Seismic assessment of the in‐plane/out‐of‐plane interaction of masonry infills in a two‐storey RC building subjected to bi‐directional shaking table tests

This paper presents a series of shaking table tests conducted on a 60%‐scaled two‐storey infilled reinforced concrete building. The main parameters include the presence of openings in the masonry infills and the effect of bi‐directional dynamic shaking to the latter. The building was subjected to four different levels of shaking intensity, ranging up to the Maximum Considered Earthquake (MCE), equivalent to 150% of the design earthquake. The study focuses on and evaluates the damage caused by combined in‐ and out‐of‐plane shaking. Under the design level earthquake, the second‐storey infill walls with openings were severely damaged, with near‐collapse of a masonry pier. For the fully infilled direction without openings instead, diagonal cracking as well as out‐of‐plane inclination of the infill wall in the second storey were observed. Due to the interaction between the two loading directions and the direct adverse effect of the in‐plane damage on the out‐of‐plane resistance of the infills, the out‐of‐plane collapse of 3/4 of the second storey masonry infill occurred during the MCE. The reported damage observations on the masonry‐infills with or without openings provide further insights on the vulnerability of non‐seismically designed reinforced concrete buildings. Additionally, the seismic performance of the structure is assessed using further structural response quantities recorded during testing, including floor accelerations, inter‐storey drifts, base shear forces and reductions in natural frequency. Future testing will focus on investigating the effectiveness of retrofitting measures in improving the connection between the infill and frame, reducing in‐plane damage, and preventing out‐of‐plane failure of the infills.


INTRODUCTION
According to the latest European Seismic Risk Model, masonry-infilled reinforced concrete (RC) buildings constitute the highest contributors to expected losses due to earthquakes, both in terms of human life and economic losses. 1 The in-plane damage and collapse of masonry infills were recently demonstrated on a full-scale, fivestorey RC building tested under multiple seismic intensities. 2Moreover, post-earthquake reconnaissance studies echo these results, with the out-of-plane (OOP) collapse of infills, particularly in higher storeys of buildings, often being one of the most significant contributor to building damage, human casualties and economic losses [e.g., in refs.3-7].
Under real earthquake conditions, the vulnerability of infill walls to OOP collapse is increased by the interaction of damage sustained due to in-plane deformations and inertial forces acting in the transverse direction. 8Despite the significance of out-of-plane collapses on human and economic losses, only limited research has been carried out to better understand this interaction [e.g., in refs.9 -11].In most cases in-and out-of-plane behaviour are investigated in separate steps, for example, by testing infilled RC frames or infill walls first to different levels of in-plane deformation, before subjecting the pre-damaged infill panels to testing in the OOP direction.The latter may be performed either through monotonic out-of-plane loading, 9,[12][13][14] airbag testing, 10,15 or shake table tests on individual infilled RC frames [e.g., in ref. 16, 17].
Tests on full buildings, and in particular, shake-table experiments to investigate the in-and out-of-plane behaviour considering dynamic loads over the whole infill area are however extremely rare.A summary of the main characteristics of previous shake-table tests on infilled RC buildings is provided in Table 1.It is noteworthy that most of the tested buildings were designed to modern seismic design guidelines, and only half considered openings, which also affect the behaviour of infilled RC frames significantly. 18The effect of simultaneous shaking in both horizontal directions has been considered by Fardis et al., 19 which highlighted the importance of bi-directional shake table testing to adequately characterise the out-of-plane vulnerability of infilled buildings, however without observation of infill collapse up to the maximum tested earthquake intensity (PGA = 0.6 g).Lourenco et al. 20 tested three buildings under bi-directional shaking, with only the non-ductile prototype collapsing through a soft-storey mechanism, following expulsion of the ground-storey infills at a PGA of 0.76 g.
In the present study, the dynamic behaviour of a non-seismically designed RC building infilled with solid concrete blocks is investigated for the first time with simultaneous loading in two directions.Parameters of investigation include different types of openings, as well as the combined effect of in-and out-of-plane damage mechanisms.The reference building studied herein represents a typical low-rise Korean school building.The corresponding test specimen is exposed to increasing levels of ground shaking from 30% to 150% of the design based earthquake (DBE).The evolution of damage under combined in-and out-of-plane shaking conditions provide valuable insights on the vulnerability of infilled RC buildings.Experimental results in terms of floor accelerations, inter-storey drifts, base shear forces and reductions in natural frequency are presented in this study.The measurements obtained during the tests enrich the available data on the seismic behaviour of infilled RC buildings, which may be useful for the further development of numerical and analytical modelling.Additionally, the tests on the reference building provide the baseline for future tests carried out within this series of experiments on buildings retrofitted with novel composite materials for combined seismic and energy retrofitting technologies, so far only tested quasi-statically on individual frames.

