This Section presents field observations of a few relevant case studies classified according to the representative typologies of the Albanian building stock discussed in Sect. 4, as well as historic buildings. Their geometric and structural features are described in detail to provide an overview of the construction practice. Damage patterns and severity are discussed and critically interpreted to highlight the strength and deficiencies of the structural and non-structural elements of the inspected constructions and identify the causes associated to the good and poor seismic performance observed in field. The following considerations and discussion on the damage scenarios are based on the visual inspection of the case studies while a detailed analysis (i.e., the use of advanced numerical models) is beyond the scope of the present paper.
Single-family rural houses
Single-family rural houses represent a widespread typology in Albania. During the mission this typology have been widely inspected considering several case studies in the village of Bubq in Krujë municipality. Some of these houses, according to information gained from local engineers, were already damaged by the earthquake in September 2019, and then collapsed in the seismic event on the 26th November. No casualties were reported in this area, although many of these structures showed a poor seismic response mainly caused by an inadequate seismic design and unauthorised interventions carried out by local artisans with little input from engineers.
The house in Fig. 9a, located in Bubq, has the typical configuration of the single-family rural houses in this area. It is a 2-storey house, classified as URM building. The 1st storey was built in 1992 by the owner by using bearing walls made of hollow concrete blocks and cement mortar (Fig. 9b) and a solid RC slab. In 1997, a portico porch made of RC columns was added, and a 2nd storey was built supported by the 1st storey and the columns of the portico porch. The bearing walls of the 2nd storey are also made of concrete blocks while the floor is made with a hollow core prefabricated concrete slab (Fig. 9c) supporting a timber truss for the roof tiles (Fig. 9a). The house, which was already damaged by the earthquake in September, completely collapsed during the following earthquake in November. The observed failure is caused by the collapse of the portico porch, which failed for soft storey due to the uneven mass distribution between the 1st and 2nd level. The rear view of the house in Fig. 9c shows a severe crack in the pier at the 1st level, due to a torsional effect in the building caused by the collapse of the portico porch. Detailing of the RC columns, supporting part of the 2nd storey, are reported in Fig. 9e, where it is possible to see the joint failure of the top end of the column. Figure 9f shows a close-up of a different joint highlighting the lack of adequate reinforcements with smooth and corroded rebars.
Similar failures and cracks were also observed in other single-family houses of the same typology with floor plans which differ in sizes. Local engineers confirmed that the 1st and the 2nd levels of these inspected houses were also built at different time, using different materials and structural floors for the 1st and 2nd level. Figure 10 shows the failure of a single-family house with a geometry plan smaller than the ones in Fig. 9. The observed damage confirms that the failure is triggered by the joint failure of columns in the portico porch, as highlighted by the horizontal cracks on the top end of the 1st storey columns and overturning of the side façade.
Multi-storey RC buildings
Among the different RC building typologies that can be found in Albania, the mission focused mainly on RC multi-storey buildings. This is justified by the following: (1) relevance, i.e., many multi-storey RC buildings are present in the Albanian territory and in many of the areas significantly affected by the earthquake; (2) such mid-rise structures were consistently damaged while low-rise RC buildings experienced none or very small damage; (3) these buildings have been designed according to the design standards and therefore, the observed damage allows a critical discussion of the code. This is not always the case with other types of low-rise RC structures that were designed outside the code or, in some cases, built by the owner even without a design; (4) these structures are characterised by some ‘unexpected’ design characteristics which make them different from typical multi-storey buildings located in other seismic areas in Europe.
The Albanian KTP-N.2-89 code has been already illustrated and discussed in Sect. 3, however it is worth highlighting a few aspects that are relevant for RC multi-storey buildings. In particular:
The strength requirements of the KTP-N.2-89 are lower than those of the Eurocode 8. Figure 11a shows a comparison of both the elastic and design spectra defined according to the KTP-N.2-89 and Eurocode 8. The example refers to a structure located in an area with PGA = 0.32 g, soil category Type II, response factor ψ = 0.25 (corresponding to a behaviour factor q = 4 in EC8) and a fundamental period T1 = 1.0 s. The corresponding design spectral acceleration are equal to Sa(T1) = 0.064 g for the KTP-N.2-89 and Sa(T1) = 0.12 g for the Eurocode 8;
Ductility requirements are included in the Albanian code and, despite less detailed, are similar to those of modern seismic design codes. In fact, the KTP-N.2-89 requires: (1) regularity checks, e.g., uniform masses and stiffness, symmetry, simplicity, etc.; (2) ductile member detailing, e.g., min longitudinal reinforcements, maximum stirrups spacing, etc.; (3) implicit capacity design, e.g., strong columns-weak beams;
Absence of stiffness requirements i.e., no considerations regarding Damage Limit State (DLS) checks. This aspect could be considered the most evident difference between the KTP-N.2-89 and the Eurocode 8 and is the main one that, in this occasion, affected the seismic response of multi-storey RC buildings. This aspect shows that insufficient consideration is made to the damage of the building under low-intensity (i.e., frequent) earthquakes, which is generally concentrated in the infill panels (for the considered building typology). However, attention is paid to the out-of-plane behaviour of the infills by the introduction of belt beams as shown in Fig. 11b.
