2-D and 3-D basin site effects in Izmir-Bayrakli during the October 30, 2020 Mw7.0 Samos earthquake

During the October 30, 2020 Mw7.0 Samos Earthquake, the Turkiye-Izmir-Bayrakli district was affected the most due to the geometry and the deep-soft alluvial nature of the basin, although the district is approximately 70 km away from the epicenter. In this study, the seismic response of the Bayrakli basin, and the role of the soil stratigraphy and basin geometry on the recorded amplifications and prolonging of seismic shakings are investigated by using 1-, 2- and 3-D non-linear finite element-based dynamic response analyses. The assessment results are presented in the form of spectral amplification ratios, AT\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A\left(T\right)$$\end{document}. The highest A(T)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A(T)$$\end{document} values were estimated at spectral periods T = 0.85 and 1 s. They exceed five in both the east–west and the north–south directions, more pronounced in the latter one. AT\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A\left(T\right)$$\end{document} was decomposed into the product of two independent amplification factors, namely rock to soil amplifications, ASoil(T)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{Soil}(T)$$\end{document}, and 1-D soil column to 3-D soil basin amplifications, ABasin(T)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{Basin}(T)$$\end{document}. ABasin(T)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{Basin}(T)$$\end{document} values for T = 0.85 and 1 s are estimated as high as 1.4 and 1.6, in the east–west and the north–south directions, respectively. These values suggest that the 3-D geometry of the Bayrakli basin amplifies the spectral accelerations by 40–60% at T = 0.85 and 1 s. The unique combination of a deep-soft alluvial site with 3-D basin geometry, and overlying 7–9 story residential buildings, consistently favors and amplifies the seismic energy in the spectral period range of 0.7–1.0 s. This multi-fold increased seismic demand, combined with poor structural design and construction details, lead to localized structural damage and over 117 life losses.


Introduction
The Bayrakli district, which is home to more than 300,000 residents, is located in the Izmir Metropolitan City of Turkiye. Despite the earthquake epicenter being approximately 70 km away, the district suffered the most damage during the M w 7.0 Samos Earthquake on October 30, 2020. It is located on a very deep alluvial basin, which is bordered by Mount Yamanlar and Mount Spil in the north, and Mount Nif in the east, as illustrated in Fig. 1. The western border of the basin is outlined by Izmir Bay. In the figure, the approximate boundary of the basin is shown with a yellow dashed line. The concentrated damage in the district is mainly attributed to local site effects, namely the amplification and prolonging of seismic waves that were trapped in the soft and deep alluvial sediments.
The Samos event was monitored densely by available strong ground motion stations (SGMS) in Izmir Bay. Their locations are shown in Fig. 1. Moreover, in Table 1, the station coordinates, their rupture distances (R rup ), and representative shear wave velocity for the Fig. 1 Bayrakli basin, and the locations of nearby strong ground motion stations in east Izmir Bay 1 3 upper 30 m, (V s,30 ) are presented. The peak ground accelerations (PGA) recorded on rock and soil sites in Izmir Bay vary in the range of 0.03-0.06 g and 0.06-0.15 g, respectively. METU EERC 2020;Cetin et al. 2021;Cetin et al. 2022). Similarly, the maximum spectral acceleration (S a ) values on rock and soil sites vary in the range of 0.09-0.17 g and 0.20-0.53 g, respectively. The predominant periods (T pre ), defined as the spectral periods corresponding to maximum amplitudes of the spectra, are estimated to vary in the range of 0.32-1.67 s. Similarly, the significant durations (D s ) vary in the range of 19.7-25.9 and 17.0-26.1 s at rock and soil sites, respectively. D s is defined as the duration of shaking, corresponding to 5-95% arias intensities of the accelerograms. Among these, SGMS #3513, #3514, and #3522, as highlighted in Table 1, are the three important stations due to the reasons that they are located i) in the Bayrakli basin, and ii) on soil and rock sites. This unique setting enabled to assess quantitatively the seismic soil site effects during the 2020 Samos event. Figure 2 presents the acceleration time histories, Fourier amplitude, and elastic response spectra of these records. As revealed by these figures, the SGMS #3514 (rock), #3513 (soil), and #3522 (soil) records exhibited different intensity levels in their east-west (E-W) and north-south (N-S) components. The rock PGA's are recorded as 0.057 g in the E-W, and 0.040 g in the N-S components. The #3513 soil record has PGA levels of 0.097 g in the E-W, and 0.108 g in the N-S components, while #3522 shows 0.065 g in the E-W, and 0.075 g in the N-S components. The significant durations of the horizontal components vary in the range of 20.16-25.90 s.
A simple proportioning of these PGA levels suggested rock to soil amplification factors, falling in the range of 1.1 ( ≅ 0.065 0.057 ) to 2.7 ( ≅ 0.108 0.040 ) due to site effects. Similarly, and more importantly, spectral acceleration (S a ) values at the 0.7-1.4 s period range reach 0.137 g versus 0.117 g, and 0.347 g versus 0.344 g, and 0.230 g versus 0.324 g, in the E-W and N-S directions at stations #3514, #3513, and #3522, respectively. Soil to rock spectral amplifications reach a factor of maximum three ( ≅ 0.344 0.117 ) . Consistent with these amplified seismic demand levels, the structural damage is mostly concentrated on structures in the Bayrakli basin with natural periods coinciding with the amplified period range of 0.7-1.4 s.
In this manuscript, site effects and corollary amplifications in seismic demand are quantitatively assessed through a set of 1-, 2-and 3-D non-linear finite element-based, dynamic site response models. The aim is defined as to evaluate the effects of the 2-D and 3-D basin geometry, geological and geotechnical setting on the amplified and prolonged shaking levels in the Bayrakli basin.

