System Identification of UCSD-NHERI Shake-Table Test of Two-Story Structure With Cross-Laminated Timber Rocking Walls

System Identification of UCSD-NHERI Shake-Table
Test of Two-Story Structure with Cross-Laminated

Timber Rocking Walls
Ignace Mugabo, Ph.D. 1; Andre R. Barbosa, Ph.D., P.E., A.M.ASCE 2; Arijit Sinha, Ph.D., A.M.ASCE 3;
Christopher Higgins, Ph.D., P.E., M.ASCE 4; Mariapaola Riggio, Ph.D. 5; Shiling Pei, Ph.D., M.ASCE 6;
Downloaded from ascelibrary.org by University Of British Columbia on 11/21/23. Copyright ASCE. For personal use only; all rights reserved.
John W. van de Lindt, Ph.D., F.ASCE 7; and Jeffrey W. Berman, Ph.D., A.M.ASCE 8
Abstract: A full-scale 2-story mass timber building was tested on the University of California San Diego Natural Hazards Engineering
Research Infrastructure (UCSD-NHERI) uniaxial shake table during the period from June 2017 to September 2017. The main objective of the
experimental program was to test the performance of mass timber building designs with different seismic lateral force–resisting systems. The
focus of this study is on a building configuration designed using self-centering post-tensioned cross-laminated timber (CLT) rocking walls
with U-shaped steel flexural plate energy dissipators. The shake-table tests were designed to subject the building to a series of earthquake
ground motions of increasing intensity, ranging from a service-level earthquake to 1.20 times the maximum considered earthquake intensity.
Between each ground motion, low-amplitude white-noise excitations were applied to the building, which responded as a quasilinear system.
In this paper, two output-only operational modal analysis methods are used to estimate the modal parameters (frequency, damping, and mode
shapes) based on acceleration data collected during the white-noise shake-table tests. The correlations of observed damage and repairs
performed during the experimental program with changes in estimated modal features are reported. The modal parameters estimated from
the testing program are also compared with a linear finite-element model that is used to validate the modal identification results and study the
performance of the two system identification methods for CLT rocking structures. DOI: 10.1061/(ASCE)ST.1943-541X.0002938. © 2021
American Society of Civil Engineers.
Introduction sustainability, and aesthetics. As an engineered wood product,

CLT and other mass timber panels have been recognized as the
The emerging use of cross-laminated timber (CLT) in the US and natural choice for buildings with enhanced sustainability goals be-
across the world is mainly attributed to its stakeholder interest, cause they are made from renewable resources and have less
multidirectional load-carrying capacity, speed of construction, embodied energy (Milaj et al. 2017). With the use of CLT contin-
uing to grow in seismic-prone regions, evaluation of the seismic
1 performance of CLT structural systems is of increased importance
Graduate Research Assistant, School of Civil and Construction Engi-
(Pei et al. 2016).
neering, Oregon State Univ., Corvallis, OR 97331; Structural Engineer,
KPFF Consulting Engineers, 111 SW 5th Ave., Suite 2600, Portland, Several experimental programs have been developed in the last
OR 97204. Email: [email protected] 2 decades to study the seismic performance of CLT wall systems. In
2
Associate Professor, School of Civil and Construction Engineering, terms of structural performance assessment and design, the most
Oregon State Univ., Corvallis, OR 97331 (corresponding author). ORCID: relevant research efforts in Europe and North America have focused
https://orcid.org/0000-0003-4547-531X. Email: andre.barbosa@oregonstate on the performance of floor and wall panels and connections for the
.edu lateral force–resisting systems (Ceccotti et al. 2006; Dujic et al.
3
Associate Professor, Dept. of Wood Science and Engineering, Oregon 2010; Popovski et al. 2010; van de Lindt et al. 2010; Ceccotti
State Univ., Corvallis, OR 97331. ORCID: https://orcid.org/0000-0003 et al. 2013; Iqbal et al. 2015; Sustersic et al. 2016; Ganey et al.
-3718-5910. Email: [email protected]
4 2017; Tannert et al. 2018; Zimmerman and McDonnell 2018;
Cecil and Sally Drinkward Professor, School of Civil and Construc-
tion Engineering, Oregon State Univ., Corvallis, OR 97331. Email: chris
Chen and Popovski 2020a, b; Chen et al. 2020). The Construction
[email protected] System Fiemme (SOFIE) project (Ceccotti et al. 2006) was one
5
Associate Professor, Dept. of Wood Science and Engineering, Oregon of the first comprehensive experimental programs performed to
State Univ., Corvallis, OR 97331. Email: mariapaola.riggio@oregonstate evaluate the lateral resistance and behavior of CLT walls. The
.edu SOFIE project comprised element-level monotonic and cyclic tests
6
Associate Professor, Dept. of Civil and Environmental Engineering, of full-size CLT walls, a 3-story shake-table test, and a full-scale
Colorado School of Mines, Golden, CO 80401. Email: [email protected] 7-story CLT structure test performed at the E-Defense shake table
7
Professor, Dept. of Civil and Environmental Engineering, Colorado in Japan.
State Univ., Fort Collins, CO 80523. Email: [email protected] The project included the design and shake-table testing of
8
Professor, Dept. of Civil and Environmental Engineering, Univ. of
3-story and 7-story CLT shake-table test specimens. For the 3-story
Washington, Seattle, WA 98195. Email: [email protected]
Note. This manuscript was submitted on May 18, 2020; approved on CLT structure (Ceccotti and Follesa 2006), the unidirectional
October 8, 2020; published online on January 23, 2021. Discussion per- [one-dimensional (1D)] shaking tests included subjecting the
iod open until June 23, 2021; separate discussions must be submitted for structure to 26 earthquake simulated ground motions, which pro-
individual papers. This paper is part of the Journal of Structural Engi- gressively increased the peak ground accelerations (PGA) from
neering, © ASCE, ISSN 0733-9445. 0.15g to 1.20g. Natural frequency reductions due to the more severe
© ASCE 04021018-1 J. Struct. Eng.
J. Struct. Eng., 2021, 147(4): 04021018

