiling, crack drifts of 2.0, 2.5 and 3.0% to further collect the data on propagation
propagation, crushing and separation of the plasterboards, cracking of the damages and possible modes of failure in both specimens.
of joints and buckling of the steel studs and tracks. During the tests, Figs. 10 and 11 present the observed residual displacements, crack
both test specimens experienced no observable damage under the drifts propagations, and the gap openings located at the wall corner close
below 1.0%, which was mainly due to the sliding of the studs through to the door opening of Specimen-1 with screw fixing and Specimen-
the top tracks. Local damage and propagation of crack through the 2 with adhesive fixing, respectively. These figures show the damages
plasterboards were initiated when the small gaps at the intersection corresponding to the last four stages of testing with 1.5, 2.0, 2.5 and
of the wall joints were closed during the stage of testing with 1.0% 3.0% peak drifts. It can be observed that the damages are increased
peak drift. This caused contact and friction between the studs, tracks, as larger drifts are applied to the specimens. Specimen-1 with screw
and plasterboards at the top and the base joints of the walls. After fixing observed minor damages around the door opening and most of
completion of this drift ratio, most of the defects were delamination the damages concentrated around the opening header. In Specimen-2

8
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Fig. 11. Observed residual displacement, adhesive failure, and gap opening around the opening of Specimen-2.

Fig. 12. Observed damages at the top and base of the junction of the walls without openings in Specimen-1.

with the adhesive fixing, the damaged parts were mostly in top corner and Specimen-2 after completion of the tests, respectively. The ob-
and the plasterboards detached form the studs due to adhesive failure. served damages in both specimens include screw pull-out, adhesive fail-
Figs. 12 and 13 present the crack propagation, gap opening between ure, fully detached plasterboard, plasterboard crushing around open-
walls, and crushing of the outer plasterboards observed at the junction ing, large gap at wall junctions, dislodgement of the stud, buckling
of the walls without opening in both tested specimens. These figures of the steel studs and tracks, and delamination of the walls and tapes
show the damages corresponding to the last four stages of testing inside the room. As shown in these figures, the types of damages in
with 1.5, 2.0, 2.5 and 3.0% peak drifts. It should be pointed out that Specimen-1 were more than the Specimen-2 and the difference was
the damage was concentrated at the top corner of Specimen-2 with due to the screw fixing, which caused more resistant and therefore,
adhesive fixing, while the top and bottom corners of Specimen-1 with
transferred more forces to the framing and plasterboards. However, in
screw fixing suffered relatively major damages. Also, the plasterboards
Specimen-2, more severe damages like larger gap and full detachment
were partially dislodged following screw head pull-through failure,
of plasterboard occurred at the end of the test.
observed at the top and the base corners of the external walls of
For Specimen-1 with screw fixing, most of plasterboard edges were
Specimen -1.
The other locations of the tested specimens also suffered damages severely crushed and broken around screws located at the wall inter-
similar to those shown in the previous figures. The multi-axis testing sections, while most of plasterboards in Specimen-2 (adhesive fixing)
of each specimen was conducted for each cycle without the repair were completely detached from the stud and no local damages observed
or changing of the damaged parts after each stage of the testing. except around the door opening. The studs were buckled and dislodged
Therefore, the observed damage of the specimen at each stage was in Specimen-1, and minor damage was found on base tracks. However,
accumulated damage. After finishing the experiments, all the damages the stud dislodgement was the most common type of damage in the
and failures for both damaged specimens were inspected and docu- framing of the Specimen-2 in addition of some minor deformation in
mented. Figs. 14 and 15 show the generated damages to Specimen-1 base tracks similar to Specimen-1. The reason that less buckling damage

9
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Fig. 13. Observed damages at the top and base of the junction of the walls without openings in Specimen-2.