The case study investigated is a typical 2000s multi-storey RC building located in Durrës just across Niko Dovana Stadium. Many buildings with similar characteristics have shown the same damage pattern described here. The structure is characterised by nine storeys with a constant inter-storey height approximately equal to 3.20 m and by 5 × 4 bays with a constant span of ~ 5 m. The structure is regular in plan and elevation and is illustrated in Fig. 12a. It is worth mentioning that, in this structure, no core wall is included and that the horizontal forces are resisted by the frames only. Based on the several observed buildings of this typology, this has been identified as a common situation where the elevators RC core is absent. The building has shops at the ground level while the upper stories is for residential use.
The structure is characterised by internal and external columns with large dimensions approximately equal to 80 × 80 cm and 110 × 35 cm, respectively. The dimensions of the external column can be observed in Fig. 12b, due to the formation of the vertical cracks in the plaster. Figure 12b shows also the cracks corresponding to the position of the beams which have a depth equal to the thickness of the floor slab (approximately 30 cm). This structural configuration of strong columns-weak beams is most probably the outcome of low design accelerations and lack of stiffness requirements.
The infilled panels are made by large hollow bricks (25 × 25 × 20 cm), with horizontal holes as shown in Fig. 12c. The interaction of the flexible structure with the stiff infills led to significant non-structural damage. In fact, infill panels in these structures are often characterised by shear cracks. Moreover, some of the unconfined panels have collapsed (Fig. 12b), or experienced heavy damage. In addition, as shown in Fig. 12d, the damage pattern is also characterised by horizontal cracks at floor level in the lower side of the floor slabs. This is related to the lack of construction detailing and a poor connection between the top side of the infill panels and floor slabs. This configuration promotes the detachment of the infills from the beams and this damage pattern is distributed over the first five storeys of the building. Figure 13 shows a closeup of the damage pattern of the eastern façade of the building which highlights the significant vulnerability of this type of stiff infill panels, especially when unconfined.
It is worth mentioning that, despite the KTP-N.2-89 is currently enforced in Albania, for more recent designs (i.e., after 2010), it became a common practice among design engineers to use recommendations from the Eurocode 8 to overcome the lack of detaining in the recommendations of the Albanian code. For example, although the KTP-N.2-89 code requires that the columns must be stronger that the adjacent beams, it does not provide a quantitative formulation to perform this hierarchy of strength requirement and in these cases, the detailed formulations provided in Eurocode 8 are usually adopted by practitioners. However, Eurocode 8 is not adopted in its entirety, i.e., recommendations in the Eurocode 8 that are not also present in the KTP-N.2-89 are not applied. For example, it is unlikely that the displacement-based checks for the damage limit state are applied. This insight is based on personal communication with local engineers. In order to investigate this aspect of the design, more recent buildings (i.e., after 2010) designed according to KTP-N.2-89 while including integrations from Eurocode 8 have been also widely investigated.
As per the case study described, also in newer buildings, the infilled panels are generally composed of large hollow bricks (25 × 25 × 20 cm) and are used both for the external and the internal walls. Consistently with the previous case study, significant non-structural damage was observed due to the interaction of the flexible structure with the stiff infills for both unconfined external and internal partitions. Clearly, the lack of confinement significantly affected the seismic response of the non-structural components, leading to shear cracks but also to out-of-plane failure of the infills. This is also highlighted in Fig. 14, showing how the belt beams for the infills are ineffective when used in unconfined frames. It is worth mentioning that many buildings with similar characteristics showed the same damage pattern.
The observed damage in these recent 2010 + multi-storey RC buildings, despite less extensive, is very similar to the one described for the 2000s structures. The damage pattern highlights a significant need for additional design requirements to be included within the KTP-N.2-89 such as the inclusion of Damage Limit States checks and the need to avoid unconfined infills. However, it is noteworthy that the Damage Limit States requirements of the Eurocode 8 have demonstrated to be ineffective in a number of occasions e.g., De Luca et al. (2018), thus more strict limits should be used to ensure the design prescribed behaviour of the infills.
