Performance of a seismically isolated building in Albania during the Durrës sequence of 2019 and a feasibility study of pendulum isolators in the region

This work evaluates the behaviour of the only seismically isolated building in Albania during the Durrës sequence of 2019 and analyses the efficiency and feasibility of pendulum isolators in the region. The behaviour of the existing building is analysed through observational data gathered during on-site inspection and is verified through numerical simulations using the impact of the main shock of 26th of November 2019. The technical efficiency and the economic feasibility of seismic isolation through pendulum isolators for non-strategic buildings are analysed by designing, detailing, and estimating the cost of an ordinary six story residential building in both approaches (capacity design and seismic isolation approach), accounting for the local factors such as typical architecture, soil characterization, the level of the seismic hazard and the cost of the construction materials and workmanship in the densely populated region of central Albania. The structural behaviour assessment highlighted the superiority of the seismic isolation, and the feasibility analysis revealed that the initial cost for delivering seismically isolated buildings is higher than that of the conventional approach by some percentage. However, the gains in structural performance and the relatively high seismic hazard of the country make seismic isolation of buildings a relevant alternative in achieving seismic protection for buildings in the region.


Introduction
With 51 casualties, over 900 injuries and an estimation of the total recovery needs that surpasses 1 billion Euros, the Durrës sequence of 2019 (main shock on 26 th of November, M w 6.4), is one of the most devastating seismic events ever recorded in central Albania [1][2][3]. As reported, most of the casualties were caused by the collapse of 9 buildings, some of which had been previously affected by the foreshock of the 21 st of September 2019 (M w 5.6) [4,5]. Moreover,  [8], with the focus in the territory of Albania, providing the peak ground acceleration (PGA) in units of gravity (g) of seismic events with return period of 475 years. Cold colours indicate comparatively low hazard areas (PGA 0 ≤ 0.1g), yellow and orange indicate moderate-hazard values (0.1g < PGA ≤ 0.25g), and red colours indicate high-hazard areas (PGA ≤ 0.25g) [4,5], and the failure of the respective authorities to take proper measures after the foreshock of 21 st of September, amplified the destructive force of the earthquake and besides the casualties, caused a significant impact in the economy of the country. Clearly, these circumstances expose the high seismic vulnerability in Albania, and therefore the need to incorporate reliable and affordable anti-seismic solutions.
As it is well known, seismic engineering provides two arguably well-established design procedures that render earthquake resisting structures: the capacity design procedure (also referred as conventional/traditional approach) and the seismic isolation approach. Based on the capacity design procedure, traditionally designed buildings are expected to dissipate the energy through the ductility of their structural members [12][13][14][15]. This implies that when the design earthquake strikes, these structures will develop severe structural damage, which might also be unrepairable, but in any case, they must remain within the limits of lifesafety. In contrast, buildings that are designed based on the seismic isolation approach [15][16][17][18][19] are expected to filter the energy that the earthquake induces to the superstructure by altering the dynamic reaction of the structure (i.e. increasing the fundamental period of vibration). This is achieved by decoupling the superstructure from the ground through special bearing devices (seismic isolators) characterized by very low horizontal stiffness that enable safe and controlled relative motion of the superstructure to the footing/substructure. This ensures that the structural members of the superstructure remain undamaged (the building remains operational during the earthquake) and the perception of the seismic event is significantly lower compared to the perception in traditional buildings. Moreover, unlike the traditionally designed structures, isolated structures can bear multiple cycles of design earthquakes while exhibiting the same expected behaviour.
When it comes to selecting the design procedure that the designer should follow, one must consider the efficiency and the feasibility of both these alternatives, obviously in the context of the challenges that each unique case represents. While there is a vast research work that analysed in detail the behaviour of traditionally designed structures under real severe earthquakes that have even sparked improvements in the rules of capacity design, in the case of the seismic isolation approach, mainly because it is not as widely spread as the conventional design is (i.e. as reported in [20], up to 2014, the number of isolated buildings in Japan exceeded 4000), there are relatively few data describing the behaviour of these buildings under real earthquakes, and they mainly refer to cases in developed countries. Such as, in [21] the performance of the USC Hospital seismically building is shown to have significantly reduced (by about 50%) the peak foundation/ground accelerations induced by the Northrige Earthquake of 1994, showing efficiency of the base isolation by keeping the superstructure in the elastic range. The response of an instrumented isolated building during the Chile Earthquake of 2010 is described in [22], where time history data are provided, showing the reduction of the peak ground acceleration induced in the structure but also a considerable reduction of the high frequency of the horizontal signals. The same effects were not observed for the vertical signals. In [23], the efficiency of HDRB is analysed making use of the sequence of low energy earthquakes that were recorded in the Operative Centre of the Civil Protection Department at Foligno (central Italy) that were recorded during period August 2016 to January 2017. In this study it was found that despite the design earthquake of the isolated building was higher than that recorded, the isolation system did activate and provided mitigation of the seismic effect when compared to the fixed based building. Further statistical data on the application and expected efficiency of base isolation in Italy are provided in [24]. The ever growing experience and expertise in base isolation in Japan is described statistically and analysed also in [25][26][27] highlighting not only the efficiency of base isolation in terms of engineering parameters, but also the human perception of these events. It is also pointed out the importance in design and maintenance of the gap around the isolated building and the need for proper considerations of the effect of the vertical earthquake component in near field events. Moreover, the effectiveness and economic efficiency of the seismic isolation varies from country to country depending on local factors such as level of seismic hazard, soil characterization, typology and architecture of the building, qualification of workmanship etc.
In this work, initially the behaviour of the only seismically isolated building in Albania under the Durrës sequence of 2019 is analysed by means of observational data gathered through on-site inspection and numerical analysis performed through the finite element software SAP2000 [28]. Additionally, the observed behaviour of the existing isolated building during the main shock, is compared to the behaviour of other similar fixed based buildings nearby. Next, the technical efficiency and the economic feasibility of pendulum isolation systems in residential buildings in Albania are investigated. This investigation consists in designing, detailing, and verifying the performance of an ordinary six story reinforced concrete building designed for residential purposes using both approaches (capacity design approach and seismic isolation procedure), while considering all the local factors such as typical architecture, seismic hazard, soil characterization, cost of construction materials and workmanship. Using the structural behaviour of the case of study solved through capacity design procedure as a benchmark (since it is the most common approach used in buildings in seismic areas), the efficiency of seismic isolation in the region is assessed while the economic feasibility is evaluated by directly comparing the costs for the initial investment needed to construct the building in both approaches and considering the long-term implications in the property value that these two approaches have (Fig. 3).