Fig. 14. Damaged parts at the end of the 3.0% drift in Specimen-1.

happened in framing of Specimen-2 was using the DH track that provide insignificant damage to walls that requires minimum repair actions like
20 mm clearance to stud and plasterboard. patching, re-taping, polishing, and painting. The next defined damage
Fig. 16 shows the ceiling residual deformation and displaced ceiling state (DS2 ) is moderate damage illustrated by the local damage of the
channels at the end of the experiments that was similar for both tested plasterboards or studs and tracks, that requires repair or removal and
specimens. As the same ceiling construction detailing was used for replacement of few elements. Third damage state, (DS3 ) is the major
both tested specimens, it can be observed that the type of fixing of damage that consists of extensive damage to plasterboards, stud to
plasterboards in drywalls has less effect in damage assessment of the board connections, and buckling of steel studs and tracks. With this
ceiling. In both specimens, the dislodged ceiling was still maintained level of damage replacement of the part or whole drywall assembly is
its vertical position and no noticeable damage was observed. required [29].
Table 2 summarizes the observed damages and the range of drift
3.4.4. Development of fragility and vulnerability curves threshold values related to each pre-defined damage state and the
Damage assessment is a key component to develop vulnerability required repair activities for each state of the damaged test specimens.
curves for drywall partitions according to the PBEE methodology. The The associated drift to each damage state is similar between Specimen-1
number of defined DSs should be at a range that cover the possible and Specimen-2 according to the same level of repair activities needed
damages at each different level of EDP. In this study, three levels of for each categorized damage in terms of material, labor, and time.
DSs are characterized based on the observed damaged parts of the The tested specimens sustained the story drifts up to 1.0% with a few
tested specimens at each drift level of the applied sequential cyclic damaged areas. In Specimen-1, first damage states (DS1 ) initiated when
loads. The damage states are defined such that they can be linked the plasterboard dust dropped, and consequently, delamination and a
to the repair actions that are needed to recover the partition walls fine visible crack appeared around the opening and the corners at the
to pre-earthquake conditions. Taghavi and Miranda [28] previously testing cycle with 1.0% peak drift. DS1 in Specimen-2 occurred when
defined three DSs for the drywall partitions. According to this study, the a small gap appeared at the wall junctions and plasterboard edges due
first damage state (DS1 ) is indicated by minor damage, which involves to partially failure of the adhesive and movement of the plasterboards.

10
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Table 2
Definition of damage states and story drifts associated with each damage state.
Damage state Description of observed damage Description of observed damage Repair activity to recover the EDP (Max IDR)
(Specimen-1) (Specimen-2) wall
𝐷𝑆 0 No damage No observed damage No observed damage No repair <1.0%
𝐷𝑆 1 Minor damage Dust dropped, delamination and Appearing a small gap in wall junctions patching, re-taping, sanding, 1.0% < 1.5%
appearing a fine visible crack and painting the plasterboards
𝐷𝑆 2 Moderate Plasterboards cracked and crushed, Steel Adhesive failure, Plasterboards repair or replacement of few 1.5% < 2.5
damage studs and tracks separation, Stud detachment, Steel studs rotation and elements
dislodgement separation from track
𝐷𝑆 3 Major damage Severe cracking and crushing, Steel Severe cracking and crushing, Steel replacement of the part or 2.5%<
studs and tracks buckling, Large gap and studs and track residual deformation, entire wall assembly
residual deformation at base corners Large gap at top corners, Full
plasterboard detachment and fallout,

Fig. 15. Damaged parts at the end of the 3.0% drift in Specimen-2.

The next defined damage state, (DS2 ), occurred at the stage with was residual displacement and small separation between the ceiling
1.5% peak drift when the plasterboards were cracked, crushed at the and walls as well as the steel stud and track connections at the wall
intersection of the wall with the other neighboring walls in Specimen- intersections.
1. The observed damages for Specimen-2 in DS2 were including the The third damage state, (DS3 ), was observed at the stage with 2.5%
adhesive failure that caused partial detachment of the plasterboard to 3.0% peak drift when the plasterboards severely cracked, crushed or
from stud, and also, larger gaps were appeared due to stud rotation and detached, and the steel studs buckled in multiple locations leading to
dislodgement at wall corners. At this stage, for both specimens, there extensive damage of the partition system. The observed damages and

11
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Fig. 16. Damaged parts at the end of the 3.0% drift in ceiling.