F I G U R E 1
Prototype building -Typical Korean school buildings.

Specimens design, geometry and materials properties
The reference building is a typical low-rise Korean school building (Figure 1) constructed in the early 1990s.The structure is a non-seismically designed RC frame with masonry infills.As is typical for non-seismically designed buildings, it features hoops evenly and widely spaced through a column and 90 degrees hook.Additionally, the masonry infills were not anchored to the RC frame with steel rods or similar means.
The test specimen was designed to simulate a classroom of the reference Korean school building, as shown in Figure 1.To accommodate the payload capacity of the shake table, the specimen scale was reduced by 60% compared to the prototype, and the geometry was adjusted accordingly.In this study, the Cauchy similitude law 26,27 was adopted and the parameter relationships are provided in Table 2.
The test specimen depicted in Figure 2 and Figure 3 is a two-storey, three-dimensional building with masonry infill walls in every RC frame.Two infills, South and North, have openings representing door and windows.While the test specimen is representative of the prototype building, its design is adjusted to account for the tests' needs.For instance, the size of the columns was designed considering the slab area supported by six columns of the prototype building.The cross-sectional sizes of the columns and beams were 200 × 200 mm and 200 × 300 mm, respectively.The columns are reinforced with four longitudinal bars, each with a diameter of 13 mm (4-D13), and hoops with a diameter of 6 mm (D6) spaced 120 mm apart (D6@120).The hoops tied the longitudinal bars with 90 degrees hook.The reinforcement details of the beam are the same as those of the columns (see Figure 2A and 2B).
The masonry walls comprised a single layer of concrete bricks with a thickness of 57 mm.The size of the infills were 2.6 m in length and 1.53 m in height, resulting in a length-to-height ratio (L/H) of 1.7.The thickness of each slab was 120 mm.The first-floor slab and beams were cast as one piece.In contract, the second-floor slab was constructed separately from the rest of the structure, and after installing the weight blocks, it was connected to the beam with high-tension bolts.
The installation of weight blocks was necessary in the shake table test of the scaled model in order to achieve the scaled total storey mass in accordance with the similitude law and hence generate the adequate corresponding inertial forces.In accordance with Cauchy's similitude law, an additional mass of approximately 96 kN was required, which was applied using a steel plate assembly of two times 48 kN per storey.This adjustment not only addressed the mass scaling discrepancy but also ensured that the columns experienced a similar axial force ratio (0.15) as those in the prototype.
The concrete used to construct the stub (base) and frame had nominal compressive strength (  ) of 24 and 18 MPa, respectively.The respective mixing ratios for the concrete used in constructing the building are shown in Table 3.The nominal yield strength (  ) of the steel rebar was 400 MPa.The results of tensile tests on the steel reinforcement bars tested to KS B 0802 28 are shown in Table 4.The concrete blocks employed to construct the infills had a standard compressive strength of 13 MPa determined to Korean Standard KS F 4004. 29 The mortar used to construct the infill walls had an average compressive strength of 20.8 MPa, based on three compression tests on 50 × 50 × 50 mm cubes according to KS L 5105. 30Moreover, three concrete block masonry prisms (dimensions: 190 × 190 × 90 mm) were tested under compressive loading with an average strength f m = 11.6 MPa, and three wallettes were tested under diagonal compression to determine the in-plane shear strength of masonry (τ max = 1.93 MPa) according to ASTM E519. 31

Experimental setup and test procedure
The foundation slab of the specimen was fixed to the shake table using high-tension bolts.Accelerometers and Linear Variable Displacement Transducer (LVDT) sensors were installed to measure the response of the structure.The sensor locations are indicated in Figure 4. Accelerometers were installed on the shake table and at the centre of the three floor slabs to capture the input and response accelerations.In order to measure the global deformation of the specimen, LVDTs were installed at two points per floor slab in both directions.In addition, eight LVDTs were placed at the corner of the infills without opening to assess their behaviour.All measurement signals were recorded at a sampling rate of 256 Hz using a Data Acquisition system.