Structural behaviour of the Seismically isolated building in Tirana during the 26 th of November earthquake
In this section, the structural behaviour of the only seismically isolated building in Albania during the earthquake of 26 th of November 2019 is analysed using the data gathered from the inspection after the earthquake and verified through numerical analysis.

General description
The existing structure under investigation, is a building of ordinary class of importance designed for residential and commercial purposes. The structure is composed of 2 underground stories (sub-structure/basement) and a combination of two blocks with respectively 8 and 9 stories above the ground (superstructure, Fig. 4-a).
The L shape configuration in plan of the structure, typically induces large irregularities and therefore, the designers introduced a combination of several shear walls (to fit the superstructure with the necessary stiffness in elevation) and the isolation system such that the torsional effects are minimised and the decoupling of the superstructure from the sub-structure is achieved (sufficient difference in the stiffness of the superstructure to the isolation layer allows for the relative movement of the superstructure as a rigid body). The structure is located in Tirana, in the crossroad of the Kavaja Street and Gjergj Fishta Boulevard, in the absolute coordinates 41 19'21.6"N 19 47'31.1"E, 1.82 km away from the TIR1 Station of Institute of Geosciences, Energy, Water and Environment -IGEWE [10], whose recording ( Fig. 3) of the ground motion of the mainshock (on 26 th of November 2019) is used in FEM analysis to quantify the structural behaviour.
Originally, this building was designed using the traditional approach of capacity design and when the investors decided to switch to the seismic isolation approach, the construction works were already underway (2011), and the raft foundation and the first underground floor was completed, which limited the choice of the isolation system. However, given the sufficient story height of the first underground floor, it was decided to duplicate the beam system below and above the isolation layer, to ensure a sufficient stiffness of the supportive system (the substructure) of the isolation layer, without thickening the previously built columns (Fig. 5). The isolation system is comprised of 77 FIP-D 1 of  Table 1). The isolation devices are positioned beneath every vertical structural member (one per each column and at least two per shear wall). Figure 6 shows an image of the isolation device used in the building and an image from the installation of the devices on site. Thus, the fundamental vibration period of the structure was shifted from T f b = 1.02s (fixed base) to T i = 2.64s  (isolated structure) with an ultimate displacement capacity 2 of the isolation system of 24cm.

Structural behaviour -Observations
Following the main shock, a quick visual inspection was conducted from which no sign of structural nor nonstructural damages was reported. Next, a careful inspection was conducted. Firstly, the integrity of the isolation system was checked in compliance with the provisions of EN 1337-10 [30], which concerns the requirements for inspection and maintenance of structural bearings, including seismic isolators. As reported in Fig. 7, some clear lines on the pavement, aligned with the steel plate covering the surrounding gap, suggested that not only the isolation system was activated and the super structure sustained displacement as expected, but also some limited residual displacements in the range of 3-4 mm had been developed. All the isolation devices were carefully checked for any damage or possible failure. The dust protection cover was perturbed, but without any visible damages. The stainless steel surfaces had no visible scratches or any residual material that could emerge from potential damages on the sliding pad, protection cover, hard encrustation or any other unforeseen anomalies. Moreover, sliding pads where in its intended position inside the convex element, while no visible damages (such as distortions, cracks or general damages) were spotted in the steel parts of the isolation units. A divergence between the position of the upper and the lower surfaces was detectable, confirming the presence of small residual displacements. After a brief consultation with the manufacturers/designers of the isolation units (FIP MEC srl), it was concluded that all the devices were in optimal condition, with no damage or need for intervention ( Fig. 8). At the ground level, all the infrastructure of the elevator, piping and electric cables were also in optimal condition. The verification of the integrity of the isolation system was followed by a detailed check of the super structure. A careful inspection of the structural and non-structural members was conducted (based on the provisions of EN1998-3 [31]), and no apparent damage were detected (namely, no visible cracks on the layers of structural members such as beams, columns or shear walls, no visible cracks on the joints of structural members with non-structural members, no visible paint deterioration, etc). The glass facade also remained intact, and the windows and doors of the building were not distorted. On the other hand, the nearby fixed base buildings of the same height exhibited a different behaviour, with multiple reports for non structural damages, such as cracks in the infill walls, especially in the joints of the infill walls with stiff vertical structural members (such as shear walls) as reported in Fig. 9. The water intake/sewage system and the fire protection system remained operative, but the elevator was reported to be temporarily out of service due to the power outage of the whole region that followed immediately after the main shock.
The first three floors of the building are occupied by service/business units which did not report any content affection, malfunction or interruption of activity. No content affection was reported also for the upper floors of the isolated building. On the contrary, in the fixed based buildings near by the content affection was considerable, from mild cases of overturned furniture, to severe cases where content affection caused injuries. However, it should be stated that the most significant difference between the performance of the case of study and the other fixed base buildings near by, was reported in terms of human perception of the seismic event. Thus, people living in the isolated building did not develop panic during the main shock or the following aftershocks, as was the case of people living in fixed based buildings.
Overall, the data gathered suggested that the isolation system was activated during the main shock, while the absence of structural and non-structural damages in the isolated building, the lack of content affection and the reduced human perception of the seismic event, suggested that despite the ground motion amplitude of the main shock in Tirana (around 0.11g) was about 40% of the design earthquake, seismic isolation showed efficiency.