Fig. 17. The defined damage states (a) fragility curves and (b) vulnerability curves.

the three defined damage states showed that the screw and adhesive state for a given EDP as expressed in Eq. (2).
fixing methods are both in same level of repair activities. Therefore,
⎧1 − F DS0 (no damage)
DS1 (EDP)
the proposed construction method with adhesive fixing to mitigate the ⎪
creaking noise in drywall system, has an acceptable seismic behavior ⎪FDS1 (EDP) − FDS2 (EDP) DS1 (minor damage)
PDS|EDP =⎨ (3)
similar to the screw fixing construction method. ⎪FDS2 (EDP) − FDS3 (EDP) DS2 (moderate damage)
Based on the defined DSs, fragility curves were developed to con- ⎪F DS3 (major damage)
⎩ DS3 (EDP)
sider a range of uncertainties that are solely related to the onset
of physical damage for a given EDP (i.e. drift). Examples of these In this equation, FDS is the fragility curve of EDP and PDS|EDP is
uncertainties include variability in the history of the damage because the probability of exceeding a damage state but not exceeding the
of the specified load protocol, construction quality, material strength subsequent damage state given an EDP. Note that the summation of
the probabilities of being in a given damage state is 1.
and expected failure modes. Accordingly, a lognormal distribution was
The computed probabilities were further integrated with the dis-
assumed for EDP corresponding to three damage states with a mean
tribution of EDPs given IM to compute the probability of being in a
value of 1.0%, 1.5%, and 2.5% drifts and a dispersion (logarithmic
post-event damage state given IM as expressed in Eq. (3).
standard deviation) value of 0.35 adopted based on the recommended
values in ATC-58 [30] for a combination of ductile (steel studs) and ∑ ∞
PDS|IM = PDS|EDP fEDP|IM dEDP (4)
brittle (plasterboards) materials. The lognormal cumulative distribution ∫0
DS
function were then used to define the fragility functions:
In this equation, PDS|IM presents the vulnerability of partition walls
( )
⎛ 𝐸𝐷𝑃 ⎞ defined as the probability of being in a damage state given IM, and
⎜ ln 𝜃 ⎟ fEDP|IM is the probability density function of EDP given IM. Fig. 17(b)
FDS (EDP) = 𝛷 ⎜ ⎟ (2)
⎜ 𝛽 ⎟ presents the vulnerability curves for all damage states that shows the IM
⎝ ⎠ values primary contributing to the exceeding a specific damage state.
where FDS (EDP) is the probability of exceeding a damage state given an As can be observed, there are no damages (DS0 ) up until IM of
EDP, 𝛷 is the standard normal cumulative distribution function (CDF), 0.10 g, which is when the probability of DS1 and DS2 starts to increase.
𝜃 is the median of the fragility function and 𝛽 is the standard deviation. The probability of DS1 peaks at 36% for an IM of 0.16 g, while
Fig. 17(a) presents the developed fragility curves, which is then used the probability of DS2 peaks at 42% for an IM of 0.24 g. Lastly,
to calculate the probability of partition walls being in a certain damage the probability of DS3 rapidly increases beyond an IM of 0.30 g and