earthquakes (0.5g–1.2g) were all 12% or less, where by design, and existed that characterized natural frequencies and correlated changes
even under such intense shaking, the structure exhibited repairable in natural frequency with observed damage for CLT self-centering
damage. rocking structural systems.
For the 7-story shake-table test (Ceccotti et al. 2013), the struc- The main objective of this paper is to present findings on system
ture was subjected to a total of 10 earthquake motions, including identification of the structure tested during Phase 1 of the UCSD-
(1) a set of five unidirectional shake-table motions, and (2) five NHERI shake-table test, which corresponds to the rocking, self-
tridirectional shake-table motions. Results from the application centering CLT structural design solution detailed by Pei et al.
of the shake-table motions indicated that the X- and Y-directions, (2019a). The shake-table tests were designed to progressively sub-
fundamental frequencies of the 7-story structure exhibited a 24% ject the structure to a series of historical earthquake ground motions
and 17% reduction from the initial values, respectively. Overall, the of increasing intensity, ranging from service-level earthquake inten-
7-story CLT structure produced ductile and self-centering behavior sity to 1.20 times the maximum considered earthquake intensity.
marked with fastener ductile bending failure and embedment fail- The acceleration data used to estimate the modal features were
ure of the connections at the wall ends. To date, no journal papers due to the low-amplitude white-noise shake-table motion tests per-
were found in the literature that focused on studying modal features formed immediately before and immediately after each of the earth-
and damage propagation of self-centering CLT structures under quake shaking tests. In addition, these white-noise shake-table
shake-table experiments. motions were also applied before and after each of the interventions
Self-centering rocking wall systems exhibit rocking deformation performed.
behavior under lateral loading. At the same time, post-tensioned In this paper, two output-only methods for structural identifica-
tendons are used to recenter the walls to their original position. tion are used, which were observed to perform well in previous
The conceptual basis behind rocking wall systems is the use of a tests performed on different structural systems (Moaveni et al.
recentering mechanism, e.g., post-tensioned tendons, coupled with 2014; Belleri et al. 2014). The dynamic characterization methods
external energy dissipators (e.g., mild steel bars) that are designed are used here to systematically document the initial modal features
to dissipate seismic energy. The rocking wall concept originated and progressive changes to the modal parameters as damage pro-
from research in precast concrete (Priestley 1991; Holden et al. gressed and repairs were performed during the testing program.
2003; Restrepo and Rahman 2007; Belleri et al. 2014; Kurama The modal parameters estimated from the testing program are also
et al. 2018) and was successfully adapted to timber systems over compared with a linear finite-element model that is used to validate
the last 15 years (e.g., Granello et al. 2020). Post-tensioned timber the modal identification results and study the performance of the
systems have been detailed in a series of subassembly experiments two system identification methods for CLT rocking structures.
performed on beam-to-column, column-to-foundation, and wall-to- Lastly, the modal results provide benchmark data that can be used
foundation energy-dissipation solutions using laminated-veneer in future designs of CLT rocking structures, including the estima-
lumber (LVL) members (Palermo et al. 2005, 2006a, b; Smith et al. tion of initial and secant stiffnesses that can be used for design-
2007) and glued-laminated timber (glulam) members (e.g., Granello basis earthquake (DBE) intensities.
et al. 2019). Out of the energy-dissipation methods tested with LVL
shear walls by Palermo et al. (2005, 2006a, b) and Smith et al.
(2007), the U-shaped flexural plates (UFP) were observed to be the Materials and Methods
most effective (Smith et al. 2008).
Baird et al. (2014) derived formulas for the yielding force, initial
Test Specimen Description
stiffness, and hardening parameter using previous analytical deri-
vations by Kelly et al. (1972), based on experimental tests and A 2-story structure with CLT rocking walls was tested at the
finite-element modeling. Kramer et al. (2015) evaluated the tension UCSD-NHERI shake-table facility in 2017. Fig. 1 shows the struc-
and cyclic performance of an external energy-dissipation connector ture on the shake table following erection. The 2-story CLT build-
to be used on self-centering rocking CLT walls. The results indi- ing had a rectangular planar geometry of 17,680 × 6,100 mm. The
cated no damage on the CLT members, low variability, and predict- total height of the structure was 6,790 mm.
able behavior of the energy dissipators. A series of other studies Glued-laminated timber (GLT) beams and columns supported
focused on the seismic performance of self-centering CLT walls gravity loads collected from the CLT floor and roof. The columns
with UFP devices (Akbas et al. 2017; Ganey et al. 2017). were attached to the steel base with slotted pin connections to allow
In 2017, a series of tests were performed in the University of for compatibility of movements due to seismic loading without
California San Diego Natural Hazards Engineering Research Infra- introducing additional lateral loading on the columns, which mainly
structure (UCSD-NHERI) shake table for mass timber structures. carry gravity loads. Two columns sizes were used: (1) GLT 273 ×
Testing performed provided benchmark data on the performance of 190 mm, located along Gridlines A2 and A3, and (2) GLT 222 ×
three lateral resisting systems. Thus, the testing was divided into 190 mm, located along Gridlines A1 and A4 (Fig. 2).
three phases: (1) a self-centering rocking wall design (Pei et al. Two different floor systems were used on the floor and roof lev-
2019a), (2) a non-post-tensioned rocking wall design (Blomgren els. On the floor level, primary beams had cross-section dimensions
et al. 2019), and (3) a platform-type CLT wall system (van de of 495 × 222 mm and spanned in the north–south (NS) direction,
Lindt et al. 2019). The three studies cited summarize the experi- secondary beams had cross-section dimensions of 495 × 171 mm
mental testing program, including design, as well as the main re- and spanned in the east–west (EW) direction, and the floor con-
sults obtained for the lateral load–resisting systems. However, a sisted of three-ply CLT spanning in the NS direction [Fig. 2(b)].
detailed analysis of system identification results had not yet been The roof floor system consisted of beams with cross-section dimen-
performed and is developed in this study. sions of 457 × 222 mm (lateral bays) and 381 × 222 mm (central
Based on the literature review and existing knowledge, only a bay) spanning in the NS direction; no beams were detailed in the
limited number of tests on rocking timber systems have been per- EW direction. The roof-level floor spanned in the EW direction
formed on shake tables. No self-centering CLT systems have been and was constructed using a five-ply CLT-concrete composite
tested on a shake table at the time the shake-table testing reported system (Higgins et al. 2017), in which the reinforced concrete
here was performed. Therefore, no full-scale, dynamic test data was 57 mm thick. Two glulam grades were specified for the
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 1. Two-story CLT rocking wall structure.
Fig. 2. CLT self-centering rocking wall test specimen: (a) roof plan view; (b) floor plan view; (c) EW elevation view; (d) foundation beam and
shake-table plan view; and (e) NS elevation view. Accelerometers are marked as square boxes with arrows indicating the direction of recording.
Dimensions are presented in millimeters.
J. Struct. Eng., 2021, 147(4): 04021018

gravity-resisting system of the structure: (1) grade L2 for the glu-
lam columns, and (2) 24F-V8 for the glulam beams (American
Wood Council 2015). All CLT floor members were graded V1
(ANSI/APA 2018) and made of Douglas fir (DF), Pseudotsuga
menziesii.
The connections and elements of the CLT walls and surrounding
members were designed to allow rocking and prevent sliding of
walls at the base. Such connections and elements included (1) post-
tensioning steel bars along the full-height of the CLT walls
[Fig. 3(a)], (2) steel tongue plates between diaphragms and the (a)
rocking walls [Fig. 3(b)], and (3) U-flexural plates (UFP) installed
between adjacent CLT wall panels and distributed along the height
of the walls [Fig. 3(c)]. The structure had two walls that consisted
of two five-ply E1 panels that were 1,520 mm long connected
structurally using five UFP devices. The lowest UFP was located
1,980 mm from the base of the wall, and the spacing between the
five UFP devices was 960, 1,140, 1,140, and 990 mm measured
from the UFP that was closest to the foundation. Each CLT panel
was post-tensioned using four 19.05-mm-diameter high-strength
rods [ASTM A449 Type 1 (ASTM 2020)] with a yield stress of
634 MPa.
An initial tension force of 53.3 kN was applied to each of the
post-tensioning bars at the top of the wall. The walls were con- (b)
nected to the floors using a vertical slot tongue plate shear key,
which coupled the horizontal in-plane displacements of the floor
diaphragm and wall. Additional steel angles with Teflon polytetra-
fluorethylene sliding faces were used to provide out-of-plane
displacement restraint to the wall from the floor. At the floor level,
the tongue plate had a cross section of 44.5 × 76 mm, and at the
roof level, the tongue plate had a cross section of 22 × 76 mm.
This connection allowed only horizontal forces to be transferred
between the floor diaphragms, thereby allowing walls to rock
and lift without providing additional vertical restraints to the wall
movement.
In order to stabilize the structure in the direction perpendicular
(c)
to shaking and avoid excessive torsional movements, cross-bracing
steel bars were added to the structure on the east and west faces
between Grids A2 and A3 [Figs. 1 and 2(c)]. The cross-braces were
19-mm fully threaded rods tying the roof and second floor to the
shake table’s floor beams. Additional construction drawings are
available for download in DesignSafe-CI (Pei et al. 2019b). The
uplift of the foundation was prevented by anchoring post-tensioned
rods from the top of the steel base beam to the bottom of the shake-
table platform.
Experimental Setup, Testing Sequence, and

Instrumentation
At the time of testing, the shake table was a unidirectional shaking (d)
platform with a total movement range of 750 mm, and a peak Fig. 3. CLT rocking wall test specimen: (a) CLT wall section showing a
acceleration capacity of 1.2g when loaded at a 400-t payload. UFP connection and four post-tensioning rods; (b) diaphragm-to-
The shake table was capable of producing shaking in a frequency rocking wall tongue plate connection; (c) close-up of a UFP connec-
range up to 30 Hz. The CLT structure was subjected to 14 input tion; and (d) close-up of a column base connection. Bolts went through
acceleration time series, based on four historical ground motion slotted holes in the column. Dimensions and other detail for this con-
records: (1) 1994 Northridge earthquake, Canoga Park, (2) 1989 nection have been reported in construction structural drawings publicly
Loma Prieta earthquake, Capitola, (3) 1987 Superstition Hills available from Pei et al. (2019b).
earthquake, Poe Road, and (4) 1979 Imperial Valley earthquake,
Delta. The ground motions were scaled to three seismic hazard
levels considered in the building design, including the service-level
earthquake (SLE), DBE, and maximum considered earthquake been provided by Pei et al. (2019a). Fig. 4 summarizes the test se-
(MCE). The SLE, DBE, and MCE intensities considered corre- quence of the input ground motions. The recorded actual shake-
spond to levels used in the design of the prototype building, which table motions were used throughout this study. The measured
was designed for a site in Oakland, California, and soil site Class B. PGAs experienced by the tested structure are shown in Fig. 4 over
Additional details on the design of the prototype structure have the different input motions.
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 4. Ground-motion testing sequence.
To understand the effects of damage (visible or not) to the struc- when there is post-tensioning (PT) force loss over 4.5 kN
ture from the increased shaking intensity, white-noise (WN) tests (1 kip)].
were carried out before and after each ground-motion test. In • Intervention 4: The intervention occurred between shake-table
addition to the WN tests after ground-motion events, additional Tests 10 and 11, and consisted of cross-bracing retightening.
WN tests followed six Interventions conducted included pre- • Intervention 5: The intervention occurred between shake-table
emptive repair or retrofitting measures, which were conducted over Tests 12 and 13 and consisted of cross-bracing retightening.
the test sequence. Thus, a total of 21 WN tests were conducted • Intervention 6: The intervention occurred between ground-
during the testing sequence. The WN signals consisted of a root- motion Tests 13 and 14, whereby the post-tensioning bars were
mean squared (RMS) acceleration of 0.03g for a duration of ap- again retensioned to their intended initial tension and cross-
proximately 120 s, except for the first three WN tests, which were bracing retightened.
conducted for 240 s. Data from these WN tests are available for Microelectromechanical system (MEMS) accelerometers with a
download in DesignSafe-CI (Pei et al. 2019b). total acceleration amplitude range of 5g (g ¼ 9.81 m=s2 ) were
Six preemptive interventions were undertaken over the test se- used to capture the structural dynamic response. The accelerome-
quence to make minor repairs or adjustments to the test structure ters were placed at seven locations on the underside of the floor and
that allow it to perform as intended in the structural design. In the roof levels, measuring the NS and EW accelerations [Figs. 2(a
interventions, the following actions were carried out: and b)]. In Figs. 2(a and b), the floor-level and roof accelerometer
• Intervention 1: The intervention taken between shake-table locations are labeled 1 through 14. In addition to measuring NS and
Tests 1 and 2 consisted of modification to the tongue plate con-
nections (a Teflon polytetrafluorethylene interface was added)
to reduce friction between the CLT wall slot and the tongue
plates connected to the floor diaphragm. The Teflon polytetra-
fluorethylene plates were effective in reducing the friction.
However, the design of the fixture of the Teflon polytetrafluor-
ethylene plate could be improved in future designs because they
tended to start to roll out of their original locations, which re-
quired that modifications and replacements had to be performed
during several of the interventions. Fig. 5(b) shows a damaged
Teflon polytetrafluorethylene plate (foreground) after being re-
moved and replaced by with new Teflon polytetrafluorethylene
plates (background).
• Intervention 2: This intervention took place between shake-table
Tests 5 and 6, and consisted of stiffening (bracing laterally) the
base beam after notable base beam deformations were observed
[Fig. 5(a)]. Fig. 5. (a) Base beam damage showing as downward deformation at
• Intervention 3: The intervention occurred between shake-table near the middle of the top flange; and (b) replacement of the Teflon
Tests 8 and 9 whereby the post-tensioning bars were retensioned polytetrafluorethylene plates during Intervention I. (Adapted from
Wichman 2018.)
to their intended initial tension [the retension was triggered
J. Struct. Eng., 2021, 147(4): 04021018