Structural behaviour -Numerical analysis
Despite being a key tool in understanding the structural behaviour of the isolated building, a simple description based on the inspections would not make proper justice to the case. For this reason, in this section the previously described behaviour is quantified in engineering parameters using the commercial finite element software SAP2000 [28].
Since the superstructure is not expected to develop any yielding (due to design criteria as specified in EN 1998 [15]) Fast Non-Linear Analysis (FNA) has been employed to investigate the behaviour of the structure under the influence of the time histories of the main shock recorded in Tirana. FNA is a modal analysis based on Ritz Vectors [28], which enables very efficient non-linear simulations [28,32] provided that the model has: (i.) primarily linear elastic elements, (ii.) a limited number of predefined nonlinear members, (iii.) concentrated non-linear behaviours in link elements. Therefore, a 3D FEM model comprising the following structural elements available in SAP2000 is created: • Frame elements, a line object with two nodes (with six degree of freedom per node), used to model beams and columns, and considering only linear properties of the material. • Shell element, a three or four-node area object (with six degrees of freedom per node), used to model floors and shear walls, with constant thickness and homogenous linear material properties. • Link elements, a two joint connection link, with nonlinear properties used to model the curved surface sliders.
For all the conventional structural members (beams, columns and slabs) the material and cross sections are taken from the official structural detailing plans of the structure. The isolator devices are modelled under the option link/support elements of type 'Friction Isolator' for curved surface sliders with non-linear properties by specifying the following parameters: • The vertical stiffness, evaluated as where N Ed is the ultimate vertical load bearing capacity of the device (Table 1). • The horizontal stiffness for both orthogonal direction evaluated as where N Sd is the vertical load corresponding to the dead load plus combination of non seismic live load(s) according to EN 1990:2002 §A. 1 [33], μ is the fast coefficient of friction as specified in Table 1. • The slow coefficient of friction evaluated as 0.7μ. • The fast coefficient of friction μ. • The rate parameter, which is constant for all the devices (50 s/m as provided by the manufacturer). • The equivalent radius of the sliding curvature (Table 1).
Despite the fact that most of the above parameters are the same for all the devices (Table 1), different positions on the isolation layer inherit different vertical load N Sd from the superstructure and therefore each position would require a specific link element. However, in common practise, and without compromising too much the accuracy of the results, the number of links defined is reduced by grouping devices with similar parameters (in this case with similar N Sd ). In the present case 12 link elements have been defined. Mass source includes the total dead vertical load (which takes into consideration the self weight of structural members and an additional surface load of g = 4.5kN/m 2 accounting for the weight of partitioning walls and layers [34]) and 30% of the vertical live load (defined as a surface load ranging from p = 2.5kN/m 2 in residential units to p = 5kN/m 2 in service/commercial units).
The load case simulating the seismic effect include two horizontal orthogonal (E-W and N-S) components of the recording. Regarding the possible site amplification, as suggested by the specialists of the IGEWE, since the recording is done on the ground level at TIR 1 [10] and the bed rock (the soil classified as Type A soil according to the EN 1998-1:2004 [15]) is relatively close to the surface (around 15m), the TIR 1 recording can be referred as the bed rock accelerogram. On the other hand, the soil characteristics of the site where the isolated building is located are such that the site is classified as Type C soil. Consequently, the site amplification is accounted, based on the provisions of the EN 1998 [15], through the scaling factor S = 1.15.
The reaction of the isolation system is verified through the displacement time histories and the hysteresis cycles of two isolation devices located in the cross axes C'-1' and C-6 (refer to Fig. 4). As it was expected, the numerical analysis confirmed the activation of the isolation system ( Fig. 10). Moreover, the numerical analysis revealed that the horizontal displacements in the isolation system reached a maximum of about 1.5cm at around 10s after the ground motion initiated, and then the amplitudes of relative motion lower as the impact of the ground motion fades. Interestingly, as observed during the on-site inspection, the numerical analysis captures the residual displacement in the North -South direction (for orientation refer to the Fig. 10 Verification of the behaviour of the isolation layer: a) Time Histories of the horizontal displacements of the isolation device located in the cross axes C'-1', b) Horizontal displacements of the isolation device located in the cross axes C'-1', c) Hysteresis cycles of the isolation device located in the cross axes C'-1', d) Time Histories of the horizontal displacements of the isolation device located in the cross axes C-6, e) Horizontal displacements of the isolation device located in the cross axes C-6, f) Hysteresis cycles of the isolation device located in the cross axes C- 6) compass in Fig. 4), in the range of 2-3mm, in the isolation device located in the cross axes C'-1'. Additionally, even though the ground motion recorded in Tirana during the main shock was relatively low (PGA that does not surpass 0.13g when accounting for possible site amplification), the hysteresis loop of the isolation devices depicted in Fig. 10-c & f indicate that the isolation system absorbed the impact and generated sufficient restoring force to revert to the initial position. It should be stated that there is a small divergence between the residual displacements measured on site and those estimated from the numerical analysis. This is to be expected since pendulum isolators are prone to small residual displacement (which does not compromise the integrity of the isolation system) and the magnitude of the residual displacement is also sensitive to the recording of the earthquake [16,17] (different recordings generated different residual displacement of the same isolation system). In the present case, the divergence can be attributed to the usage of a recording which is not measured on site, but in stead, as previously mentioned is done at the TIR 1 Station, 1.82 km away. Moreover, the region was rocked by at least two major aftershocks with magnitude of around 5 [1][2][3][4], which is an additional factor that must be accounted for the observed divergence. However, these results provide concrete evidence over the activation of the isolation system and its ability to generate the necessary restoring force to revert into the initial position.
Following, the behaviour of the superstructure is verified. As it would be expected, the low peak ground acceleration of the main shock in Tirana was sufficient to activate the system, but surely was far from being able to claim the maximum horizontal displacement that the isolation system can accommodate (24cm, which accounts for the safety factor imposed by the EN 1998-1:2004 [15]). However, the superstructure response in this regard, as expected, was almost as if it was a massive rigid block on top of the isolation system. The sway between the top floor and the isolation layer was just below 3cm while the building rises in elevation up to more than 32m ( Fig. 11-a). Most importantly, the numerical analysis revealed that the distribution of the relative floor displacements was such that the interstory drifts 3 , which quantify the structural and nonstructural damages that the structure can develop, are well below the limits set by the European norms for buildings having non-structural elements of brittle materials attached to it ( Fig. 11-b). The very low value of the interstory drifts explains why the building did not develop any structural or non-structural damage in the first place and had little to no content affection at all.
As previously mentioned, the most significant difference between the isolated building and the similar non isolated buildings nearby, was in the perception that the inhabitants had about the seismic event. Figure 11-c depicts the profile of floor accelerations generated in the superstructure during the mainshock. As it is reported in the description of the building, the friction coefficient of the isolation devices is 5.5%, and the literature of seismic isolation clearly states that as the friction coefficient of the isolation system increases, the influence of the higher modes increases, and therefore the profile of the floor accelerations is notably affected [16][17][18]. In the present case, the numerical analysis revealed a slight increase in the floor acceleration of the superstructure, which remains almost constant until the last floor, where it is increased up to 2.4m/s 2 . Moreover, this behaviour can also be attributed to the fact that the building has a significant shrink of the area in the top floor. This shrink renders a considerable reduction of the stiffness of the top floor (some of the perimetral walls of the structure are absent) and therefore, the top floor is more flexible than the rest of the building, resulting in an increase of the floor acceleration. However, despite the above, the profile of floor accelerations is kept well unamplified in elevation, as it would be the case in a fixed base building, which 3 In compliance with the design practice enforced by the local codes [15], the interstory drifts are computed as the difference of the average lateral displacements at the top and bottom of the storey (d s ), divided by the story height (H Story ) and expressed in percentage as follows Interstory Drift (%) = ds H Story 100 explains why the perception of the seismic event was so low compared to other conventionally designed buildings in the vicinity.
Overall, the numerical analysis confirms the activation of the isolation system and quantifies the performance observations done on-site, explaining the absence of damages, the low perception and lack of content affection observed in the building. Despite having been activated and provided protection from the seismic event with the epicentre in 35km from the building, the analysed building falls short in capturing the full efficiency of the seismic isolation approach in the region, since the recorded amplitudes of the ground motion in Tirana are way much lower than the potential that the seismic hazard has in the region. Moreover, as previously mentioned, the building was under construction when it was decided to switch to base isolation and therefore, this structure is not relevant to evaluate the economic feasibility of seismic isolation in the region. For this reason, the following section completes the verification of the technical efficiency of seismic isolation in the region and extends it to a feasibility analysis of the base isolation in the region (Fig. 12).