Input signal
An artificial earthquake acceleration record was generated specifically for the shake table test, based on the Korean seismic design spectrum. 32The bi-directional earthquake record used for the test was scaled based on the similitude law outlined in Table 2.The duration of the excitation was reduced to 0.6 times its original length, while the acceleration was increased by a factor of 1.67.Spectrum matching of the X-and Y-components was done independently, ensuring a cross-correlation Seismic design spectrum reference response spectrum (RRS) and pseudo-acceleration spectra of the artificial record's X and Y components (5% damping).Acc (g) Time (sec) value between the two components below 0.16 according to the Korean Design Standard. 33Furthermore, a lower-bound of 90% and an upper-bound of 130% from 0.04 to 10 s were applied according to the requirements of the standard.The seismic design spectrum and the pseudo-acceleration spectra of the artificial records are shown in Figure 5, while the scaled input motions in the X-and Y-directions are shown in and Figure 6.A dynamic identification test was performed to evaluate the frequency characteristics of the structure before the earthquake simulation tests, using a random excitation with frequency range from 0.1 to 50 Hz for 30 s.The earthquake simulation tests were conducted simultaneously in two orthogonal directions while gradually increasing the acceleration intensity of DBE (Design Based Earthquake) from 30% to 150%.The test program is shown in Table 5.

TA B L E 6
Definition of damage states adapted from ref. [34].The resonance frequency was determined by calculating the transfer function of the response acceleration (Unit, b) at each position of the specimen for the acceleration input (Base, a) from the shaking table.As indicated in Equation (1), the transfer function (T ab ) was calculated by the Cross Power Spectral Density (P ba ) of the input/output signal to the Power Spectral Density (P aa ) of the input signal.To improve the accuracy of resonance analysis, a symmetric hamming window was applied to each signal.
3 Experimental results and discussion

Failure mechanisms and damage states
In this section the observed damage at the different stages of the shaking table tests is described in detail.During the tests, increasing level of damage was observed and qualitatively associated to different damage states with reference to the criteria proposed by Rossetto and Elnashai, 34 from which five damage states were considered, as summarised in Table 6.In Table 7, the observed damage states in the X-and Y-directions are linked with the main recorded results in the positive and negative directions during the shake table tests.Next to the peak ground acceleration (PGA), the acceleration response recorded at the roof (top of the second floor), that is the peak floor acceleration (PFA2), the floor acceleration factor (FAF2), that is the ratio between peak roof and peak ground accelerations, as well as the peak inter-storey drift ratios of the first and second floor, Δ max,1 and Δ max,2 , respectively, are presented.The table also provides the equivalent natural frequency f eq , which was evaluated during the seismic excitation, rather than from a random vibration test. 35This was done to avoid any further damage in the vulnerable infills by using a transfer function of the measured roof acceleration