EW accelerations, the accelerometers placed in the corners of the which was used to implement the SSI and the EFDD modal analysis
floor plans (at locations labeled 1, 2, 6, 7, 8, 9, 13, and 14) also methods. Of the several SSI techniques available in ARTeMIS
captured vertical accelerations. One accelerometer was placed at Modal, the SSI-extended unweighted principal component (SSI-
the top of the CLT-concrete roof diaphragm to compare accelera- UPCX) (Mellinger et al. 2016) was used for all the SSI-obtained
tion values with the accelerometers at Location 11 placed on the modal parameters. Because output-only OMA methods were used,
underside of the roof CLT members. Ten additional accelerometers the input excitation was not included in the analysis, although
were placed on the CLT walls. For the analysis conducted in the all input data are publicly available for download from Pei et al.
current study, only the14 accelerometers located on the CLT floor (2019b).
and roof panels are used, corresponding to a total of 36 channels
(18 channels on the floor level and 18 channels on the roof level). Data Postprocessing
Additional instrumentation drawings are available in Design-Safe The WN acceleration data were collected at 240 Hz during the
(Pei et al. 2019b). shake-table tests. However, the modal results were obtained by fo-
cusing on a frequency range that could be used to identify the first

six significant horizontal modes of vibration, including the first and
Operational Modal Analysis and Damage Assessment second modes in the NS, EW, and torsional directions. For consis-
Procedure tency, data were organized into three sensor groups for analysis.
Two operational modal analysis (OMA) methods were used for First, Group 1 considered all floor level and roof channels, whereas
analyzing the WN acceleration data, namely enhanced frequency- Group 2 and Group 3 considered channels measuring accelerations
domain decomposition (EFDD) and the stochastic subspace iden- in EW and NS directions only, respectively. Group 3, for example,
tification (SSI). was used to improve identification of the NS modal features be-
The EFDD method is derived from the frequency-domain cause the NS direction is perpendicular to the shake-table shaking
decomposition, a method in which the power spectral densities direction and therefore saw little to no shake-table excitation. For
of a linear time-invariant system are performed (Brincker et al. the modal feature extraction using both EFDD and SSI methods,
2001). The EFFD method decomposes the frequency content of the acceleration data were downsampled using a low-pass filter
the output data into independent single degrees of freedom acting with an upper-frequency cutoff at 80% of the downsampled fre-
along the frequency range of interest. A key assumption in the quency. The acceleration data were downsampled to 12, 6, and
EFDD method is that the input data are of a white-noise type, 3 Hz to focus on different frequency ranges of interest based on
meaning that their power spectral densities are constant. Secondly, a preliminary inspection of power spectral density results, which
the damping model used in the EFDD method follows the viscous revealed several frequency peaks of interest below 12 Hz. After
damping model for linear-elastic systems. In the EFDD method, the a preliminary analysis of the data, a model order of n ¼ 40 was
natural frequencies and damping ratios are obtained by transform- consistently used to extract modes using the SSI method.
ing the single-degree-of-freedom frequency data into the time do-
Finite-Element Modeling
main and finally using the crossing time method and logarithmic
decay approach for the natural frequencies and damping ratios, Initial-State FE Model. A linear-elastic finite-element (FE) model
respectively. of the structure was developed for the initial state of the structure
The SSI method relies on the use of a state space model, effec- before it was subjected to any strong shaking using general-purpose
tively turning the second-order differential equation of motion into finite-element structural analysis software SAP2000 version 19
a first-order differential equation (Brincker and Andersen 2006). (CSI 2017). The FE model included seven different categories
The following assumptions of the SSI method should be noted: of frame and shell elements representing: (1) glulam columns,
(1) the system is a linear time-invariant type, (2) the system’s re- (2) glulam beams, (3) three-ply CLT diaphragm for the floor level,
sponse can be represented by a Gaussian stochastic process, and (4) CLT-concrete composite diaphragm, (5) five-ply CLT shear
(3) a viscous damping model is adopted. Following the formulation wall, (6) post-tensioning steel rods, and (7) experimental features
of the state space model, the SSI method uses discretization to such as steel base beams and cross-bracing. A geometric represen-
create time-lapsed series of the system output time-domain data. tation of the FE model is shown in Fig. 6.
The time-lapsed series are used to create a model of the system by Due to the orthotropic behavior of CLT, all CLT members were
computing covariances between the future subsets of the output modeled as orthotropic thin-shell elements. Orthotropic shell mod-
data to the prior subsets of the output data. The correlation phase eling entailed estimating shear and elastic moduli for in-plane and
allows for the system matrix to be obtained by regression, and out-of-plane directions. The in-plane moduli of elasticity (E1 and
subsequently allowing for the modal parameters to be obtained E2 ) along the strong and weak axes of a CLT panel were assigned
by eigenvalue analysis. based on E0 compression tests listed by Barbosa et al. (2019) on
Extended explanations of the EFDD and the SSI methods have five-ply DF (Pseudotsuga menziesii) CLT. Barbosa et al. (2019)
been given by Brincker et al. (2001) and Brincker and Andersen obtained the strong-axis modulus of elasticity E1 of five-ply DF
(2006), respectively. The uncertainty associated with the applica- CLT from testing of five specimens and resulted in a mean value
tion of these methods directly to the shake table used in these ex- of 9.0 GPa with a coefficient of variation equal to 3.1%. To derive
periments was summarized by Moaveni et al. (2014). The authors the weak-axis modulus of elasticity (E2 ), the composite theory
are aware that such differences and identification errors in damping (k-method) defined by Blass and Fellmoser (2004) was adopted.
ratio values have been observed in other studies using OMA meth- The k-method in-plane moduli of elasticity relationships are
ods (Magalhães et al. 2010; Moaveni et al. 2014). For example,
multiple factors influence the identifiability of the damping ratios,
including amplitude of the signal, spatial density of the measure- E1 ¼ E0 · k3 ð1aÞ
ments, length of the signals, and model order selected (Moaveni
et al. 2014).
E a − am−4 þ · · · a1
The methods used are available in ARTeMIS Modal version 5.2 k3 ¼ 1 − 1 − 90 · m−2 ð1bÞ
(Structural Vibration Solutions 2017) modal identification software, E0 am
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 6. Geometric representation of the FE model of the structure: (a) isometric view; (b) north–south elevation; and (c) east–west elevation.
E2 ¼ E0 · k 4 ð1cÞ CLT Handbook (Gagnon and Popovski 2011). Table 1 summarizes