The efficiency and feasibility of the seismic isolation in the region
As mentioned before, besides the seismic hazard (magnitude and distance to the epicentre and/or fault), other local factors such as soil characterisation, the type of the building and its occupation defines the suitability of applying the seismic isolation. Therefore, to complete the technical efficiency analysis and to verify the economic feasibility of the seismic isolation of buildings in the region, an ordinary six story (one floor underground serving as a basement and parking lot, and five floors above the ground, designed for accommodation units) reinforced concrete building is analysed in both approaches. Initially the structure is designed and detailed according to the rules of capacity design (also referred as fixed based building) whose performance is used as a benchmark, and then the same architectural configuration is designed and detailed using the seismic isolation approach. The building represents a typical and simple solution designed for residential purposes in central Albania and to emphasize the effects of the local conditions of the densely populated area of Tirana, the structure is set in a real construction site. Based on the site investigation, the evaluated design earthquake with a return period of 475 years has a PGA in bed rock a g = 0.27g [9], and the soil characteristics are such that the site is classified as Type C soil according to the provisions of section §3.1.2 in EN 1998-1:2004 [15].
Both structures are designed using a three-dimensional linear analysis through SAP 2000 [28], with the seismic action represented according the provisions of Section §3.2.2 in EN 1998-1:2004 [15]. Following, the design specifications, performance assessment and costs evaluation for initial investments are provided.

Design specifications for the capacity design approach
The structural solution adopted for the non-isolated structure, consists in a dual structural system comprised of a combination of columns and shear walls as vertical primary structural members and a system of light weighted flat slab and perimetral beams stiff enough to ensure the diaphragmatic behaviour of the floors. The structure contains a rigid elevator case, made of reinforced concrete walls, which induces high torsional effects in the structural behaviour, but necessary to give the structure enough stiffness to withstand the design earthquake. To reduce the torsional effects induced by the core, a shear wall with relatively large dimensions (300cm x 30cm) is incorporated in the 5 th axis of the structure (Fig. 13). Also, considerably large columns (cross sections 30cm x 100cm) were incorporated to achieve a better balance.
A Response Spectrum Analysis (RSA 4 ) on a 3D FEM model on SAP2000 [28] is employed to perform the design of the structural members. Columns and beams are modelled as frame objects with elastic properties, while as slabs are modelled as a shell element with constant thickness and homogenous elastic properties. The material properties consist of concrete of strength class C25/30 and reinforcing of class B500c [35]. Mass source includes the total dead vertical load (which accounts for the self weight of the structural members and an additional surface load of g = 4.5kN/m 2 to consider the weight of the partitioning walls and layers [34]) and 30% of the vertical live load (defined as a surface load of p = 2.5kN/m 2 ). As a standard procedure in RSA, the seismic effect is considered through a response spectrum of the design earthquake, which as previously mentioned takes into account seismic hazard, site characterization and the behaviour factor which accounts for the ductility level as defined in EN 1998-1:2004 [15] and shown in Fig. 12 marked with red line the design spectrum (equivalent viscous damping ξ = 5% and behaviour factor q = 3) and black line the elastic spectrum (equivalent viscous damping ξ = 5%) of the design earthquake (a g = 0.27g, Type C soil). All structural members are designed and detailed in compliance with the provisions for the capacity design rules with a behaviour factor of q = 3 (Fig. 12), as enforced in EN 1998 [13,15]. 4 RSA is a statistical linear method that takes into account the contribution from each natural mode of vibration to render the possible maximum seismic response of an elastic structure [28].