Geological and geophysical setting of the Bayrakli basin
Bayrakli basin is composed of Quaternary alluvial deposits, while their foothills extend along the rising rock outcrops in the north and south. Alluvial bedding and its features are spatially variable, as shown in Fig. 3. The bedrock of the northern slope descends towards the south and reaches a depth of approximately 1000 m on the alluvial floor immediately under the Bayrakli district center. Additional geological descriptions are available in Akgun et al. (2011), Akgun 2012, Pamuk et al. (2017, 2018a, Kıncal et al. (2009), Ozdag et al. (2015 and Cetin et al. (2022).
A field survey was performed to develop idealized geophysical profiles of Izmir Bay, more specifically of the Bayrakli basin. The surveys employed multi-channel analysis of surface wave (MASW), refraction microtremor (ReMi), and microgravity measurements.

3
The geophysical studies encompassed 10 km long profiles with sampling intervals of 500 m, selected by considering the land use and urbanization limitations (Akgun et al. 2011(Akgun et al. , 2012. The geophysical survey points and lines are shown in Fig. 3, where yellow and purple dots indicate the locations of MASW and ReMi studies. Within the scope of the field studies, the basin is split into grids in the N-S and E-W directions. Figure 4 shows the grid layout along with the locations of SGMS #3514, #3513, and #3522. In the same figure, the approximate boundary of the basin is shown by a yellow  dashed line. The yellow dots show the scatter of the collapsed or heavily damaged buildings in the basin. The N-S (N-S1 through 6) and E-W (E-W3 through 7) extending blue and red solid lines, respectively, are selected as the sections, for which 2-D seismic site response analyses are to be performed. Additionally, 1-D site response analyses are performed on the soil profiles corresponding to the intersection points of these E-W and N-S lines. Last but not least, the green box designates the boundary of the 3-D basin model, which will be discussed in detail, later in the manuscript.
The representative shear wave velocity profiles are generated based on MASW measurements. The density of each sublayer is estimated by benefiting from the shear wave velocity correlations presented in Akgun et al. (2012). Then, they are calibrated with the microgravity measurements. A sample of the soil and shear wave velocity profiles is shown in Fig. 5 for the cross-section N-S2. A complete documentation of the 2-D basin shear wave velocity profiles is also presented in the Electronic Supplement A. The resulting idealized 2-D soil and shear wave velocity profiles were then consolidated into a 3-D basin profile.