the material properties used in the FE model for various elements.
The tongue plates effectively acted as the boundary elements for
E90 E90 a − am−4 þ · · · a1
k4 ¼ þ 1− · m−2 ð1dÞ the CLT floors, transferring load to the shear walls. The tongue
E0 E0 am
plates were idealized as a rigid element with rotational springs
where k3 and k4 = composite factors for loading parallel and at the diaphragm end with stiffness given by
perpendicular to the strong axis of the CLT panel, respectively;
am = total depth of the CLT panel; a1 = thickness of the middle 3Es I TP
K TP ¼ ð3Þ
lamella; and am−2 and am−4 = thickness of the total CLT depth l3TP
subtracted by the thickness of the two and four outer lamellae,
respectively. The modulus of elasticity of a single lamella in the where Es , I TP , and lTP = tongue plate material modulus of elasticity,
direction perpendicular to grain, E90 , and its ratio to the strong axis second moment of inertia about the direction perpendicular to the
lamellar modulus of elasticity, E0 , E90 =E0 , has been estimated to be diaphragm floor, and the assumed length between the point of
0.05 and 0.068 for the radial and the tangential axes for DF lumber, fixity of the tongue plates and wall connection, respectively. The
respectively. The CLT Handbook (Gagnon and Popovski 2011) length of the tongue plate, shown in Fig. 3(b), was 912 mm, and the
recommends an E90 =E0 ratio of 1/30, which was assumed for the tongue plate was attached to a steel wing plate using several hex
computation of E2 . The recommended values from the CLT Hand- bolts. The wing plate, in turn, was attached to the CLT floor using
book were used because it is commonly adopted in the design the 45° inclined screws shown in the figure. Thus, a simplified
procedures used in North America. model as the one shown in Eq. 3 requires the assumption of the
The in-plane shear modulus G12 was computed using the effective length lTP , which is a function of the assumed fixity for
method developed by Bogensperger et al. (2010), given by the tongue plate and a single point of load transfer between the wall
and the tongue plate. The length lTP was assumed to be 304 mm,
G0;l;mean
G12 ¼ ð2aÞ
1 þ 6 · αT · ða1 =wl Þ2
Table 1. Summary of material properties used in this study
where
q Mechanical properties
a1
αT ¼ p · ð2bÞ Orthotropic Isotropic
wl
Material/members E1 (GPa) E2 (GPa) G12 (GPa) E (GPa)
and G0;l;mean , a1 , and wl = shear modulus, average thickness, and CLT, 3-ply 9.94 5.21 0.59 —
average board width of the CLT lamellae. The parameters p and q CLT, 5-ply 9.00 6.16 0.64 —
were determined by reported values from Dröscher and Brandner Glulam beamsa — — — 12.41
(2013), and calibration from shear modulus experimental testing by Glulam columnsa — — — 11.72
Barbosa et al. (2019). In determining G12 , G0;l;mean was set equal to Post-tensioning bars — — — 196.5
0.0625 × E0 according to estimates provided in the CLT Handbook All other steel members — — — 200
(Gagnon and Popovski 2011). The out-of-plane stiffness properties a
For the purpose of simplifying the FE model, glulam beams and columns
were computed using the shear analogy method, as provided in the were modeled as isotropic materials.
J. Struct. Eng., 2021, 147(4): 04021018

which is equivalent to the distance from the first hex bolt (closest to Damaged-State FE Model. Several sources of damage were iden-
the wall) and the center of the wall. However, the results of some tified to have occurred during the testing sequence (Pei et al.
modal frequencies of the FE model are sensitive to this assumption, 2019a). Thus, the initial FE model described in the previous section
as discussed subsequently. Furthermore, the vertical stiffness was was adjusted to incorporate the effects of damage by modifying the
set to zero to capture the design features, which decoupled the hori- stiffness of various elements. The most sources of damage are sum-
zontal motion of the diaphragm and walls from the vertical motion marized here in the interest of completeness and readability of the
of the walls. paper. The effects of damage were considered in modeling of
Similar to the tongue plate connections, the UFPs were modeled (1) base beams, with minor residual deformations of the base beams
as spring elements. As the walls rocked, the UFPs coupled the following the first few shake-table tests, requiring the base beam to
movement of adjacent walls. Following recommendations from be reinforced with stiffeners, as described in Intervention 2, and
Baird et al. (2014), the initial stiffness of the UFPs is given by (2) CLT wall corners, with minor crushing of the CLT wall panel
corners at the base.
16Es bu tu 3
In addition to the observed structural damage, other sources af-