Design specifications for the seismic isolation approach
In the case of the seismic isolation approach, the structural solution is simplified since the structural members of the superstructure are not required to partially dissipate the energy of the earthquake (Fig. 14). Instead, the seismic isolation system significantly reduces the impact of the earthquake in the superstructure, thanks to the increase of the fundamental period of vibration to a range where the elastic spectrum is characterised by very low accelerations (Fig. 12) and leaving the superstructure undamaged.

Preliminary design of the isolation system
Since the building is characterised by an underground story (which is not expected to suffer severely from earthquake events), the isolation layer is positioned in the level between the super structure and the underground floor at the relative elevation of −0.5m, to facilitate the installation and most importantly to ensure accessibility for it to be inspected (and/or repaired/replaced should it be needed) [30] after the building has been commissioned. The design procedure, is an iterative one (therefore heavily depends on the experience of the designers) and starts with the selection of the isolation technology that will be used. Based on the experience of the senior authors, for the given level of seismic hazard it is convenient to use Double Concave Surface Sliders (pendulum isolators) [15,29], which would be positioned beneath every vertical structural member, respectively one beneath each column and a sufficient number of devices beneath the shear walls that ensures safe transmission of the vertical load to the ground and guaranties the functionality of the device (since these devices generate their stiffness and reverse force based on friction, they must always be in compressive state). Next, to define the sliding units to be used, a preliminary design is performed for a single degree of freedom system with an equivalent mass to that of the super structure. The latter is derived as the total vertical dead load (the self weight of the assumed structural members and the additional surface loads to account for partitioning walls and layers which are identical to the case of capacity design) and 30% of the vertical live load (which is also the same magnitude as the surface load considered in the capacity design case). Next, the desired fundamental period of vibration and maximum displacement allowed are selected. For the present case the targeted fundamental period for the super structure is T = 3s. This choice is done to shift the fundamental period of vibration by a significant margin compared to the fixed base solution targeting an intersection with the response spectrum of the design earthquake, Fig. 12, in reasonably low accelerations while remaining outside of the resonation range with the ground frequency of vibration [15][16][17]. Moreover, since the space around building does not pose any restrictions and targeting the most convenient solution based on practical experience, an initial value for the effective equivalent viscous damping of ξ = 15% is assumed, and then the expected displacement, d, of the equivalent mass is evaluated as the spectral displacement S De (T ) corresponding to the targeted period of vibration as follows where, S e (T ) is the spectral acceleration corresponding to the target period T , computed from the elastic response spectrum with 15% equivalent viscous damping [15][16][17].
On the other hand, the effective (initially assumed) period of vibration and equivalent viscous damping for the friction double pendulum isolators are evaluated as where g is the gravity constant, R is the equivalent curvature of the isolation system and μ is the equivalent fast friction coefficient of the isolation system. Depending on the experience of the designers, one can assume an equivalent radius and compute the equivalent friction coefficient by accordingly transforming one of the equations from Eq. 4 (or vice versa). In the present case, based on the practical experience of the authors and targeting low friction pendulum isolators (envisaging a system able to be activated even for frequent seismic hazard), the equivalent friction coefficient is chosen μ = 1.76%, and by transforming accordingly the first equation in Eq. 4 and substituting g = 9.81m/s 2 , T eff = T = 3s, d = 0.16m, the equivalent radius is computed as follows For simplicity and to conclude the first iteration, the equivalent radius is accepted as R = 3m. In the next iteration, the equivalent effective damping ξ eff , period T eff and displacement are computed considering R = 3m. After the first iteration, at the end of each iteration, the relative error of the displacement is computed as follows where d i is the displacement at the i-th iteration. According to the provisions of EN1998-1:2004 §10.9.2 [15], the iterative procedure can be concluded if the relative error of displacements between the two consequent iterations is less than 5%, and the computed parameters of the last iteration conclude the preliminary design of the equivalent isolation system [16,17,29]. In the present case the parameters of the equivalent single degree of freedom system are μ = 1.766%, R = 3m, T eff = 3.03s,