12
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

Table 3 peak drift to ultimate value of 3.0%. The experiments showed that
IDR ranges corresponding to different damage states obtained from selected past the assembled drywall systems experienced minor damage up to 1.0%
experimental studies [31].
story drift ratio. The initial damage was observed at drift ratios above
Authors DS1 DS2 DS3
1.0% and increased considerably at the 1.5% drift. The experiment was
IDR range IDR range IDR range
(%) (%) (%) continued and the observed damaged parts at the end of cycle with
Pali et al. [31] 0.06–3.12 0.50–2.84 0.69-4.30
2.5% and 3.0% peak drifts were recorded. The first damage stage, DS1 ,
Restrepo and Bersofsky [32] 0.05–1.00 0.50–1.50 0.50-3.00 initiated when the plasterboard dust dropped and consequently a fine
Retamales et al. [11] 0.10–0.56 0.40–1.84 0.62-2.66 visible crack appeared in the specimen (1.0–1.5% drift). The second
Petrone et al. [12] 0.11–0.78 0.28–2.14 0.66-5.00 damage state, DS2 , was observed when the plasterboards cracked,
Jenkins et al. [15] 0.26–2.64 0.97–2.64 1.86-2.64 crushed and detached (1.5–2.0% drift). The third damage state, DS3 ,
Wang et al. [33] – 0.11–1.09 1.24-2.75
was observed when the plasterboards started to severely crack, crush,
and fully separate from the framing, and also the steel studs and tracks
started to buckle (2.5% < drift). The observed damages showed that
peaks at 77% for an IM of 0.53 g. Note that all the computations are the proposed construction method with adhesive fixing to mitigate the
conditioned on no occurrence of collapse of the building. creaking noise in drywall system, has an acceptable seismic behavior
similar to the screw fixing construction method. The data from the
3.4.5. Comparison with previous experimental work numerical model of the building and the results of the experiments
Several experimental studies have conducted fragility analysis re- were then used to develop vulnerability curves to demonstrate the
porting the IDR ranges corresponding to different damage states of probability of being in a post-event damage state given IMs. The results
drywall partitions [11,12,15,31–33]. In these studies, a number of showed that the onset of damage occurred at around 1.0% story drift
large-scale drywall partitions with different configurations were tested ratio. In addition, the DS1 , DS2 and DS3 reached to the peak at IMs of
to evaluate the seismic response, failure mechanism and fragility of 0.16 g, 0.24 g and 0.53 g, respectively. The demonstrated framework
these components considering various uncertainties. Table 3 shows the can be used to define acceptance criteria for drywall partitions that
range of IDRs corresponding to the three previously defined damage can be used for provision of design, construction and maintenance
states. The comparison highlights that the IDR ranges identified in the guidelines for designers and manufactures. They can also be used to
current research (refer to Table 2) are withing the reported ranges by involve the stakeholders in risk-informed decisions and post-hazard
the previous studies. strategic planning.
In particular, Pali et al. [31] investigated the in-plane seismic
CRediT authorship contribution statement
behavior of drywall partitions with different construction details un-
der a quasi-static cyclic load protocol. In these experiments, different
Hamidreza A. Yazdi: Writing – original draft, Methodology, In-
construction parameters including connection of the wall to the sur-
vestigation, Conceptualization. Javad Hashemi: Writing – review &
rounding elements, stud spacing, type of plasterboard linings and joint
editing, Writing – original draft, Validation, Supervision, Software,
finishing were considered. From these experiments, the specimens with