Idealized basin soil profiles
The resulting 2-and 3-D geophysical profiles are used to develop idealized models for the numerical assessments. Figures 6, 7, and 8 show a sample of the 1-D, 2-D, and 3-D profiles, respectively. The irregularities or discontinuities, as already marked with blue circles in Fig. 5, are smoothed when developing these idealized profiles. The resulting 1-D models have approximately 500 elements, whereas 2-and 3-D models have 17,000 and 1,800,000 elements, respectively. Due to very long run times (21 days for the 3-D model), the 3-D basin model has to be downsized to cover only the sections bordered by E-W3 to 7 and N-S1 to 4, where most of the structural damage was concentrated.

The constitutive model and input parameters
The hardening soil (HS) model with small-strain stiffness (HS-small), available in Plaxis 2-D and 3-D software, is used as part of the assessments. HS is an advanced model for simulating the behavior of both soft and stiff soils (Schanz et al. 1999). Some basic characteristics of the model are: (1) stress-dependent stiffness following to a power law (input parameter m), (2) plastic straining due to primary deviatoric loading (input parameter E 50,ref ), (3) plastic straining due to primary compression (input parameter E oed,ref ), (4) elastic unloading/reloading (input parameters E ur,ref , ν ur ), (5) failure according to the Mohr-Coulomb failure criterion (parameters c′, ϕ′ and ψ). HS-small constitutive model is based on the HS model and uses almost entirely the same set of parameters. Only two additional parameters are needed to describe the variation of stiffness with strain: (1) the initial or very small-strain shear modulus, G 0 , and (2) the shear strain level (γ 0.7 ) at which the secant shear modulus, G S , is reduced to 70% of G 0 . The response simulated by HS-small constitutive model is comparatively shown in Fig. 9, along with Vucetic and Dobry (1991) modulus degradation curves for soils with plasticity index (PI) of 0, 15, and 30%. The damping response in the HS-small model is governed by stress-dependent unloading modulus, hence a direct comparison with corollary shear strain-dependent Vucetic and Dobry (1991)  damping curves cannot be presented. A list of input parameters for the HS-small constitutive model is presented in Table 2.

Meshing, boundary conditions and the deconvolution procedure
The mesh and element sizes are determined considering the maximum frequency (minimum period) of the seismic wave, which can propagate through the mesh. The minimum period of the wave is selected as approximately 0.5 s, to assure reliable predictions at the natural period of the residential buildings. Then, as per the findings of Kuhlemeyer and Lysmer (1973), the average mesh size is estimated as one-fourth of the wavelength. The resulting average mesh sizes are estimated to vary in the range of 30-400 m, considering the shear wave velocity of the soil or rock sublayers. The element sizes were automatically adjusted by the mesh generator available in Plaxis software, on the basis of consistently defined local coarseness factors (Plaxis 2021).
The side boundaries in 1-D assessments are defined as "tied-degree of freedoms". They are used to connect the same elevation nodes at the left and right boundaries of the mesh window. This choice enables that both left-and right-side boundaries are subjected to the same horizontal and vertical displacements. In 2-D and 3-D models, the "free-field" lateral boundary elements are used to extend the side boundaries to infinity, simulating far-field conditions by applying equivalent normal and shear forces. In the normal and shear directions, two dashpots are added at each node to absorb the reflected secondary waves from the internal structures. In all models, the base boundary conditions were defined as "compliant-base". It is a combination of line-prescribed displacement and a viscous boundary, and is used at seismic bedrock interfaces (Plaxis 2021; Joyner and Chen 1975). More detailed discussions on these boundary elements are available in Plaxis (2021), Joyner and Chen (1975) and Amorosi et al. (2010).
The #3514 outcropping rock records were de-convolved by using Deepsoil software (Hashash et al. 2020) to a "within motion" at 1500-2000 m depths, where seismic bedrock was located. When deconvolving the outcrop records, the #3514 profile, presented in Fig. 10, was used along with the modulus degradation and damping curves listed in Table 3. Figure 11 shows comparatively the "outcrop" motions recorded at the #3514 site, along with the "within" motions, estimated by the deconvolution process.