Final design and detailing of the isolation system and super and sub structure
Based on the preliminary design, some initial parameters for each device are determined, in order to proceed with the RSA of the design of the entire isolation system and super structure. Thus, all the devices must have the same equivalent radius of curvature, R = 3m, while as the fast friction coefficient of each device is defined in such a way that the modal analysis is optimised (minimal torsional effects in the first two modes of vibration) [15][16][17]29]. Thus, 24 pendulum isolators (one beneath each column and 4 beneath the elevator case) of two types, characterized by different friction coefficient are chosen in such a way that the latter is higher in the devices assumed to be installed beneath the perimetral columns (2.8%) and lower for devices assumed to be installed in some of the internal columns (0.8%) to optimise the performance, Fig. 14, and render an equivalent (weighted) friction coefficient of 1.766%. Next, the super structure and the isolation system are modelled in a 3D FEM model in SAP2000. Since the design is done through a RSA, a linear analysis with identical vertical loads and mass properties as in the case of capacity design approach, not only the super structure (frame and shell elements with linear properties only) but also the isolation system is modelled as a linear element. In this case, SAP2000 [28] provides the option to model link/support elements of type 'Friction Isolator', for pendulum sliders with linear properties only, by specifying an effective stiffness evaluated as follows [29] K e = N Sd 1 R + μ d (7) where N Sd is the vertical load used to model the curved surface evaluated usually as the quasi-permanent load of the structure (evaluated as the dead load plus combination of non seismic live load(s) according to EN 1990:2002 §A.1 [33]), R is the equivalent radius of curvature, d is the preliminary displacement capacity of the isolator and μ is the specific friction coefficient of the device. Since the vertical load and the friction coefficient are different for different devices, multiple link elements are defined. In this case, due to the similarities in vertical load and accounting for the different friction coefficient 6 link elements are defined. Although link elements in SAP2000 provide the option of specifying also the equivalent damping, based on the provisions of EN 1998-1:2004 §10.9.2 (3) [15], that functionality is not used. Instead, the response spectrum of the design earthquake is modified in such a way that for periods of vibration T ≥ 0.8T eff = 2.4s (corresponding to the isolated modes of vibration) the equivalent viscous damping in the linear spectrum is considered that of the isolation system ξ eff = 15%, and for periods T ≤ 0.8T eff = 2.4s (corresponding to higher modes of vibration) the equivalent viscous damping is considered that of usual fixed based super structure ξ eff = 5% (Fig. 12). Additionally, the design norms do not exclude the possibility of using the a behaviour factor 5 q ≤ 1.5, but, for the sake of simplicity and with the aim of "remaining on the safe side", in this work both the superstructure and the substructure are designed and detailed for q = 1 (Table 2). With the FEM model completed as above described, the RSA is performed and the final parameters of the isolation system are defined. The modal analysis confirm the adequate positioning, friction coefficient and equivalent curvature radius of the isolation devices. More specifically, the first two isolated modes return a perfect translational displacement as shown in Table 3, with a mass participation ratio of 97% each. The maximum horizontal displacement from the RSA, as expected [16,17], are higher (17.5cm) than that of the preliminary design (16.6cm), and based on the provisions of section §10.3 2(P) in EN1998-1:2004 [15], the ultimate displacement capacity of each device is magnified by a safety factor of γ x = 1.2. The vertical capacity of the sliders is defined and verified in compliance with the codes provisions described in EN 1998-1:2004 and EN 15129 [15,29]. Thus, the isolating devices must have a vertical load bearing capacity that is able to resist the maximum combination of loads (both static and seismic combinations) as defined in EN 1990in EN :2002. In the present case, since we have selected two types of devices, the vertical load that they will be able to bear will be defined by the position that requires the maximum capacity. Eventually, this design practice along a rigorous Factory Production Control test [29], fit the isolation with an increased reliability and, in the unlikely event that one device fails, the over-strength in vertical load and horizontal displacement capacity of the isolation system avoid a global collapse of the structure (similar with the redundancy for a fixed base structural solution in the capacity design approach). Finally, the design procedure of the isolation system is concluded and consists of 10 units FIP-D of Type 1 and 14 units FIP-D Type 2, with their main parameters summarized in Table 2.
Next, the design of the super structure is completed. Given the significant reduction of the seismic forces, as previously mentioned, the structural solution for the superstructure results into a much simpler solution. Moreover, since it is detailed to work in linear range only [15] the construction process would be much easier to be delivered fast and according to the standards [13]. Particular attention is posed to the layer above the isolation system, since it will serve as an intermediate support of the super structure to the isolation system, and therefore it must be sufficiently stiff to enable a safe transmission of the loads. Similarly, the columns beneath the isolation layer are designed with the sufficient stiffness to ensure the stability and safe support of the superstructure. Furthermore, since it is assumed that the elevator serves to the underground floor as well, the elevator room is isolated in such a way that it permits the movement of the elevator along with the superstructure (Fig. 15).

Structural behaviour comparison -Linear analysis
As it would be expected, the most eminent distinction between the fixed base solution and the isolated one stands in the principal modal parameters, as reported in Table 3. The increment of the fundamental period of vibration in the isolated structure enables the reduction of the corresponding elastic spectral acceleration from S a (T = 0.71s, ξ = 5%) = 6.55m/s 2 to S a (T = 3.03s, ξ = 15%) = 0.72m/s 2 which yields a proportional reduction of the shear force induced in the structure. Therefore, the maximum design shear force, considering the different behaviour factors (q = 3 for the fixed base solution and q = 1 for the base isolation solution) results respectively around 5200kN for the fixed base solution and 2400kN for the base isolated solution (Fig. 16-b). Consequently, the reduced shear force allows for the reduction of the cross section and the rebar area of the structural members ( Fig. 16-a), which translates not only into a reduction of the costs of the superstructure but also in an increase of the property value, since the actual useful space in the building is increased. Moreover, the modal analysis reveals that the isolation system allows for almost a perfect balancing of the structure, avoiding completely torsional effects, which further facilitates the design, detailing and delivering of the superstructure of the isolated building. Additionally, provided that the design is conducted correctly, it can be stated (based also on the reported superior performance of the isolation buildings under real earthquake events [21,22,26,27]) that seismic isolation approach is more reliable compared to the fixed base approach. That is because it is easier to deliver one critical component such as the isolation layer, rather than having multiple critical components or structural members distributed in elevation such as in the case of capacity design approach. Moreover, installation of the isolation system is usually done by highly qualified workmanship and bringing to attention that the design provisions [15] are based on an over-strength approach in conjunction with factory production control testing foreseen in [29], periodic inspection as foreseen in [30], increase the reliability of base isolation. On the other hand, as it is known in literature [13], it is much more difficult and complicated to detail structural members with the ability to develop controlled plastic deformations, and is often prone to errors due to unqualified workmanship, especially in developing countries [4,5] As it is well known, the shift in the vibration period enabled by the isolation system, yields relatively large horizontal displacement that must be accommodated in the isolation layer. Therefore, proper considerations must be done to ensure the relative motion of the superstructure (accounting also for the safety factor recommended by EN 1998 [15]). That includes design considerations not just for the isolation system, but also for the gap around the building and the elevator case, as well as the flexibility of water supply, sewage, gas, and electricity services. Additionally, it is critical that the gap around the building remains free and unblocked throughout the entire life of the building, which is why it is usually recommended that the periodic checking to be conducted not only for the isolation system but also for the gap [30]. Figure 17-a depicts the profile of horizontal displacements in elevation of the structure, where it can be seen that the decoupling of the superstructure is achieved, given that the sway of the top floor to the isolation interface is less than 0.15%, while in the case of the fixed base solution, the horizontal displacements are lower in their maximum value but distributed in elevation which yields larger interstory drifts. And indeed, as depicted in Fig. 17-b, the interstory drifts in the case of the base isolation solution are up to 4 times lower than the case with fixed base solution  [15] for limitation of the interstory drifts in buildings having nonstructural elements of brittle materials attached to it). This ensures that the superstructure in the case of the seismic isolation approach remains completely unaffected, and the building remains in the operational phase even during a strong earthquake. Moreover, the isolation system can bear multiple cycles of design earthquakes, unlike the capacity design approach which is expected to develop structural and non-structural damages which might result to be nonfeasible to repair (in any case, provided that the design is done correctly, the lives of the occupants must remain safe).