Resources, Project administration, Methodology, Investigation, Funding
horizontal fixed connections to the surrounding elements (#9 and #10)
acquisition, Data curation, Conceptualization. Emad Gad: Writing –
were closely similar to the room-assembly tested in current study in
review & editing, Supervision, Project administration, Investigation,
terms of the detailing. The obtained results for these specimens show
Funding acquisition, Conceptualization.
median values of 0.79%, 1.18%, and 1.44% for DS1 , DS2 , and DS3 ,
respectively, that are slightly lower than the values obtained in the Declaration of competing interest
current study, which are 1.0%, 1.50%, and 2.50% for DS1 , DS2 , and
DS3 , respectively. The difference could be attributed to the effect of out- The authors declare that they have no known competing finan-
of-plane movements under bidirectional clover leaf cyclic loading that cial interests or personal relationships that could have appeared to
could induce more damage to the drywall as opposed to pure in-plane influence the work reported in this paper.
loading configuration.
The abovementioned results can be further used to assess the risk Acknowledgments
of failure and propose new approaches for minimizing the damage and
improve the resilience of drywall partitions during a seismic event. This The authors acknowledge the contribution of Siniat (Etex Australia
study can also be extended to quantity the losses associated to the sus- Pty Ltd) in conducting the experimental program by providing the
tainability metrics of repair activities such as the cost (economic), time plasterboard and metal framing and the structural engineers Alireza
(social) and energy consumption and carbon emissions (environmental) Bataghva and Pourya Pazandeh for their technical support. Also, the
as described in detail in [34]. authors thank the Smart Structure Laboratory staff for their assistant
with the experiments.
4. Conclusion
References
The primary aim of this study was to perform seismic vulnerability
[1] R. Whitman, S. Hong, J. Reed, Damage Statistic for High-Rise Buildings in
assessment of non-structural drywall partitions, constructed with screw
the Vicinity of the San Fernando Earthquake. Optimum Seismic Protection and
and adhesive fixing method, using the PBEE methodology. In this Building Damage Statistics, Massachusetts Institute of Technology, Cambridge,
framework, a prototype limited-ductility RC building was numerically MA, 1973b.
modeled and calibrated based on previous experimental results. IDA [2] S. Taghavi, M.M. Miranda, Response Assessment of Nonstructural Building
was then conducted on 3D model of the building by applying two Elements, Pacific Earthquake Engineering Research Center., 2003b.
[3] A.S. Tasligedik, S. Pampanin, A. Palermo, Low damage seismic solutions for
components of 22 ground motion records that were scaled to incremen- non-structural drywall partitions, Bull. Earthq. Eng. 13 (4) (2015) 1029–1050.
tally increasing IM levels to collapse. For the damage assessment, two [4] D. Perrone, P. Calvi, R. Nascimbene, et al., Seismic performance of non-structural
large-scale room specimens with screw fixing as a common construction elements during the 2016 central Italy earthquake, Bull. Earthq. Eng. 17 (10)
method, and adhesive fixing with the application of mitigating the (2019) 5655–5677.
[5] R. Whitman, S. Hong, J. Reed, Damage Statistic for High-Rise Buildings in
creaking noise under serviceability wind loads, were constructed and the Vicinity of the San Fernando Earthquake. Optimum Seismic Protection and
experimentally tested. Bidirectional cloverleaf loading protocols were Building Damage Statistics, Massachusetts Institute of Technology, Cambridge,
applied to the specimens in 9 sequential levels with increasing the MA, 1973a.