K UFP ¼ ð4Þ
27π Du fecting the frequencies of the structure were identified and adjusted.
In the direction perpendicular to the shaking, the cross-braces ex-
where Es , bu , tu , and Du = modulus of elasticity, width, thickness, hibited loosening during shaking. They were retightened at several
and diameter of the UFP steel plates, respectively. The UFPs in the stages of the test sequence (as described previously in terms of the
tested structure were manufactured with a bu value of 114 mm, tu various interventions). Although minor damage was also observed
value of 10 mm, and Du value of 92 mm. The Young’s modulus of in the column base and some beam-to-column connections, these
steel Es was set to 200 GPa. were assumed not to have a large impact on the initial stiffness of
The last set of springs discretized the behavior of the base of the the structure because most of these connections were designed with
CLT wall panels in compression during rocking. Following recom- slotted connections to allow for displacement compatibility be-
mendations from Akbas et al. (2017), the springs at the interface tween the LFRS and the gravity system (Pei et al. 2019a). Thus,
between the bottom of the CLT wall and the steel base beam are these were neglected for the low levels of excitation studied as part
given by of the WN testing.
Additionally, a few CLT wall post-tensioned rods exhibited
E 1 · Ac some loss of tension during the testing sequence and were sub-
K ew ¼ ð5Þ
hc sequently retensioned to the specified design tension force, having
eventually yielded and losing 30% of the initial tensioning force
where Ac = cross-sectional area of the compression block; and hc = after 1.20 × MCE test (Pei et al. 2019a). However, when the struc-
compression block height. The compression block cross-sectional ture was subjected to small deformations under the white-noise ex-
area Ac was computed as the product of the thickness of the five-ply citations, the post-tensioning rods provide a small contribution to
CLT panel, am , and the width assumed to be 3/8 of the length of the the stiffness of the structure, as observed from a preliminary sen-
wall, Lw , at the effective elastic limit, per recommendations from sitivity study that was part of this research but not shown here in the
Akbas et al. (2017). The compression block height hc was assumed interest of brevity. In other words, even though the steel rods were
to be equal to 0.5 × am , a value that was approximately equivalent observed to yield following the 1.2 × MCE motions, up to that
to the height of the observed CLT wall damage at the compression point, the magnitudes of the exhibited losses in post-tensioning
block locations (Pei et al. 2019a). The base of CLT wall spring el- force did not alter the white-noise-induced small vibration building
ements and the post-tensioning bars acted as the boundary elements frequencies significantly.
for the CLT walls.
To capture the modal response of the shake-table structure in the
NS direction as well as in torsion, the steel rod cross-braces were Results and Discussions
also modeled. The steel rods had a diameter of 19 mm, consisting
of A36 steel grade. Thus, the rods were assigned a Young’s modu-
Experimental Test Results
lus of 200 GPa.
The base beam was observed to have affected the behavior of
the structure (Pei et al. 2019a), and therefore the base beam was System Identification of the Structure
also included as part of the FE model. The base beam was modeled The frequency peaks of the structure were initially evaluated using
as steel shell elements of the same dimensions used in the physical power spectral density (PSD) with a focus on a frequency range
experiment (the “Test Specimen Description” section gives the below 6 Hz and without data filtering. Fig. 7 displays the time
measurements of the base beam steel plate members). The boun- series and PSDs of Channels 14E and 14N located at the southeast
dary conditions of the base beams and gravity columns in the elastic corner of the roof level, which measured accelerations in the EW
model were all idealized as pinned supports to the base to replicate and NS directions, respectively. The figure shows time signals and
the low-amplitude boundary conditions of the experimental study. PSDs for WN tests at the start (WN 01) and the end (WN 21) of the
The post-tensioning bars were modeled as 19-mm-diameter ten- shake-table testing sequence. It can be observed that the Channel
dons with a modulus of elasticity of 196.5 GPa. 14E acceleration time series shows higher amplitudes compared
The estimated mass of the structure included the self-weight of with the Channel 14N acceleration time series, which is mainly
all structural members in addition to the added seismic weights because the shake-table shaking was in the EW direction only.
(steel plates) that were used to represent the weight of the super- The PSDs show a distinct frequency peak at 1.40 Hz for Channel
imposed dead loads. All of the masses were realistically distributed 14E data. Results from Channel 14N display a less distinctive peak
according to building element locations and added seismic weights at 1.55 Hz. From this analysis alone, the frequency peak observed
at the appropriate locations to match the experimental test condi- in the Channel 14N data indicates that a natural frequency from
tions. The added seismic weights were 156 kN for the roof and either a torsional or NS translational mode of vibration is being
313 kN for the floor level (Pei et al. 2019a). produced despite the shake-table shaking only in the EW direction.
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 7. White-noise time signals and power spectral densities: (a) time signals of Channels 14E and 14N for white-noise Test 1 at the start of the
shake-table test sequence and white-noise Test 21 at the end of the testing sequence; and (b) power spectral densities of Channels 14E and 14N at
for white-noise Test 1 and Test 21 of the shake-table test sequence.
Other peaks are also observed at approximately 4.3 Hz, as dis- modes of vibration, and finally, Mode shapes 3 and 6 are the first
cussed in detail subsequently. and second modes in the NS direction, respectively.
Using the SSI and EFDD methods, six modes of vibration are
identified using the first WN trial (WN 01) data, which was con- Modal Features during the Shake-Table Testing Sequence
ducted at the beginning of the shake-table testing sequence. Changes in the natural frequencies and damping ratios for the iden-
Table 2 summarizes the identified natural frequencies and tified modes were evaluated at the end of each ground-motion
damping ratios. Small variations between the natural frequencies shake-table test as well as after interventions were made. Fig. 9
identified using EFDD and SSI methods are observed, the largest shows the changes in identified natural frequencies from each of
of which amounts to a 3.7% difference for Mode 1. The damping the 21 WN tests performed, as listed in Fig. 4. Results in Fig. 9
ratios identified using the EFDD and SSI methods vary consider- show that the natural frequency of Mode 1, f̄1 , decreased by ap-
ably, whereby the identified damping ratios for Mode 1 are 3.7% proximately 27% at the end of the testing sequence, with other
and 7.4% using the EFDD and SSI methods, respectively. Simi- modes showing smaller reductions and even slight increases, as
larly, the damping ratios identified for Mode 4 are approximately shown in Modes 3, 4, and 6. Before Intervention 1, a decrease
8.5% and 16.4% using the EFDD and SSI methods, respectively. of 13% in f̄1 was observed following the first ground-motion test
Such variations in damping ratios have been documented and jus- (Loma Prieta SLE intensity). This initial reduction is observed in
tified before in other shake-table studies (e.g., Moaveni et al. 2014) f̄2 , f̄ 4 , and f̄5 , although to a lesser extent. The initial reduction was
and ambient vibrations studies (e.g., Mugabo et al. 2019) when not apparent for Modes 3 and 6, which were perpendicular to the
different structural identification methods are used. As can be direction of shaking. Interestingly, f̄ 3 and f̄ 6 exhibited a slight in-
observed from Fig. 8, vibration Mode shapes 1 and 4 are the first crease. The most considerable reduction of f̄ 5 amounted to a 9%
reduction following Intervention 1 (WN Test 3), during which
and second translational modes of vibration in the EW direction,
modifications were made to tongue plates to reduce friction-
Mode shapes 2 and 5 correspond to the first and second torsional
induced transfer of vertical forces.
Moderate reductions in f̄1 are observed after the remainder of
the SLE round motions, as shown in the graph starting with WN 02
Table 2. Natural frequencies and damping ratios for the white-noise test at through WN 06.
the start of the shake-table testing sequence Similar to the case of SLE ground-shaking events, f̄1 following
DBE and MCE ground motions shows a gradual reduction of up to
Natural frequency, Damping
f̄ (Hz) ratio (%) 20% relative to the initial natural frequency. Following the first
Mode DBE ground motion corresponding to WN 07, f̄ 1 displayed reduc-
number Direction FDD SSI EFDD SSI tions up to 25% relative to the initial frequency value. WN 08 oc-
Mode 1 EW 1.39 1.34 3.64 7.35 curred after Intervention 2, which consisted of reinforcing the base
Mode 2 Torsional 1.49 1.51 2.24 2.11 beam with steel stiffeners. The addition of steel stiffeners caused an
Mode 3 NS 1.95 1.96 3.18 1.45 increase in the frequency, returning it to approximately 90% of the
Mode 4 EW 2.83 2.87 8.53 16.39 frequency identified prior to subjecting the structure to earthquake
Mode 5 Torsional 4.25 4.18 3.11 4.50 shaking, as can be seen in Fig. 9. This demonstrates that the base
Mode 6 NS 6.31 6.26 1.70 4.71
beam stiffeners, added to prevent the permanent deformation under
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 8. Identified mode shapes and frequencies using the SSI method.
the CLT walls, resulted in stiffening the overall structural system. minimal structural damage to the structure except for the interface
The increase in natural frequencies after Intervention 2 (between region of the CLT walls with the base beam (Pei et al. 2019a).
WN 07 and WN 08) is also observed in Mode 4 and Mode 5. The damping ratios showed substantial changes for all four
Following Intervention 2, f̄ 4 exhibited a 46% increase, identified modes throughout the testing sequence. The damping ra-
resulting in a frequency of approximately 130% of the frequency tio associated with Mode 1 varied between 5.4% and 9.6% based on
estimated at the start of the test sequence. This notable increase in results of the EFDD analysis (diamonds) and between 3.9% and
frequency points to the importance of foundation effects from soil– 8.6% based on the results of the SSI analysis (circles in Fig. 10).
foundation–structure interactions that can alter the dynamic behav- In Fig. 10, it can be seen that the damping ratio for the main EW
ior of a structure, in this case originating from the stiffening of the modes (Modes 1 and 4) increases between Interventions 1 and 2,
base beam. The authors were not able to conclusively pinpoint the which is an opposite trend seen in the decrease of the natural
main reason for the observed erratic changes between WN 7 and frequencies. Assuming the mass is constant, the critical damping
WN 11 because this was not logged during testing. Nonetheless, may decrease as the stiffness decreases, which is implied as the
the data clearly indicate that changes were made at the end of frequency decreases. Thus, the damping coefficient, i.e., the esti-
Day 3, which are likely related to checking and adjusting of speci- mated viscous damping, would tend to be constant, even though
men, mainly through tightening of the transverse rods, following the damping ratio increases due to the decrease in the critical damp-
the public test performed on Day 3 of testing. ing. The testing and analysis cannot confirm this explicitly, and
During WN 18, after Intervention 5, f̄2 , f̄3 , and f̄6 experienced further research could be developed to answer this specific hypoth-
increases of up to 12%, 13%, and 22%, respectively. The frequency esis in a controlled small-scale experiment.
increases after Intervention 5 are only observed in the NS directions More considerable differences in damping ratios were observed
modes (f̄3 and f̄ 6 ) and in the first torsional mode (f̄ 3 ), demonstrat- as a result of using the two methods of analysis. As an example,
ing the effects of tightening the steel cross-brace rods. The steel Mode 4 displayed large differences in damping ratio when ex-
cross-brace rods were added on the north and south center bays tracted using the two modal analysis methods for the majority
of the testing sequence. The difference is more pronounced in
to reduce torsional responses in the structure during shake-table
the first half of the testing sequence, where there is a ratio of almost
testing. The cross-braces were retightened during several of the in-
four times between the damping ratios observed using the EFDD
terventions to maintain their effectiveness.
method and the ones observed using the SSI method [Fig. 10(d),
The most considerable reduction in the frequency for Mode 1
WN 04]. Such differences in damping ratios are most likely due to
was observed following the 1.2 × MCE Northridge event at the end
current limitations in the estimation of damping under low excita-
of the testing sequence. It resulted in a 28% reduction relative to the
tion levels when OMA methods are used (Magalhães et al. 2010;
initial frequency.
Moaveni et al. 2014). The structure is expected to respond in a
Other than Mode 1, the estimated natural frequencies of other
quasi-linear manner in response to the low-amplitude system iden-
identified modes exhibited only moderate decreases at the end of tification white-noise acceleration shake-table tests performed.
the shake-table experiment, with the second-highest reduction Thus, it is expected that the damping ratio value estimates are in-
being 10% for Mode 2. Interestingly, at the end of the testing se- dependent of the level of damage observed, unlike what would be
quence, Modes 3 through 6 displayed natural frequencies that were expected if the base motion induced large nonlinearities, as ex-
equal or slightly above the frequencies estimated at the beginning pected during large intense ground shaking.
of the testing sequence. This observation points toward the limited
effects of damage in modes perpendicular to the direction of shak-
ing and due to the effective retightening of the steel cross-braces Finite-Element Modeling Results
over the testing sequence. The moderate changes in frequencies Fig. 11 shows the frequencies estimated using the FE model
were also consistent with the field observations reporting only normalized by the identified frequencies obtained using the SSI
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 9. Changes in natural frequencies for (a) Mode 1; (b) Mode 2; (c) Mode 3; (d) Mode 4; (e) Mode 5; and (f) Mode 6. Y-axis values are natural
frequencies normalized to the natural frequencies at the beginning of the testing sequence.
method. In the interest of brevity of this paper, comparison are only developing the elastic structural models that account for the effects
made with results based on the SSI analysis. Overall, there is a good of damage, four effects were considered: (1) damage to the base
agreement in most of the FE model frequencies, except f̄4 , which beam, (2) damage to the rocking wall base, (3) loosening of
exhibits a 46% difference between the FE model frequencies and cross-braces, and (4) combined effects of Effects 1–3. These are
the SSI-identified frequencies. detailed in the next subsections.
In identifying reasons for the differences in f̄ 4 , a sensitivity
study of the FE model revealed that a change in the effective stiff- Damage to the Base Beam
ness of the tongue plates resulted in a considerable shift in the The base beam damage was idealized by creating plastic hinges at
frequencies of the nonmatching modes. At the same time, the the two edges where the top flange met the two web plates of the
change in effective stiffness of the tongue plates produced a small base beam (quasi-box beam). The plastic hinges were assumed to
effect on other FE model modes. Although a single rotational stiff- yield and assigned a postyield hardening modulus of elasticity of
ness cannot fully represent the exact mechanics of the tongue plate 1% Es . The modeled reduction on the base beam stiffness resulted
connections, the best match in f̄ 4 and f̄ 5 occurred with the use of a in natural frequencies that were 86% and 94% of the initial values
single rotational stiffness with an effective length of the plate equal for the EW and torsional frequencies, respectively. The stiffness
to 304 mm, which corresponded to the distance between the first reduction of the base beam did not result in a reduction in the
bolt of the tongue plate and the centerline of the shear wall. When NS fundamental frequency (f̄3 ). The lack of frequency reduction
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 10. Changes in damping ratios for the identified modes: (a) Mode 1; (b) Mode 2; (c) Mode 3; (d) Mode 4; (e) Mode 5; and
(f) Mode 6.
for Mode 3 is consistent with the experimental results, where no as a reduction in the cross-sectional area of the compression block.
reduction was observed following Intervention 2 before WN 08, As a result, a spring stiffness reduction factor of 1/20 was intro-
when the base beam was reinforced with additional steel stiffeners. duced by fitting the loss of stiffness due to crushing on five-ply
The effect of considering the damage to the base beams in terms of CLT samples in tests conducted by Barbosa et al. (2019). The
the natural frequencies are presented in Fig. 12 as horizontal solid effect of this damage resulted in natural frequencies that were
lines (level 1 lines). 84% and 93% of the experimental EW and torsional frequencies,
respectively. The NS fundamental frequency exhibited a 2% reduc-
Damage to the Base of the Rocking Wall tion due to this effect. Fig. 12 shows these results as horizontal
The second effect considered was the stiffness reduction of the wall dash-dotted lines (level 2 lines).
corner springs at the intersection between walls and the base beam.
Using Eq. (5), the wall spring stiffness was computed to be approx- Loosening of Cross-Braces
imately 5.25 × 106 kN=m. The damage observed in the wall cor- The third effect considered was the reduced axial stiffness of
ners included splitting, local crushing, and minor gaps between the cross-braces. The axial stiffness was considered to change
the base of the wall and base beam. These features can be modeled by a factor of 0.25 to match the total reduction range observed
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 11. FE modes shapes. The parameter f̄ represents the FE modal frequencies normalized by the corresponding modal frequencies identified using
the SSI analysis (Fig. 8).
for the NS direction because damage to the base beam and the rock- The MAC is defined by the following equation:
ing walls did not affect the NS-direction frequencies. The reduction
in cross-brace stiffness resulted in 93% of the experimental fre- ðϕTa ϕb Þ2
MACab ¼ ð6Þ
quency in the NS direction. The torsional fundamental frequency ðϕTa ϕa ÞðϕTb ϕb Þ
showed a reduction of up to 88% of the experimental frequency at
the start of the shake-table tests. The EW fundamental frequency where ϕa = mode shape a; and ϕb = mode shape of mode b. For the
remained unchanged as a result of reducing the axial stiffness of the purpose of this study, the MACab is used to compare the mode
cross-braces. The effect of loosening cross-braces is shown in shape at the beginning of the experimental testing program with
Fig. 12 with horizontal solid lines (level 3 lines). the same mode shapes at several stages of the experimental tests
as well as the FE mode shapes.
Combined Damage Effects Fig. 13 features a selection of four MAC plots to show a com-
The next and last case considered was introduced to provide an parison of experimental mode shapes taken at different stages of the
encompassing case that captures the observed damage by combin- testing sequence, as well as comparisons of experimental and FE
ing the effect of loss of stiffness due to the damage at the steel base model mode shapes at the start of the shake-table sequence. The
beam and CLT wall base as well as the loosening of the cross-braces experimental MAC plots shown represent the MAC values between
that were described previously. A modification was made to the modes identified in WN 01, at the start of the test, and modes iden-
rocking wall base damage effect by introducing a reduction factor tified in a subsequent WN test, ranging from WN 03 to WN 21
of 1/600 (through calibration) to simulate more significant local- [e.g., Fig. 13(b) corresponds to MAC values between modes iden-
ized damage as observed at the end of the shake-table testing se- tified from WN 01 and WN 03 accelerations]. For all stages, Modes
quence. The combined damage effects resulted in frequencies of 1, 3, and 6 diagonal MAC values remained above 0.90 for the test-
75%, 84%, and 93% of the initial experimental frequencies in ing sequence. Modes 2, 4, and 5 recorded lowest diagonal MAC
the EW, torsional, and NS directions, respectively. The last case values of 0.8 (for WN 08), 0.76 (for WN 10), and 0.86 (for
combining all damage effects is presented in horizontal dashed WN 15), respectively. A few notable trends were observed between
lines (level 4 lines) in Fig. 12. changes in mode shapes and the schedule of interventions carried
Level 1 damage is consistent to the damage to the base beam, out during testing. Mode 2, which is the first torsional mode, dis-
which was partially remediated during Intervention 2. This obser- played its lowest consistency to its initial mode shapes following
vation explains the increase in f̄1 after Intervention 2 and points to Interventions 1 and 2. The MAC values between Mode shape 2
the subsequent reductions being related to the damage to the base of following Interventions 1 (WN 03) and 2 (WN 08) and the initial
the wall. estimates for Mode shape 2, were 0.85 and 0.80, respectively.
For Mode 2, lower MAC values following these interventions
point to the effect of retightening the cross-braces and making
Mode-Shapes Analysis and FE Validation
modifications to the tongue plate connections. Mode 4, being
The modal assurance criteria (MAC) analysis (Pastor et al. 2012) the second EW mode, displayed abrupt reductions in MAC values
was performed for all WN tests to evaluate the consistency of the following Interventions 1 and 4. The corresponding diagonal
experimental mode shapes and observe changes in mode shapes MAC values dropped to 0.92 and 0.79 following Interventions 1
due to the damage in the structural system. The comparison (WN 03) and 4 (WN15), respectively. Mode shape 4 appears to
consisted of computing MAC values between the modes identified be largely influenced by the load-transfer mechanism at the tongue
at the beginning of the testing sequence and all subsequent plate connections, which were designed to transfer lateral transla-
WN tests. tional forces without transferring vertical forces or moments.
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 12. Damage FE modeling for the fundamental natural frequencies: (a) Mode 1, EW direction; (b) Mode 2, torsional direction; and (c) Mode 3,
NS direction. Levels 1, 2, and 3 indicate the damage to the base beam, damage to the rocking wall base, and the loosening of cross-braces,
respectively. Level 4 considers the combined effects of Levels 1–3 of the damage modeling.
The observed changes in Mode shape 4 justify the modifications Conclusion