Verification of structural behaviour through non-linear analysis
As it is well known in literature [15][16][17][18], for design purposes, provided that the non-linearity of the isolation system is low enough to make the influence of the higher modes of vibration neglectable in the structural response of the building, a simple linear analysis as previously described is sufficient to properly design through the isolation approach. While, if this condition is not met, EN1998-1:2004 §10.9.5 (1)P [15] requires the verification of the behaviour of the structure through dynamic nonlinear analysis. In the present case, the linearity conditions posed by EN 1998 are satisfied, and therefore it would be sufficient to rely on linear analysis only. However, to better assess the efficiency of seismic isolation and accounting for real recorded strong motion signals in the region, the performance of the isolated building is verified through Fast Non-linear Analysis carried out through the finite element software SAP2000 [28]. For this reason, firstly the 3D FEM model used in the previous section is modified and adopted to conduct a FNA, by introducing non-linear properties to the link elements modelling the isolation devices specified in Table 2 (the parameters of the non-linear link elements are defined in the same way as described in "Structural behaviour -Numerical analysis" section). Then, the model is exposed to two earthquake scenarios (accounting for the horizontal components only), assuming the recording in TIR 1 as the case when the building is exposed to a frequent earthquake (given the relatively low PGA of the signal recorded in Tirana) and the recording in DURR station as the case when the building would have been in the same site characteristics but exposed to a stronger ground motion. It is worth to bring into attention that the signal in Durrës was partially recorded due to a power outage 17s after the recording started ( Fig. 3-b). Although the specialists of the IGEWE ensured that the maximum peak ground acceleration is already captured in the real recording, it is worth bringing to attention that systems with long period of vibration might reach their maximum response long after the PGA is reached. Figure 18 depicts the behaviour of the isolation system (verified by checking the behaviour of the isolation device located under the column in cross axes A-2) under the signal recorded in the Tirana station (TIR 1) and the one in Durrës (DURR). The displacement time histories, as in the case of the existing isolated building in Tirana, revealed that even for small impacts, the isolation system would activate and therefore provide protection. Similarly, some residual displacements in the range of 1.5mm are observed in through numerical analysis when considering the record from TIR 1. The recording obtained from the DURR station imposes a much greater impact in the isolation system, in particular in the N-S direction. Thus, even though the peak ground acceleration of the East-West Component of the DURR record is only 0.125g, the nature of the frequencies is such that it triggers a relative displacement of the superstructure that reaches up to 5 cm in the corresponding direction. While the North-South component of the DURR recording (that has a peak ground acceleration of about 0.195g) induces a much greater impact in the isolation system, which reaches a maximal relative displacement that goes up to 13.5cm. It is worth noting that the response spectrum of the N-S component recorded in Durrës on 26 th of November of 2019, almost converges (in the range of high periods) with the response spectrum of the design earthquake used in the present case, while as the E-W component inflicts almost half of the seismic excitation (reported with orange continuous and dashed lines in Fig. 12). Such a difference is clearly captured also by the hysteresis cycles depicted in Fig. 18-f, which on the other hand emphasises the adequate behaviour of the isolation system in absorbing the seismic impact. Further, the non-linear analysis reveals that for all the components considered, the superstructure exhibits relative displacements almost as a single rigid block on top of the isolation layer ( Fig. 19-a, as it was predicted also from the linear analysis) and sustaining very small interstory drifts ( Fig. 19-b). Additionally, similarly as in the case of the existing building in Tirana analysed in section 3 of this work, it is evident that the isolation system prohibits the amplification of the floor accelerations as it would be the case of a fixed based building. Also, it can be noted, that for the small amplitude components, the floor accelerations are not significantly reduced in value (nevertheless as stated it remains almost constant), while in the case of the N-S component of the main shock recorded in Durrës during 26 th of November, the actual floor accelerations induced in the building are reduced by 50% in their max value (from 1.9m/s 2 on the ground to 1m/s 2 on the top floor of the building).
Overall, numerical analysis on the case of study presented in this section, confirm the efficiency of seismic isolation in the region, which can provide protection not only for large amplitude expected seismic events, but is able to activate and provide an enhanced structural performance even for typical small amplitude seismic events of the region.