Validation of the 3-D site response model
The 3-D Plaxis basin model and the input parameters are cross-validated against the responses recorded in the basin. Validation with multiple recordings from the basin would help in understanding spatially variable seismic basin response. However, only a single station, i.e.: SGMS #3522, is available in the basin model. Figure 12 shows the recorded and the resulting elastic response spectra evaluated by the 3-D basin model. The solid and dashed lines show the recorded and the estimated responses, respectively. In the figure, the natural period range of damaged residential buildings (0.5-1.0 s) is also marked. The agreement between the recorded and the estimated responses are observed to be good, despite the #3522 SGMS located to be in the vicinity of the 3-D model boundaries. This reasonable agreement also eliminated the possible concerns regarding the compliance effects at the boundaries.

Amplification ratios by 1-, 2-and 3-D site response models
With the intent of letting the variability in the basin geometry and the soil profiles be the only independent variables of the problem, all the other attributes of the numerical assessments are kept constant. Simply, the same set of software, constitutive model, parameter sets, and element sizes, along with the seismic input motions, were used for the analyses. Numerical assessments were performed by using Plaxis 2-D or 3-D software. The same HS-small constitutive model with Mohr-Coulomb failure criterion was used in the The results are presented in the form of amplification ratios, A(T) , expressed in terms of spectral accelerations. A(T) is defined as 3-D soil basin spectral accelerations normalized by the reference rock spectral accelerations, as given in Eq. 1. A(T) is then decomposed into the product of two amplification factors, aiming to quantify independently the site amplifications due to: a) 1-D soil stratigraphy, and b) 3-D (or 2-D) basin geometry. The first of these amplification factors, A Soil (T) , represents the spectral amplification of a 1-D soil column as compared to a reference rock column. Hence, it is defined consistently, as presented in Eq. 2. The latter of the two, A Basin (T) , simply represents the spectral soil basin accelerations estimated by a 3-D model normalized by the ones of a 1-D model (i.e.: 1-D soil column vs. 3-D soil basin). Similarly, it is given in Eq. 3.
The reference rock site was chosen as the outcrop rock recording at SGMS #3514. Consistent with these definitions, A(T) , A(T) Soil , A(T) Basin values were estimated for spectral  periods, T , of 0.5, 0.85, 1, 1.4 s. Their spatial distributions are mapped as given in the Electronic Supplement B. Note that, as discussed earlier due to the coarseness of the mesh size, estimated spectral acceleration responses for periods shorter than 0.5 s, are judged to be not dependable. Hence, they are not presented. The highest amplification factors were observed at the spectral period range of 0.85-1 s. Hence, A(T) values are presented in Fig. 13 and 14 for T = 0.85 and 1 s, respectively.
As illustrated by these figures, A(T) values for the spectral periods T = 0.85 and 1 s were assessed to exceed a value of five in both the E-W and the N-S directions, more pronounced in the latter one. In simple terms, the SGMS #3514 outcrop record spectral accelerations were amplified by a factor of exceeding five in the critical parts of the basin, confined by sections N-S1-3 and E-W4-6.  1 3 The highest amplified region (shaded by warmer colors) correlates well with the scatter of the moderate to highly damaged buildings, shown with yellow dots. Note that as discussed elsewhere (Cetin et al. 2021(Cetin et al. , 2022, the majority of these moderate to highly damaged buildings were 7-9 story residential buildings with natural periods falling into the spectral period range of 0.7-1.1 s. Their natural periods coincide with most-amplified spectral periods suggesting a resonating soil-structure response. The correlation between the amplified seismic demand and the structural damage scatter is shown to be more pronounced in the N-S direction than in the E-W direction. This observation is not surprising since the Bayrakli basin is two-way confined by Mount Yamanlar and Mount Nif in the N-S direction. In the E-W direction, the basin is only confined in the east and opens to Izmir Bay in the west (i.e.: one-way confined in the E-W direction).  Alternatively, the A Basin (T) values were also estimated by using the results of the 2-D basin models, as given in the Electronic Supplement B. A Basin (T) values are estimated as high as 1.4 and 1.6, in the east-west and the north-south directions, respectively. These values suggest that the 3-D geometry of the Bayrakli basin produces 40-60% higher spectral accelerations than the values estimated by the 1-D soil column model. Again, geometric amplifications were more pronounced in the N-S direction for the same to as high as 2.3. These amplification levels are judged to be comparable with the values estimated as part of this study. Despite the consistency in the estimated amplification levels, it should be noted that the basin amplification responses are basin-and seismic event-specific due to nonlinear nature of seismic basin responses, and cannot be