Among those investigated, very few RC buildings showed structural damage. It is important to highlight that no reference is herein made to the collapsed buildings. Indeed, such buildings were already demolished at the time of the mission and no considerations are reported here due to lack of direct observations. Figure 15a, b show respectively the shear and axial load failure of one external and one internal columns of a RC multi-storey building in Durrës. Although the year of construction is unknown, it is evident that the quality of both materials and structural details is very poor for this case study. Figure 15 shows the presence of weak concrete (possibly with low percentages of cement), smooth aggregates, smooth longitudinal bars, small stirrups missing the 135° hook and showing a particularly large spacing. Based on this information, it is fair to say that likely this building was designed and/or constructed without abiding the code and with poor construction materials and execution.
Pre-1990 prefabricated large-panel buildings
As many other countries in Eastern Europe, in the 1960–1970s Albania responded to the growing demand for new houses utilising the emerging trends for industrialization of the construction process and mass construction of prefabricated residential buildings (see Fig. 5d) based on large-panel prefabricated RC elements as illustrated in Fig. 16. These buildings were built according to standardised templates approved by the Albanian governmental authorities and hence, they represent a widespread standardised technology in Albania. During the 1970s large-panel buildings spread throughout the country and become the main type of construction in the Albanian cities such as Shkodër, Tirana, Durrës, Lushnjë, Burrel, Elbasan, Berat, Pogradeci, Laç, Lezhë, Korçë, Tepelenë, Gjirokastër (Abazaj 2019). Most of these buildings have five or six storeys and comprise different modules, the number of which depended on the urban project.
Under earthquake loading the prefabricated wall behaves as one structural unit composed of interacting wall elements as shown in Fig. 16c, d. This structural interaction within the wall needs to be secured by structural connections that resist the required shear, tensile and compressive forces. If the strength of the horizontal and vertical joints exceeds the forces in the interface between the panels, the prefabricated panel wall will have monolithic behaviour under lateral load as shown in Fig. 16e, and the structural damage will be concentrated in the form of diagonal cracks in the panels. However, in prefabricated concrete wall systems there may be significant slip (shear displacements) along the vertical and horizontal joints. Figure 16e also shows horizontal actions applied to a large-panel cantilever wall in which slip has occurred along both vertical and horizontal joints due to shear forces transferred along these joints. The effect of slip along the horizontal and vertical joints is to reduce the stiffness of the system. It is difficult to say which will be the prevailing seismic response and failure mode of large-panel walls under earthquake loads. The observations in past earthquakes in Romania, Armenia and Bulgaria suggest that slippage between panels is likely to occur since most of the reported damage was in the form of horizontal and vertical cracks in the interfaces between panels. This situation with ‘strong panels—weak joints’ is also related to the high-quality control of the prefabricated elements which provides the prefabricated panels with a high concrete resistance. However, the situation with ‘strong joints—weak panels’ cannot be excluded and should be checked during the assessment procedures. In addition, in the most common situations with ‘strong panels—weak joints’, the slippage and rocking between panels, is likely to provide a significant friction damping which is beneficial for the seismic performance of these structures and could be one of the reasons for the good seismic performance of large-panel buildings observed in past earthquakes.
The team visited two neighbourhoods with large-panel buildings, in Laç and Durrës. All buildings seemed to have poor maintenance and many signs of deterioration were visible on the facades. However, there were no external signs of earthquake-induced damage in any of the buildings. An inspection from inside showed minor cracking in the interface between the panels in the first two floors—vertical cracks in the contact zone between two wall panels, and horizontal cracks in the contact zone between slab and wall panels. These are typical crack patterns in large-panel buildings in the onset of structural damage when the damage is mainly in the form of cracking of the grouting between the panels and does not affect the structural safety. Cracks in the wall panels and damage in the dowels were not observed in any of the inspected large-panel buildings in Laç and Durrës.
Pre-1990 masonry buildings
In Albania, masonry buildings were widely used for residential and public purposes between 1945 and 1990, when the Albanian building design codes underwent several changes, as discussed in Sect. 3. The first masonry buildings were 1 or 2 storeys and constructed using different types of masonry such as stone, and clay bricks. Buildings from 1945 to 1963 were mostly constructed based on experience of engineers and simplified calculations. The first standardised design template was approved by the Albanian governmental authorities in 1949 for a 2-storey adobe building. Successively, a large variety of standardised design templates for buildings of 3 to 5 masonry storeys were implemented between 1963 and 1978. After the 1979 earthquake near Shkodra, many 5-storey masonry buildings were severely damaged and therefore the code was revised with the publication of KTP-N.2-89 which is the current seismic design code in Albania (Bilgin and Hysenlliu 2020). As reported in Sect. 4.1, masonry buildings together with RC buildings are still in use and represent a high percentage in the residential building stock of the country. These buildings can be classified in two different typologies: (1) URM buildings with load bearing walls as shown in Fig. 17a; (2) CM buildings made of load bearing walls confined with RC tie-elements as shown in Fig. 17b.