made to the tongue plate stiffness during the FE model calibration
phase. In a similar way to Modes shapes 2 and 4, Mode shape 5 A seismic experimental program was conducted on a full-scale
recorded the lowest MAC value following Intervention 4 (WN 15), 2-story CLT rocking wall structure at the University of California
further highlighting the influence of retightening the cross-braces San Diego shake-table testing facility. The shake-table program
and making modifications to the tongue plate connections. consisted of 14 ground motions of increasing intensity ranging
The FE model mode shapes were compared with the experimen- from SLE to 1.2×MCE ground motions for a hypothetical site lo-
tal mode shapes at the start of the shake-table testing sequence. The cated in Oakland, California. In general, the 2-story CLT structure
initial FE model and the combined damage effect model mode with the post-tensioned rocking walls performed well in the sense
shapes were compared with the experimental mode shapes at that it provided a seismic resilient design. In this study, the resil-
the beginning of the testing sequence. All diagonal MAC values iency of the design was confirmed because there was no significant
between the initial experimental and the initial FE model were damage observed. Even though the structure was subjected to 14
above 0.92, except for Mode 6, which had a diagonal MAC value intense shake-table tests, only moderate reductions (at most 28%
of 0.88 [Fig. 13(e)]. The diagonal MAC values between the initial reduction in frequency) in the fundamental natural frequency in
experimental and FE combined damage effects modes were above the direction of shaking were observed.
0.94, except for Mode 6, which had a MAC value of 0.77. Addi- The current study explored changes in natural frequencies and
tionally, Mode 4 of the FE combined damage effects was not de- damping ratios as a result of the damage sustained as shaking in-
tectable, most likely due to its proximity to local modes. tensity increased as well as features of the testing program, which
J. Struct. Eng., 2021, 147(4): 04021018

Fig. 13. MAC between experimental Modes 1–6 for WN 01 and (a) itself; (b) WN 03; (c) WN 15; (d) WN 21; and (e) the mode shapes of the
FE model of the initial, undamaged structural model.
included repair and retrofits during the experimental program, in illustrated the impact of damage at the base of the rocking wall
between earthquake shaking tests. The first and second modes structural system as well as the effect of repair and retrofit inter-
in the NS, torsional, and EW directions were identified using ventions performed in the tested structure.
two output-only methods (EFDD and SSI). The first fundamental A linear-elastic FE model was developed to capture the initial
frequency value reduced by 28% at the end of the testing program. modal parameters and subsequent levels of damage consistent with
Reduction in natural frequencies for the first five ground motions the damage observed during testing. The FE model suggests that
was attributed partly to the boundary conditions at the base of the the lateral stiffness of the tongue plate connections is appreciably
CLT walls, including deformation in the steel base beam. The base lower than predicted by an idealized cantilever beam stiffness. Ad-
beam was retrofitted with the addition of stiffeners at high CLT wall ditionally, the tongue plate stiffness has a strong influence on the
bearing force locations in the second intervention. Following this second EW natural frequency due to the added flexibility at its
second intervention, a gradual and less abrupt reduction in the fun- location.
damental natural frequency was observed. The second natural fre- The shake-table test also provided insights into the influence of
quency in the direction of shaking initially decreased but increased foundation–structure effects (stiffening of the base beam), retight-
after Intervention 2 to levels above its initial value. The increase in ening of the cross-braces, and modifications made to the tongue
the second natural frequency in the direction of shaking originated plates. The impact of each intervention could not be evaluated sep-
from the modifications made at the tongue plate connections be- arately because multiple actions were done simultaneously for
tween the CLT walls and the diaphragms. some of the interventions. Further research on the CLT rocking wall
It is well-known that the global modal features may not always system is needed to determine the effects of the diaphragm to wall
detect the effects of local damage; however, results in this study connections for higher modes of vibrations so that interferences
J. Struct. Eng., 2021, 147(4): 04021018

between higher modes and fundamental modes of vibrations can be Blomgren, H., S. Pei, Z. Jin, J. Powers, J. Dolan, J. van de Lindt,
avoided, especially as taller buildings with similar lateral systems A. Barbosa, and D. Huang. 2019. “Full-scale shake table testing
and details are designed. of cross-laminated timber rocking shear walls with replaceable compo-
nents.” J. Struct. Eng. 145 (10): 04019115. https://doi.org/10.1061
/(ASCE)ST.1943-541X.0002388.
Bogensperger, T., T. Moosbrugger, and G. Silly. 2010. “Verification of
Data Availability Statement CLT-plates under loads in plane.” In Proc., 11th World Conf. on Timber
Engineering 2010, editor by A. Ceccotti. Trentino, Italy: Trees and
Some or all data, models, or code generated or used during the Timber Institute.
study are available in a repository online in accordance with funder Brincker, R., and P. Andersen. 2006. “Understanding stochastic subspace
data retention policies (Pei et al. 2019b), or upon reasonable request identification.” In Proc., 24th Int. Modal Analysis Conf. (IMAC-XXIV):
to the corresponding author. A Conf. and Exposition on Structural Dynamics. Society for Experimen-
tal Mechanics, 461–466. Red Hook, NY: Curran Associates.
Brincker, R., L. Zhang, and P. Andersen. 2001. “Modal identification of