Cost evaluation of the proposed solutions for the case of study
The previous sections provides an insight over the efficiency and enhancement of structural behaviour enabled by seismic isolation (realized through pendulum isolators) in the densely populated area of central Albania. Nevertheless, especially in developing countries, economic feasibility becomes the decisive factor on the way how a building is finally delivered. For this reason, the present section provides an estimation of the differences in building materials and cost of the isolation system for the case of study so far investigated, for both proposed solutions, the fixed base approach, and the base isolation one, as shown in Table 4.
Thanks to the reduction of the shear force induced in the super structure, the structural solution based on seismic isolation allows for the reduction of the cross sections and the rebar area in both primary and secondary structural members ( Fig. 16-a). Additionally, releasing the superstructure from any "duty" of energy dissipation allows for a simple detailing of its structural members, as in the case of non-seismic areas EN 1992 [35]. This gives the opportunity for some more gaining's in the workmanship costs and can facilitate and speed up the construction progress, which translates in an increment of the property value and consequently will increase the profit of the developer. Moreover, this approach reduces the uncertainties associated with the proper detailing of   [13] as in the case of the capacity design approach (which is reported as an important role in considerable damages developed in the newly constructed building in the city of Durrës during the main shock [4,5]). But it must be pointed out that concentrating the seismic resisting component of the building in one single layer requires a careful and accurate design, qualified manufacturing and testing of isolation devices and appropriate installation of the devices onsite. However, it is easier to deliver properly one critical component of the building rather than having them distributed in multiple critical components as in the case of capacity design approach. Considering all the above, in terms of construction material and workmanship savings (accounting also for the specific type of piping and other infrastructure adjustments to ensure the "decoupling" of the superstructure from the ground/substructure), the seismic isolation approach in the region saves about 15,000 Euro (Table 4) in the superstructure of the case of study.
On the other hand, the cost of the isolation system, including the mandatory factory production control tests required by the European standard EN 15129:2009 [29], in this case is higher than the savings generated in construction materials (43,000 Euro 6 −14, 773.4 Euro = +28, 266.6 Euro), as reported in Table 4. Considering all the above, the cost of the initial investment for a seismically isolated residential building in central Albania is about 2.5% higher than the final cost for the same building delivered according of the principles of the capacity design (it should be stated that the final cost includes the overall cost of construction and not just the cost of construction materials and isolation system reported in Table 4). However, given the performance enhancement, increased flexibility in architectural design, a reduction in the construction time, the increased useful space in the superstructure and the overall increased property value suggests that the seismic isolation is fairly feasible in the region. Moreover, it is worth 6 Note that the price for the isolation system includes the shipping costs. Additionally, the price is evaluated by filtering the influence of extraordinary spike in steel prices due to reduced production caused by the pandemic and the 'shock' caused by the war in Ukraine, aiming to provide a more realistic evaluation in the long term. reminding that these two different approaches target two different performance objectives, as the fixed base building is expected to yield in the range of life safety while base isolation approach renders a building that remains in the elastic range and consequently in operational phase during and after the design earthquake, for multiple cycles of design earthquake. Considering the latter, seismic isolation is far more superior compared to the conventional design approach in the region.

Conclusions
This paper provided an insight over the structural behaviour of the only seismically isolated building in Albania during the Durrës sequence of 2019 and exploited the technical efficiency and economic feasibility of the base isolation (realized through pendulum isolators) in the region. Even though the amplitude of ground motion recorded in Tirana was relatively low, the existing isolated building exhibited a significantly enhanced behaviour compared to similar nearby fixed base buildings, especially in terms of content affection and human perception of the seismic event. Numerical analysis conducted confirmed the activation of the isolation system, quantified and explained through engineering parameters the observed behaviour on-site. This is an important evidence that seismic isolation can provide protection even for small amplitude ground motion in the region.
The efficiency of seismic isolation in the region was further exploited by designing and detailing an ordinary six story building designed for residential purposes in both approaches and considering all the local factors such as seismic hazard, site characterisation, typical architecture, and the typology of the building. The behaviour of the fixed base solution, as the typical anti seismic solution used in the area, was used as benchmark to evaluate the efficiency of base isolation using both, linear and non-linear analysis. As expected, the numerical analysis revealed a significant enhancement in structural performance enabled by seismic isolation. Additionally, emphasises were put to proper measures that must be taken to ensure the functionality of the isolation system throughout the entire life of the building, which even though are minimal (a clear sufficient gap around the building) remain critical.
Finally, in the feasibility study the cost of construction materials; cost of workmanship; cost for manufacturing, mandatory testing, shipping, and installing the seismic isolators and cost of specific infrastructure that allows for the decoupling of the superstructure were considered. The study revealed that the savings enabled by the reduction of the impact that the earthquake has in the superstructure, partially compensate the cost for the initial investment. However, the reduced cross sections in the superstructure increases the useful surface of the building which along with the ability to bear multiple cycles of design earthquake without sustaining any damage, increase the value of property. Overall, the findings in this study suggest that seismic isolation in the region represents an effective and feasible alternative to provide adequate seismic protection of buildings in the region.

Declarations
Competing interests M.G. Castellano is an employee at FIP MEC srl, a company that manufactures and sells structural devices, including seismic isolators.
N. Hima is a PhD student of the 35th cycle of Civil, Environmental and Mechanical Engineering at the University of Trento, Italy, and in compliance with the provisions of the Marie Sklodowska-Curie grant agreement 'INSPIRE -Innovative ground interface concepts for structure protection' PITN-GA-2019-813424-INSPIRE, his research work is being conducted in collaboration with FIP MEC srl, a company that manufactures and sells structural devices, including seismic isolators.
E. Ç aushi & L. Murtaj declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/.