13
H.A. Yazdi, J. Hashemi and E. Gad Thin-Walled Structures 176 (2022) 109303

[6] H.-J. Kim, D.-H. Shin, Shake table test program of cold-formed steel in-plane [20] H.A. Yazdi, A. Bataghva, P. Pazandeh, et al., Wind-induced creaking noise
partition walls, Structures (2021) 503–517. assessment of a large-scale assembly of drywam system using multi-axis testing,
[7] F.A. Arifin, T.J. Sullivan, G. MacRae, et al., Lessons for loss assessment from the Aust. Struct. Eng. Conf. (2020: Online) 23 (2020) 2–239.
canterbury earthquakes: a 22-storey building, Bull. Earthq. Eng. 19 (5) (2021) [21] M.J. Hashemi, R. Al-Mahaidi, R. Kalfat, et al., Development and validation of
2081–2104. multi-axis substructure testing system for full-scale experiments, Aust. J. Struct.
[8] K.M. McMullin, D. Merrick, Seismic Performance of Gypsum Walls: Experi- Eng. 16 (4) (2015) 302–315.
mental Test Program, Consortium of Universities for Research in Earthquake [22] F. McKenna, Opensees: A framework for earthquake engineering simulation,
Engineering, 2002. Comput. Sci. Eng. 13 (4) (2011) 58–66.
[9] T.H. Lee, M. Kato, T. Matsumiya, et al., Seismic performance evaluation of non- [23] M. Hashemi, H. Tsang, Y. Al-Ogaidi, et al., Collapse assessment of reinforced
structural components: drywall partitions, Earthq. Eng. Struct. Dynam. 36 (3)
concrete building columns through multi-axis hybrid simulation, ACI Struct. J.
(2007) 367–382.
114 (2) (2017) 437–449.
[10] C. Ramirez, E. Miranda, Buildingspecific Loss Estimation Methods and Tools
[24] Applied Technology Council and Federal Emergency Management Agency,
for Simplified Performance-Based Earthquake Engineering Report (171) Ph.D.
Quantification of building seismic performance factors, 2009, Reportno. Report
Dissertation, John A. Blume Earthquake Engineering Center, Stanford University,
Number|, Date. Place Published| : Institution|.
United State, 2009, Ph. D. Dissertation, John A. Blume Earthquake Engineering
Center, Stanford.. [25] H.A. Yazdi, M.J. Hashemi, R. Al-Mahaidi, et al., Multi-axis testing of concrete-
[11] R. Retamales, R. Davies, G. Mosqueda, et al., Experimental seismic fragility of filled steel tube columns forming ductile soft-story in multi-story buildings, J.
cold-formed steel framed gypsum partition walls, J. Struct. Eng. 139 (8) (2013) Construct. Steel Res. 183 (2021) 106736.
1285–1293. [26] Y. Suzuki, D.G. Lignos, Development of collapse-consistent loading protocols for
[12] C. Petrone, G. Magliulo, P. Lopez, et al., Seismic fragility of plasterboard experimental testing of steel columns, Earthq. Eng. Struct. Dynam. 49 (2) (2020)
partitions via in-plane quasi-static tests, Earthq. Eng. Struct. Dynam. 44 (14) 114–131.
(2015) 2589–2606. [27] CEN CEdNT, 340, European Code UNI EN 15129: 2009 anti-seismic devices.
[13] S. Swensen, G. Deierlein, E. Miranda, Behavior of screw and adhesive connections Brussels, 2009.
to gypsum wallboard in wood and cold-formed steel-framed wallettes, J. Struct. [28] S. Taghavi, E. Miranda, Response Assessment of Nonstructural Building Elements,
Eng. 142 (4) (2015) E4015002. Pacific Earthquake Engineering Research Center, 2003a.
[14] J. Ye, X. Wang, H. Jia, et al., Cyclic performance of cold-formed steel shear [29] G. Araya-Letelier, E. Miranda, Novel sliding/frictional connections for improved
walls sheathed with double-layer wallboards on both sides, Thin-Walled Struct. seismic performance of gypsum wallboard partitions, in: 15th World Conference
92 (2015) 146–159. on Earthquake Engineering, 2012.
[15] C. Jenkins, S. Soroushian, E. Rahmanishamsi, et al., Experimental fragility [30] P. Fema, 58-1 (2012) seismic performance assessment of buildings (volume
analysis of cold-formed steel-framed partition wall systems, Thin-Walled Struct. 1-methodology), 2012, Reportno. Report Number|, Date. Place Published| :
103 (2016) 115–127. Institution|.
[16] L. Fiorino, T. Pali, R. Landolfo, Out-of-plane seismic design by testing of non-
[31] T. Pali, V. Macillo, M.T. Terracciano, et al., In-plane quasi-static cyclic tests
structural lightweight steel drywall partition walls, Thin-Walled Struct. 130
of nonstructural lightweight steel drywall partitions for seismic performance
(2018) 213–230.
evaluation, Earthq. Eng. Struct. Dynam. 47 (6) (2018) 1566–1588.
[17] G. Araya-Letelier, E. Miranda, G. Deierlein, Development and testing of a fric-
[32] J.I. Restrepo, A.M. Bersofsky, Performance characteristics of light gage steel stud
tion/sliding connection to improve the seismic performance of gypsum partition
partition walls, Thin-Walled Struct. 49 (2) (2011) 317–324.
walls, Earthq. Spectra 35 (2) (2019) 653–677.
[33] X. Wang, E. Pantoli, T. Hutchinson, et al., Seismic performance of cold-formed
[18] L. Fiorino, B. Bucciero, R. Landolfo, Evaluation of seismic dynamic behaviour of
drywall partitions, façades and ceilings through shake table testing, Eng. Struct. steel wall systems in a full-scale building, J. Struct. Eng. 141 (10) (2015)
180 (2019) 103–123. 04015014.
[19] H. Hasani, K.L. Ryan, Experimental cyclic test of reduced damage detailed [34] M.J. Hashemi, A.Y. Al-Attraqchi, R. Kalfat, et al., Linking seismic resilience
drywall partition walls integrated with a timber rocking wall, J. Earthq. Eng. into sustainability assessment of limited-ductility RC buildings, Eng. Struct. 188
(2021) 1–21. (2019) 121–136.

14

You might also like