Acknowledgments output-only systems using frequency domain decomposition.” Smart
Mater. Struct. 10 (3): 441–445. https://doi.org/10.1088/0964-1726/10/3
This work was financially supported by the USDA Agricultural /303.
Research Service in cooperation with the Tallwood Design Institute Ceccotti, A., and M. Follesa. 2006. “Seismic behaviour of multi-storey
XLam buildings.” In Proc., COST Action E29 Workshop Earthquake
under Grant No. 58-0204-6-002. Additional thanks to Simpson
Engineering on Timber Structures. Coimbra, Portugal: Community
Strong-Tie and DR Johnson for support. The National Science
Research and Development Information Service.
Foundation also supported this research project through several col- Ceccotti, A., M. Follesa, M. P. Lauriola, and C. Sandhaas. 2006. “Sofie
laborative awards, including CMMI 1636164, CMMI 1634204, project–test results on the lateral resistance of cross-laminated wooden
and CMMI 1634628. The use of the NHERI experimental facility panels.” In Proc., 1st European Conf. on Earthquake Engineering and
is supported by the National Science Foundation’s Natural Hazards Seismicity. Red Hook, NY: Curran Press.
Engineering Research Infrastructure (NHERI) Program. The Ceccotti, A., C. Sandhaas, M. Okabe, M. Yasumura, C. Minowa, and N.
authors would like to especially thank the NHERI at UCSD site Kawai. 2013. “SOFIE project: 3D shaking table test on a seven-storey
management and staff, who helped greatly in the construction full-scale cross-laminated timber building.” Earthquake Eng. Struct.
and testing program. The authors also would like to acknowledge Dyn. 42 (13): 2003–2021. https://doi.org/10.1002/eqe.2309.
individual industry collaborators and students who worked on this Chen, Z., and M. Popovski. 2020a. “Material-based models for post-
project. These include Sarah Wichman, Jace Furley, Brian DeMeza, tensioned shear wall system with energy dissipators.” Eng. Struct.
Gabriele Tamagnone, Daniel Griesenauer, Ethan Judy, Steven 213 (Jun): 110543. https://doi.org/10.1016/j.engstruct.2020.110543.
Kordziel, Aleesha Busch, Ali Hansan, Joycelyn Ng, Monica Liu, Chen, Z., and M. Popovski. 2020b. “Mechanics-based analytical models for
balloon-type cross-laminated timber (CLT) shear walls under lateral
and Ata Mohseni. The opinions presented herein are solely those
loads.” Eng. Struct. 208 (Apr): 109916. https://doi.org/10.1016/j
of the authors.
.engstruct.2019.109916.
Chen, Z., M. Popovski, and A. Iqbal. 2020. “Structural performance of
post-tensioned CLT shear walls with energy dissipators.” J. Struct.
References Eng. 146 (4): 04020035. https://doi.org/10.1061/(ASCE)ST.1943-541X
.0002569.
Akbas, T., R. Sause, J. M. Ricles, R. Ganey, J. Berman, S. Loftus, J. D. CSI (Computers and Structures). 2017. SAP2000: Integrated software for
Dolan, S. Pei, J. W. van de Lindt, and H. E. Blomgren. 2017. “Analyti- structural analysis and design, version 19. Berkeley, CA: CSI.
cal and experimental lateral-load response of self-centering postten- Dröscher, J., and R. Brandner. 2013. PA13-903-1 Abschaetzung der
sioned CLT walls.” J. Struct. Eng. 143 (6): 04017019. https://doi.org/10 Scheibenschubfestigkeit mit dem Pruefverfahren nach Kreuzinger
.1061/(ASCE)ST.1943-541X.0001733. und Sieder 2013. Graz, Austria: Institute of Timber Engineering and
American Wood Council. 2015. National design specification (NDS) sup- Wood Technology.
plement: Design values for wood construction 2015 edition. Leesburg,
Dujic, B., K. Strus, R. Zarnic, and A. Ceccotti. 2010. “Prediction of
VA: American Wood Council.
dynamic response of a 7-storey massive XLam wooden building tested
ANSI/APA (American National Standards Institute/Engineered Wood
on a shaking table.” In Proc., 11th WCTE World Conf. on Timber
Association). 2018. Standard for performance-rated cross-laminated
Engineering, editor A. Ceccotti. Trentino, Italy: Trees and Timber
timber. ANSI/APA PRG-320. Tacoma, WA: ANSI/APA.
Institute.
ASTM 2020. Standard specification for hex cap screws, bolts and studs,
Gagnon, S., and M. Popovski. 2011. “Structural design of cross-laminated
steel, heat treated, 120/105/90 ksi minimum tensile strength, general
use. ASTM A449-14. West Conshohocken, PA: ASTM. timber elements.” In CLT handbook: Cross-laminated timber, edited by
Baird, A., T. Smith, A. Palermo, and S. Pampanin. 2014. “Experimental S. Gagnon and C. Pirvu. Pointe-Claire, QC, Canada: FPInnovations.
and numerical study of U-shape flexural plate (UFP) dissipaters.” In Ganey, R., J. Berman, T. Akbas, S. Loftus, J. Daniel Dolan, R. Sause,
Proc., New Zealand Society for Earthquake Engineering Annual Conf. J. Ricles, S. Pei, J. V. D. Lindt, and H. E. Blomgren. 2017. “Experi-
Wellington, New Zealand: New Zealand Society for Earthquake mental investigation of self-centering cross-laminated timber walls.”
Engineering. J. Struct. Eng. 143 (10): 04017135. https://doi.org/10.1061/(ASCE)ST
Barbosa, A. R., R. Soti, A. Sinha, C. Higgins, R. Zimmerman, and E. .1943-541X.0001877.
McDonnell. 2019. In-plane shear and compression behavior of large Granello, G., C. Leyder, A. Frangi, A. Palermo, and E. Chatzi. 2019.
scale cross-laminated timber panels. Framework Project Testing Rep. “Long-term performance assessment of an operative post-tensioned
Corvallis, OR: Oregon State Univ. timber frame structure.” J. Struct. Eng. 145 (5): 04019034. https://doi
Belleri, A., B. Moaveni, and J. I. Restrepo. 2014. “Damage assessment .org/10.1061/(ASCE)ST.1943-541X.0002308.
through structural identification of a three-story large-scale precast con- Granello, G., A. Palermo, S. Pampanin, S. Pei, and J. van de Lindt.
crete structure.” Earthquake Eng. Struct. Dyn. 43 (1): 61–76. https://doi 2020. “Pres-lam buildings: State-of-the-art.” J. Struct. Eng. 146 (6):
.org/10.1002/eqe.2332. 04020085. https://doi.org/10.1061/(ASCE)ST.1943-541X.0002603.
Blass, H. J., and P. Fellmoser. 2004. “Design of solid wood panels with Higgins, C., A. R. Barbosa, and C. Blank. 2017. Structural tests of
cross layers.” In Proc., 8th World Conf. on Timber Engineering. composite concrete-cross-laminated timber floors. Corvallis, OR:
Helsinki, Finland: Finnish Association of Civil Engineers. Oregon State Univ.
J. Struct. Eng., 2021, 147(4): 04021018

Holden, T., J. Restrepo, and J. B. Mander. 2003. “Seismic performance of Pei, S., J. van de Lindt, A. Barbosa, J. Berman, E. McDonnell, J. Dolan,
precast reinforced and prestressed concrete walls.” J. Struct. Eng. H. Blomgren, R. Zimmerman, D. Huang, and S. Wichman. 2019a. “Ex-
129 (3): 286–296. https://doi.org/10.1061/(ASCE)0733-9445(2003) perimental seismic response of a resilient 2-story mass-timber building
129:3(286). with post-tensioned rocking walls.” J. Struct. Eng. 145 (11): 04019120.
Iqbal, A., S. Pampanin, A. Palermo, and A. H. Buchanan. 2015. “Perfor- https://doi.org/10.1061/(ASCE)ST.1943-541X.0002382.
mance and design of LVL walls coupled with UFP dissipaters.” Pei, S., J. van de Lindt, A. Barbosa, J. Dolan, and J. Berman. 2019b.
J. Earthquake Eng. 19 (3): 383–409. https://doi.org/10.1080/13632469 Two-story wood building test. Shake table test of a two-story mass tim-
.2014.987406. ber building with post-tensioned rocking walls. Thessaloniki, Greece:
Kelly, J. M., R. I. Skinner, and A. J. Heine. 1972. “Mechanisms of energy DesignSafe-CI. https://doi.org/10.17603/ds2-zcb9-ry11.
absorption in special devices for use in earthquake resistant structures.” Popovski, M., J. Schneider, and M. Schweinsteiger. 2010. “Lateral load
Bull. N. Z. Soc. Earthquake Eng. 5 (3): 63–88. resistance of cross-laminated wood panels.” In Vol. 4 of Proc., 11th
Kramer, A., A. R. Barbosa, and A. Sinha. 2015. “Performance of steel en- WCTE World Conf. on Timber Engineering, editor by A. Ceccotti,
ergy dissipators connected to cross-laminated timber wall panels sub- 3394–3403, Trentino, Italy: Trees and Timber Institute, National Re-
jected to tension and cyclic loading.” J. Struct. Eng. 142 (4): E4015013. search Council.
https://doi.org/10.1061/(ASCE)ST.1943-541X.0001410. Priestley, M. J. N. 1991. “Overview of PRESSS research program.” PCI J.