A(T), A(T) Soil , A(T)
Basin values can also be estimated by using the results of the 2-D site response models, as an alternative to results of the 3-D model. Their contour maps are also documented in Electronic Supplement B. As illustrated by these figures, the amplifications estimated by 1-D, 2-D, and 3-D models were observed to be significantly different. For example, A(T) amplification predictions by 1-D models suggest that they were mostly amplified at the northern and southern edges of the basin, whereas the 3-D model suggests maximum spectral acceleration amplification responses occurring at the deepest part of the alluvial basin, confined by the sections N-S1-3 and E-W4-6. Similar non-consistent amplification predictions by 1-D and 2-D and 3-D models were also observed at spectral period of 1.0 s, and other periods are no exception. These inconsistent amplification predictions by 1-D, 2-D, and 3-D models suggest that simplifications of a 3-D basin geometry through 1-D or even 2-D idealized models may not guarantee consistent predictions with the ones by 3-D models.

Summary and conclusions
During the October 30, 2020, M w 7.0 Samos earthquake, the most affected district was reported as Turkiye-Izmir-Bayrakli, although its epicenter is approximately 70 km away. Izmir-Bayrakli is founded on a very deep alluvial basin bordered by the hills to the north and the south. The unique nature of the basin soil stratigraphy and geometry were speculated to be the governing reasons behind this unusually localized structural damage levels observed during a relatively distant seismic event. The Samos event was monitored densely in Izmir Bay by available strong ground motion stations. SGMS #3513, #3514, and #3522 are the three of these stations, which are all located in Izmir-Bayrakli, on soil and rock sites. This setting enabled to assess quantitatively the seismic soil site and basin effects during the 2020 Samos event.
To quantitatively assess the role of soil site conditions on the observed amplifications and prolonging of the seismic demand, 1-D soil column, 2-and 3-D basin models were developed. The 3-D finite element model, the adapted constitutive models, and their input parameters were validated with the recordings at SGMS #3522. The agreement with the estimated and recorded responses at #3522 was judged to be good and mutually supportive. All the variables of the model (e.g.: the software, model mesh size, constitutive model, model input parameters, seismic within bedrock motion, etc.) were kept the same to assess uniquely the basin geometry effects on spectral amplifications. The estimated overall spectral amplification, A(T) , defined as the 3-D soil basin spectral accelerations normalized by reference rock spectral accelerations, was decomposed into the product of two independent amplification factors: A Soil (T) and A Basin (T).
The first of these amplification factors, A Soil (T) , represents the spectral amplification of a 1-D soil column as compared to that of a reference rock column (site). The reference rock site response was chosen as the outcrop rock recording at SGMS #3514. The latter of the two, A Basin (T) simply represents the spectral soil amplifications due to the 3-D soil basin. The numerical simulation results were summarized in the form of A(T) , A Soil (T) , and A Basin (T) contour maps.
The highest amplification factors were observed at the spectral periods of 0.85-1 s range. A(T) values at these spectral periods exceed five in both the E-W and the N-S directions, more pronounced in the latter one. In simple terms, the SGMS #3514 outcrop record spectral accelerations were amplified by a factor exceeding five in the critical parts of the basin, confined by sections N-S1-3 and E-W4-6. Highly amplified region correlates well with the scatter of moderate to heavily damaged buildings. The majority of these buildings were 7-9 story residential buildings with natural periods falling into the spectral period range of 0.85-1.0 s. Their natural periods coincide with the most-amplified spectral periods suggesting a resonating soil-structure response. Hence, it can be concluded that increased seismic demand levels due to site effects (due to both soil stratigraphy and 3-D basin geometry) could be listed as one of the major contributors to the observed overall structural damage, along with other structural design and construction factors discussed elsewhere (Yakut et al. 2021).