(A) (B)
F I G U R E 7 Graphical representations of the peak inter-storey drift ratios recorded at each level of seismic intensity at the first and second floor.
response and the ground acceleration 1 .Finally, the peak base-shear force (V max ), as well as the observed damage states are also summarised in Table 7.
Graphical representations of the peak inter-storey drift ratios and accelerations recorded at each level of seismic intensity at the base, first and second floor are provided in Figure 7 and Figure 8, respectively.
In terms of observed damage, during the first tests at 30% of the DBE corresponding to an average PGA of 0.12 and 0.155 g in X and Y directions, respectively, cracking initiated along the RC frame-infill interfaces in both X-and Y-directions.As shown in Figure 9A, in the X-direction small stair-stepped diagonal cracks at the upper and lower corners of the openings were also observed in the first and second storey, with a thin horizontal crack also extending between the window and the door opening in the South frame (Figure 9B).Additionally, limited vertical cracking was also seen in the first floor beams, as well as cracks between the slab and the beam (Figure 9C). 1 In the case of the 100% and 150% DBE a Fast Fourier Transform of the recorded roof displacements was used instead, as the accelerometer readings presented too much noise for an accurate natural frequency detection.At 60% DBE (0.26 and 0.32 g PGA in X and Y, respectively), the diagonal cracks at the corners of the openings extended to the corners of the infill walls, as shown in Figure 10A for the North frame.In the Y-direction without opening, the first stepped-diagonal and horizontal cracks appeared on the infill wall of the East frame.Increased cracking in the beams and along the infill-frame interface was also observed (Figure 10B).In particular, in the second storey of the East side gap-openings between column and infill up to 1 mm were recorded, with a residual opening of 0.6 mm.At the same time, detachment from the top beam and slight inwards leaning of the infill wall can be observed in Figure 10B.This was also confirmed by the LVDT reading (D18), which presented a small residual gap opening (0.4 mm) in the vertical direction due to the leaning of the infill wall.It worth's noting that, although cracking is still minor at this stage of testing and the integrity of the infills and RC frame is fully preserved, the peak inter-storey drift of +0.49% and +0.23% reached in the X and Y-direction, respectively, are higher than the drift limits for DS2 and DS1 set-out in ref. [36], respectively.
The first observation of severe damage can be made at 100% DBE (0.46 and 0.59 g PGA in X and Y, respectively), with the second floor masonry pier in the South side separating and twisting, remaining in a position of near-collapse (Figure 11A).In the corners of the infill walls, spalling and separation from the RC frame were observed and cracks in the beams became more noticeable.During the test, significant opening of the cracks around the windows, with separation between bricks was noticed.The observed damage in the X-direction visibly corresponds to DS3, for which the drift threshold (0.75% 36 ) was in fact exceeded with a peak inter-storey drift of 0.81% in the second floor.
In the stiffer Y-direction at 100% DBE, the diagonal stepped cracks in the infills increased, particularly in the second floor, but remain limited.Moreover, additional horizontal cracks due to shear sliding appeared at the lower third part of the infill in the East frame, as can be seen in Figure 11B.Increased out-of-plane bulging of the entire second floor East wall was observed (Figure 11C), meaning DS3 was reached.Out-of-plane failure was initially prevented as the infill was forming an arching mechanism to resist it.However, during the test, failure of the mortar bond between the infill and the upper beam was observed, with a residual gap opening measured at ca. 1 mm, leading to the infill to lean due to rigid body motion and out-of-plane failure vulnerability for higher intensity shaking.This was however not the case in the West frame in which no residual gap opening was measured.
Finally for the 150% DBE (0.74 and 0.88 g PGA in X and Y, respectively), corresponding to the Maximum Considered Earthquake (MCE), collapse of masonry infill walls on three sides of the second floor was reached, as shown in Figure 12.First, the already precarious South-side masonry pier between window and door collapsed, leading to out-of-plane collapse of the entire South wall (Figure 12A).The twisting of the pier which led to failure of the infill was a consequence of bidirectional shaking.Out-of-plane collapse of the North-side wall ensued directly afterwards (Figure 12B).In the stiffer Y-direction, the East side infill wall experienced increasing diagonal cracking, with out-of-plane collapse initiating in the upper-third of the wall towards the inside of the building.
Next to the collapse of the infill walls, as shown in Figure 13, significant longitudinal splitting cracks followed by limited spalling of the concrete cover were also observed at the base of the second storey RC columns along the lap-splice length (520 mm) of the main reinforcement (Figure 13A), as well as flexural cracking and concrete crushing at the upper end (Figure 13B).
To summarise, Table 8 provides for each panel, at each intensity level, the observed damage and crack pattern.

Reduction in natural frequency
With increasing damage in the infill walls, the stiffness of the building is reduced in both directions, and it is hence also expected to observe a reduction in its natural frequency.The natural frequency in both directions was initially determined before testing, with the resonant frequency in the X direction of 4.0 Hz (T n = 0.25 s), and the resonant frequency in the Y

F I G U R E 1 4
Evolution of the first-mode natural frequencies of building CT-1 after each testing stage.
direction of 8.25 Hz (0.12 s).In the case of the X direction, it is judged that the natural frequency is reduced by about two times compared to the Y direction due to the opening in the infill wall.As shown in Figure 14, a reduction in equivalent natural frequency as a result of the degradation of the lateral in-plane stiffness of the infilled frames with increasing levels of damage was observed at higher seismic intensities.The difference in frequency between the X-and Y-direction remained significant in the 30% DBE tests.With increased detachment of the infills in both directions at 60% DBE, a strong reduction in natural frequency was observed at 60% DBE.After failure of the infills at 150% DBE, the natural frequency in both directions are very similar with f = 1.6 and 2.2 Hz, for the X-and Y-direction, respectively.The small difference can be attributed to the West-frame not collapsing.