Kurama, Y. C., S. Sritharan, R. B. Fleischman, J. I. Restrepo, R. S. Henry, 36 (4): 50–57. https://doi.org/10.15554/pcij.07011991.50.57.
N. M. Cleland, S. K. Ghosh, and P. Bonelli. 2018. “Seismic-resistant Restrepo, J. I., and A. Rahman. 2007. “Seismic performance of self-
precast concrete structures: State of the art.” J. Struct. Eng. 144 (4): centering structural walls incorporating energy dissipaters.” J. Struct.
03118001. https://doi.org/10.1061/(ASCE)ST.1943-541X.0001972. Eng. 133 (11): 1560–1570. https://doi.org/10.1061/(ASCE)0733-9445
Magalhães, F., Á. Cunha, E. Caetano, and R. Brincker. 2010. “Damping (2007)133:11(1560).
estimation using free decays and ambient vibration tests.” Mech. Syst. Smith, T., F. Ludwig, S. Pampanin, M. Fragiacomo, A. Buchanan, B.
Signal Process. 24 (5): 1274–1290. https://doi.org/10.1016/j.ymssp Deam, and A. Palermo. 2007. “Seismic response of hybrid-LVL
.2009.02.011. coupled walls under quasi-static and pseudo-dynamic testing.” In
Mellinger, P., M. Döhler, and L. Mevel. 2016. “Variance estimation of Proc., 2007 New Zealand Society for Earthquake Engineering Conf.
modal parameters from output-only and input/output subspace-based Wellington, New Zealand: New Zealand Society for Earthquake
system identification.” J. Sound Vib. 379 (Sep): 1–27. https://doi.org/10 Engineering.
.1016/j.jsv.2016.05.037. Smith, T., S. Pampanin, A. Buchanan, and M. Fragiacomo. 2008. “Feasibil-
Milaj, K., A. Sinha, T. H. Miller, and J. A. Tokarczyk. 2017. “Environmen- ity and detailing of post-tensioned timber buildings for seismic areas.”
tal utility of wood substitution in commercial buildings using life-cycle In Proc., 2008 New Zealand Society for Earthquake Engineering Conf.
analysis.” Wood Fiber Sci. 49 (3): 338–358. Wellington, New Zealand: New Zealand Society for Earthquake
Moaveni, B., A. R. Barbosa, J. P. Conte, and F. M. Hemez. 2014. “Uncer- Engineering.
tainty analysis of system identification results obtained for a seven-story Structural Vibration Solutions. 2017. “ARTeMIS modal.” Accessed April
building slice tested on the UCSD-NEES shake-table.” Struct. Control 15, 2018. http://www.svibs.com/products/ARTeMIS_Modal.aspx.
Health Monit. 21 (4): 466–483. https://doi.org/10.1002/stc.1577. Sustersic, I., M. Fragiacomo, and B. Dujic. 2016. “Seismic analysis of
Mugabo, I., A. R. Barbosa, and M. Riggio. 2019. “Dynamic characteriza- cross-laminated multi-story timber buildings using code-prescribed
tion and vibration analysis of a four-story mass timber building.” Front. methods: Influence of panel size, connection ductility, and schematiza-
Built Environ. 5: 86. https://doi.org/10.3389/fbuil.2019.00086. tion.” J. Struct. Eng. 142 (4): E4015012. https://doi.org/10.1061
Palermo, A., S. Pampanin, and A. Buchanan. 2006a. “Experimental inves- /(ASCE)ST.1943-541X.0001344.
tigations on LVL seismic-resistant wall and frame subassemblies.” In Tannert, T., M. Follesa, M. Fragiacomo, P. Gonzalez, H. Isoda, D. Moroder,
Proc., 1st European Conf. on Earthquake Engineering and Seismology. H. Xiong, and J. van de Lindt. 2018. “Seismic design of cross-laminated
Red Hook, NY: Curran Press. timber buildings.” Wood Fiber Sci. 50: 3–26. https://doi.org/10.22382
Palermo, A., S. Pampanin, A. Buchanan, and M. Newcombe. 2005. /wfs-2018-037.
“Seismic design of multi-storey buildings using laminated veneer van de Lindt, J., J. Furley, M. Amini, S. Pei, G. Tamagnone, A. Barbosa,
lumber (LVL).” In Proc., 2005 New Zealand Society for Earthquake D. Rammer, P. Line, M. Fragiacomo, and M. Popovski. 2019. “Exper-
Engineering Conf. Wellington, New Zealand: New Zealand Society imental seismic behavior of a two-story CLT platform building.” Eng.
for Earthquake Engineering. Struct. 183 (15): 408–422. https://doi.org/10.1016/j.engstruct.2018.12
Palermo, A., S. Pampanin, M. Fragiacomo, A. H. Buchanan, and B. L. .079.
Deam. 2006b. “Innovative seismic solutions for multi-storey LVL tim- van de Lindt, J. W., S. Pei, S. E. Pryor, H. Shimizu, and H. Isoda. 2010.
ber buildings overview of the research program.” In Proc., 9th World “Experimental seismic response of a full-scale six-story light-frame
Conf. on Timber Engineering WTCE 2006. Portland, OR: Oregon State wood building.” J. Struct. Eng. 136 (10): 1262–1272. https://doi.org/10
University Conference Services. .1061/(ASCE)ST.1943-541X.0000222.
Pastor, M., M. Binda, and T. Harčarik. 2012. “Modal assurance criterion.” Wichman, S. 2018. “Large-scale dynamic testing of rocking cross lami-
Procedia Eng. 48: 543–548. https://doi.org/10.1016/j.proeng.2012.09 nated timber walls.” M.Sc. thesis, Dept. of Civil and Environmental
.551. Engineering, Univ. of Washington.
Pei, S., D. Rammer, M. Popovski, T. Williamson, P. Line, and J. W. Van De Zimmerman, R. B., and E. McDonnell. 2018. “Framework—Innovation in
Lindt. 2016. “An overview of CLT research and implementation in re-centering mass timber wall buildings.” In Proc., 11th US National
North America.” In Proc., WCTE 2016 World Conf. on Timber Engi- Conf. on Earthquake Engineering. Oakland, CA: Earthquake Engineer-
neering. Vienna, Austria: Vienna Univ. of Technology. ing Research Institute.
J. Struct. Eng., 2021, 147(4): 04021018

System Identification of UCSD-NHERI Shake-Table Test of Two-Story Structure With Cross-Laminated Timber Rocking Walls

Uploaded by

Copyright:

Available Formats

System Identification of UCSD-NHERI Shake-Table Test of Two-Story Structure With Cross-Laminated Timber Rocking Walls

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

System Identification of UCSD-NHERI Shake-Table Test of Two-Story Structure With Cross-Laminated Timber Rocking Walls

Uploaded by

Copyright:

Available Formats

System Identification of UCSD-NHERI Shake-Table

Test of Two-Story Structure with Cross-Laminated

Introduction sustainability, and aesthetics. As an engineered wood product,

© ASCE 04021018-1 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

© ASCE 04021018-2 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

Fig. 1. Two-story CLT rocking wall structure.

© ASCE 04021018-3 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

Experimental Setup, Testing Sequence, and

© ASCE 04021018-4 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

Fig. 4. Ground-motion testing sequence.

© ASCE 04021018-5 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

cusing on a frequency range that could be used to identify the first

© ASCE 04021018-6 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

E2 ¼ E0 · k 4 ð1cÞ CLT Handbook (Gagnon and Popovski 2011). Table 1 summarizes

© ASCE 04021018-7 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

In addition to the observed structural damage, other sources af-

© ASCE 04021018-8 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

© ASCE 04021018-9 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

© ASCE 04021018-10 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

© ASCE 04021018-11 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

© ASCE 04021018-12 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

© ASCE 04021018-13 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

The observed changes in Mode shape 4 justify the modifications Conclusion

© ASCE 04021018-14 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

© ASCE 04021018-15 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

Brincker, R., L. Zhang, and P. Andersen. 2001. “Modal identification of

© ASCE 04021018-16 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

https://doi.org/10.1061/(ASCE)ST.1943-541X.0001410. Priestley, M. J. N. 1991. “Overview of PRESSS research program.” PCI J.

© ASCE 04021018-17 J. Struct. Eng.

J. Struct. Eng., 2021, 147(4): 04021018

You might also like