The correlation between the amplified seismic demand and the structural damage scatter is shown to be more pronounced in the N-S direction than the one in the E-W direction. This observation is not surprising since the Bayrakli basin is two-way confined by Mount Yamanlar and Mount Nif in the N-S direction. In the E-W direction, the basin is only confined in the east and opens to Izmir Bay in the west (i.e.: one-way confined in the E-W direction). Consistently, induced seismic waves were trapped inside the basin in the N-S direction, producing higher amplifications in the N-S direction.
3-D basin geometry-induced amplifications, A Basin (T) , were assessed after normalizing the 3-D basin model results with those of the 1-D model. A Basin (T) values for T = 0.85 and 1 s are estimated as high as 1.4 and 1.6, in the E-W and the N-S directions, respectively. These values suggest that the 3-D geometry of the Bayrakli basin produces 40-60% higher spectral acceleration demands than the ones estimated by the 1-D soil column model. In simpler terms, the same soil column is subjected to 40-60% higher intensities because the soil column is located in a basin, more specifically in the Bayrakli basin. Again, the basin geometry-induced amplifications were more pronounced in the N-S direction due to the two-way confined nature of the basin in this direction. Moreover, these geometric amplification factors also reach their highest values at the spectral period range of 0.85-1.0 s. Hence, both the soil stratigraphy-and basin geometry-induced amplifications reach their maximum values at the period range of mostly damaged 7-10 story residential buildings of Bayrakli.
Alternatively, A(T) , A(T) Soil , A(T) Basin values were also estimated by using the results of the 2-D site response models, as opposed to the results of the 3-D model. The basin amplifications by 2-D and 3-D models were observed to be significantly different. These inconsistent amplification predictions suggest that simplifications of a 3-D basin geometry through 1-D or even 2-D models may not guarantee consistent predictions with the ones by 3-D models.
As the concluding remark of the trilogy of manuscripts (Cetin et al. 2022(Cetin et al. , 2023 and this manuscript), the relatively slower rupture mechanism of the Samos event (Kiratzi et al. 2021) produces mid to longer period-(0.5-1.0 s) rich seismic source energy. This unusually long period rich seismic energy was not damped but preserved when propagating through mostly fractured regional bedrock towards Izmir Bay. When reached Izmir Bay, this longer period-rich bedrock level seismic demand was further amplified by deep soil columns of the Izmir-Bayrakli basin, with coinciding natural periods varying in the range of 0.7-1.4 s. Hence, bedrock level seismic shaking resonated the overlying soil columns at the period range of 0.7-1.4 s. The 3-D nature of the basin further favored the spectral seismic energy at this period range and amplified it by a factor of 1.4 and 1.6 in the E-W and N-S directions, respectively. Then, finally, this already twice amplified (first by the soil column and then by the 3-D basin geometry) seismic shaking resonated the overlying 7-9 story residential buildings, on which the majority of building damage was concentrated. The unique combination of a) relatively slower seismic rupture mechanism, b) propagation in highly faulted/fractured bedrock (path effects), c) site soil column, d) 3-D basin geometry, and e) overlying 7-9 story residential buildings, consistently favor and amplify the seismic energy in the spectral period range 0.7-1.0 s. This multi-fold increased seismic demand levels, combined with poor structural design and construction details, lead to poor structural performances and over 117 life losses.
Funding The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.
Data availability All data and models analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors have no relevant financial or non-financial interests to disclose.