Acceleration response
Figure 15 displays the acceleration time-histories for the two-storey building at four different levels of earthquake intensity (ranging from 30% to 150% DBE) in both shaking directions.The ground and roof accelerations are shown (measured at the top of the second storey).Up to the 150% DBE, the response in the Y-direction without openings exhibits lower floor accelerations compared to the X-direction with openings.Only at the 150% DBE, when the collapse of the second storey infills with openings occurred prior to the collapse of the infill in the Y-direction, the peak floor acceleration in the Y-direction surpassed that of the X-direction.This is also reflected by the FAF, that is the ratio between peak roof and peak ground accelerations (as presented in Table 7).The second storey FAF reduces progressively with increasing PGA.For the softer X-direction, it reduces from 3.1 for the lowest PGA to 1.3 for the 150% DBE.This indicates increased energy dissipation due to damage of the infill and frames, as also observed by. 35In the stiffer Y-direction the FAF is lower than in the X-directions, with the peak roof acceleration being about double of the ground acceleration at 30% DBE (FAF2 = 2.0), reducing to values between 1.0 and 1.3 for higher earthquake intensities.

Force-displacement response
Based on the acceleration time-histories provided in the previous section, the base shear force was estimated by considering inertial forces with acceleration readings in both X-and Y-directions for both storeys, utilizing storey masses of 16.44 and 15.99 tonnes for the first and second storey, respectively.Note that this implicitly includes the damping forces.Figure 16 illustrates the hysteretic response of the building, representing the base shear force (kN) against the second storey inter-storey drift, under increasing levels of seismic intensity (ranging from 30% to 100% DBE) in both shaking directions.The envelope curve corresponding to each level of seismic intensity is also presented.It is important to note Ground and roof acceleration time-histories for the four levels of earthquake intensity.
that during the final test (150% DBE), an accurate calculation of the base shear force was not feasible due to the progressive failure and collapse of the infills, which made it impossible to assess the actual storey masses.
A symmetric in the positive and negative directions of shaking can clearly be observed for the three levels of seismic intensity for the X-direction, with the peak forces obtained at 100% DBE of +115.7 and −125.6 kN.The stiffness response can be seen to reduce from 30% DBE to 100% DBE, echoing the trend in the natural frequency identification.In the Y-direction, the hysteresis loops are less easily distinguished, but a similar trend to the X-direction is observed and higher sustained base shear forces of +122.1 and −151.7 kN.
One important difference in the hysteretic response is the reduced inter-storey drift ratios in the Y-direction (without openings).As expected, due to the reduced stiffness of infills with openings, the peak inter-storey ratios are significantly higher in the X-direction (see Figure 16 and Figure 17).In both directions, the highest inter-storey drift ratios are reached in the second floor for all four stages of testing.Starting at 60% DBE, the inter-storey drift between the first and second storey exhibits a significantly larger difference in the X-direction compared to the Y-direction, mainly due to the rotation and loss of connection of the masonry pier in the South wall of the second storey, with the X-direction drift nearly twice that of the Y-direction.Once the out-of-plane collapse of the second storey infill walls occurs at 150% DBE, F I G U R E 1 7 Recorded peak inter-storey drifts in the first (1F) and second floor (2F) in both directions of shaking, compared with damage limits for DS1 to DS4 proposed by 36 for infills with and without openings, respectively.
the peak inter-storey drift ratio in the in-plane direction increases drastically, as the frames act as bare ones.Figure 17 illustrates this phenomenon clearly, showing that at 150% DBE, the peak inter-storey drift reaches 3.4%, compared to 0.8% at 100% DBE.In the Y-direction, a similar pronounced effect is observed, with the inter-storey drift increasing from 0.4% to 1.8% at 100% and 150% DBE, respectively.The difference in observed damage between the East and West infill walls (without openings) is also observed in the recorded inter-storey drift time-histories shown in Figure 18.In the East frame higher inter-storey drifts compared to the West frame are recorded from the 30% DBE test, however at very low levels.The difference becomes more significant at 100% DBE (double) and 150% DBE (more than triple) due to bulging and later collapse of the East infill.A comparison between North (window opening) and South frames (window and door opening), instead, highlights no significant differences apart from the peak inter-storey drift recorded at the 150% DBE.The latter may be a consequence of the coincidence with the peak differential displacement between East and West frames at the same time leading to torsional rotations.

Infill separation and out-of-plane collapse
The recorded difference in inter-storey drifts between East and West frames without openings in Figure 18 echoes differences in observed damage and disconnection of the infill from the frame.The latter being more significant in the East frame, leads to a lower stiffness and hence higher drifts.It is well known that the degradation of the boundary between frame elements and masonry infills can reduce their out-of-plane capacity, as it prevents an adequate arching mechanism Inter-storey drift time history recorded during the four testing stages.
to form. 18Next to visual observations, differences level of are also confirmed by the higher values of opening in the horizontal and vertical direction for the East frame, as shown in Figure 19.For the East frame, more significant residual displacements were recorded, indicating that the gap between frame and infill is not closed after the tests.The reason for increased infill-separation and hence lower out-of-plane capacity in the East-side may be a consequence of differences in workmanship, as also noted in other studies. 37nset of out-of-plane inclination can be seen from vertical infill-frame gap measurements in the East frame from 60% DBE.The residual displacement (ca.0.5 mm) in the vertical gap measurement indicates the infill is leaning at the end of the test.At 100% DBE this is more significant (with a residual opening of −1 mm).This was also confirmed by visual inspection (Figure 11C).
Although no accelerometers were utilized to measure the maximum sustained acceleration at the centre of the infill panels, the measurements of gap openings at 150% DBE were employed to determine the occurrence of out-of-plane expulsion of the infills.This expulsion took place when the displacement transducers failed at t = 4 s, as clearly depicted in Figure 19D.After experiencing significant in-plane damage, as shown in Figure 11A, the masonry infills with openings subsequently collapsed out-of-plane (Figure 12) following a PGA of −0.77 g and PFA of 0.95 g in the X-direction.A similar F I U E 1 9 Measurements of vertical and horizontal infill-frame separation in the East and West frames.pattern was observed in the Y-direction, where the solid infills on the East collapsed out-of-plane (Figure 12) at a higher PGA of 0.95 g and PFA of 1.04 g, following previous in-plane damage illustrated in Figure 11B and C.
It is worth noting that at 100% DBE, the infill without openings did not collapse for a PGA of 0.63 g (PFA = −0.54g), which is in line with previous experimental results on a similar building tested under simultaneous horizontal shaking in two directions, 19 for which no out-of-plane collapse of the second-storey infill was observed at a PGA of 0.60 g.

CONCLUSIONS
This study aimed to evaluate the seismic behaviour of a two-storey reinforced concrete building with masonry infills through a series of shaking-table tests.The research focused on investigating the differences in response between infills with and without openings, as well as examining the impact of bi-directional horizontal excitations on the vulnerability of the infill walls to out-of-plane failure.Due to higher stiffness in the fully infilled Y-direction compared to the X-direction of the building, differences in peak inter-storey-drift ratios, floor acceleration, as well as sustained base shear forces were observed.The degradation in natural frequencies was accelerated for the fully infilled Y-direction, particularly at higher earthquake intensities when the connection between frame and infill was reduced.
The evolution of damage exhibited significant variations between the two directions of the building.In the fully-infilled frames, in-plane diagonal cracking, infill-frame separation, as well as out-of-plane inclination were observed, leading ultimately to the collapse of one infill.Conversely, in the frames with openings, cracking initiated at the corners of windows and doors, and subsequent out-of-plane failure ensued after significant rotation and damage to the masonry pier.At the design level earthquake, the building already sustained severe damage, including near-failure of the masonry pier between openings in the X-direction, and significant out-of-plane inclination and degradation at the boundaries in the Y-direction.At the MCE, out-of-plane collapse was reached in both directions, at peak floor acceleration of 0.95 and 1.04 g in the infills with and without openings, respectively.The observation of out-of-plane failure in the second-storey infills confirmed the increased vulnerability of higher storey infills when subjected to simultaneous in-plane and out-of-plane loading for the case of non-seismically designed reinforced concrete buildings.Similar findings were reported for masonry-infilled reinforced concrete buildings designed according to EC8. 19 Finally, the observed damage and failure mechanism of the infills highlights the need for bi-directional shake table testing to replicate realistic conditions.While the out-of-plane response of the infills is a consequence of their inertia in the transverse direction, degradation of the infill-frame boundary due to in-plane displacements accelerates their outof-plane failure.In the fully-infilled direction (Y-direction), the East infill panel resisted out-of-plane failure through an arching mechanism at lower intensities.However, due to in-plane loading and the separation from the boundary frames, it collapsed at higher intensity under inertial forces.Bi-directional shaking conditions also accelerated the failure of the masonry pier in the X-direction, hence further confirming the need for bi-directional shake table testing to adequately characterise the vulnerability of infilled RC buildings.
The findings from this experimental study offer valuable insights for future numerical modelling efforts, particularly due to the limited availability of data investigating the in-and out-of-plane behaviour of infilled RC frames under bidirectional dynamic loading.Additionally, future research will investigate OOP-collapse prevention though different integrated seismic and energy retrofitting schemes using composite materials and pre-cast panels on the same building prototype.

A C K N O W L E D G E M E N T S
The experiments presented in this paper were carried out in the framework of the Collaborative Research Arrangement No. 34064 between the European Commission, Joint Research Centre (JRC) and the Korea Construction Engineering Development, Collaboratory Management Institute (KOCED-CMI) within the International Collaborative R&D Programme Call (KAIA-2-2019), organised by the Korean Agency for Infrastructure Technology Advancement (KAIA).These experiments were also financially supported by the Technology Promotion Research Programme through the research grant 21CTAP-C152795-03, funded by the Ministry of Land, Infrastructure and Transport of the Korean Government.The data post-processing work was partially funded by the JRC in the framework of the iRESIST+ research project.The authors dedicate this work to the late Hyoungsuk Choi, who contributed to this research.

D ATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

F I G U R E 2
Geometry of the specimen.

8 F I G U R E 9
Graphical representations of the peak accelerations recorded at each level of seismic intensity at the base, first and second floor.Note: First floor accelerometer malfunctioning at 150%DBE.Observed damage at the end of the 30%DBE test: (A) North view; (B) South second storey; (C) North first storey.DBE, design based earthquake.

F I G U R E 1 0
Observed damage at the end of the 60%DBE test: (A) North view; (B) East second storey.DBE, design based earthquake.F I G U R E 1 1 Observed damage at the end of the 100%DBE test: (A) South-East view; (B) Second floor East frame; (C) visible detachment of infill panel in East frame.DBE, design based earthquake.

F I G U R E 1 2
Observed damage at the end of the 150% DBE test: (A) South-East view; (B) North-East view.DBE, design based earthquake.F I G U R E 1 3 Damage to RC columns after 150% DBE test: (A) lower end; (B) upper ends of the second floor columns.DBE, design based earthquake; RC, reinforced concrete.

6
Hysteretic responses in both directions of shaking (A) X-direction: Infills with Openings; (B) Y-direction Solid infills. 25 Characteristics of previous shake table tests on infilled RC buildings.
TA B L E 1Abbreviation: RC, reinforced concrete.

F I G U R E 3 Isometric view of the test specimen. TA B L E 3
Mixing ratios for the concrete in the frame and slabs of the building specimen.

TA B L E 4 Steel material properties. Bar diameter f y f t E s [mm] [MPa] [MPa] [GPa]
F I G U R E 4Location of the main sensors (LVDTs and accelerometers) placed on the test specimen.LVDT, linear variable displacement transducer.
Adopted shaking table test sequence of inputs.
TA B L E 5Abbreviation: DBE, design based earthquake.
Summary of results.
TA B L E 7Abbreviation: DBE, design based earthquake.