Typical masonry buildings are built as isolated constructions with regular plan and regular elevations. The majority of these buildings have a rectangular plan or irregular plan geometry with symmetrical distribution of the load bearing walls. Their plan can vary in size, while the floor height is generally 2.8 m. Openings have regular layout and concrete lintels over the openings are commonly used to transfer gravitational loads. The floor systems are either RC slabs or hollow core prefabricated concrete panels. The former is commonly a 150 mm thick slab, and the latter consists of a 220 mm thick slab on top of prestressed RC joints. The wall thickness is 380 mm in the first two storeys and 250 mm for the remaining ones, while partition walls have thickness of 120 mm. These are made of solid fired clay bricks with dimensions 250 × 125 × 60 mm or silicate bricks with dimensions 250 × 125 × 65 mm. Bricks are bonded using cement or silicate mortar. The typical masonry types observed on site are reported in Fig. 18a–c. Tie-elements in the CM buildings (see Fig. 18d, e) consist of RC columns with typical cross section of 380 × 380 mm and RC beams with depth of 250 mm or 380 mm, depending on the wall thickness at their locations. The concrete class is typically C16, or lower, while the typical steel grade is S220 (Bilgin and Korini 2012).
During the survey it was observed that both URM and CM buildings made of fired clay bricks and RC slabs showed a good seismic performance. Severe damage was not observed, and buildings were not tagged as (red) unsafe in these cases. Amongst others, the good interlocking of the bearing walls in URM buildings was highlighted by the absence of cracks along the façade edges, which is typical of masonry buildings with poor connections among adjacent walls. In addition, the good seismic response is also associated to the typical box behaviour which occur thanks to the good connection and stiffness of the RC slabs distributing the lateral loads on the resisting walls. In the CM buildings, a RC system embraces the masonry walls with frame elements. Their good seismic performance is well known and was observed in various past seismic events (Brzev et al. 2010; Borah et al. 2019). Their capacity is provided by the efficient interactions between the masonry walls and the confining elements given by the shrinkage in the concrete casted after the construction of the walls. These effective connections ensure that CM typologies behave as a whole up structure which is capable to sustain larger deformation and allow high strength and ductility.
The major deficiencies observed for these buildings are related to irregularities in elevations due to unauthorised interventions such as the closure and creation of new openings, use of different masonry for reparation, and introduction of additional floors carried out by the owners of the single dwellings as showed respectively in Fig. 19a–c. These alterations, together with the lack of maintenance, degradation of the mechanical properties of the material, and change of the load paths, can become the origin of damage and deteriorations, which may increase the seismic vulnerability of the structure over its lifetime.
Despite the good performance observed for these buildings made of fired bricks and RC slabs, structural failures were reported for three URM (one 3-storey and two 5-storey) buildings made of solid silicate bricks and hollow core prefabricated concrete slabs, located in Thumanë. The causes of collapse for these buildings are associated with the poor mechanical properties of the silicate bricks and the lack of connections between the prefabricated concrete slabs and the bearing walls failing in overturning. Inspections in Thumanë were carried out after the buildings were already demolished, therefore information related to these failures were provided by local engineers.
URM buildings made of silicate bricks with hollow core prefabricated concrete slabs were also inspected in the city of Lezhë, located in the northwest of Albania, about 50 km from the epicentre. In this area, only a few buildings were (red) tagged as unsafe by local engineers, although the damage consisted of non-structural damage on external bearing walls and light shear diagonal and X-shape cracks on internal bearing walls.
More severe damage was observed on buildings of the same material and floor types which had undergone large interventions carried out with an inadequate seismic design. Figure 20 reports the damage due to the extension of a 5-storey URM building built in 1974 (green plan in Fig. 20a) with a 4-storey RC building built in 1980 (red plan in Fig. 20a). The different stiffness the URM and RC buildings caused severe pounding damage visible on the external bearing walls (Fig. 20b), the cracks along the opened gap between buildings observed during the internal inspection (Fig. 20c), and the detachment of the staircase due to torsion (Fig. 20d).
A similar seismic performance was also noted for a few CM buildings with silicate bricks and with hollow core prefabricated concrete slabs inspected in Laç (see Fig. 1). The inspection revealed that these buildings (red) tagged as unsafe by local engineers, were already extremely deteriorated before the earthquake due to a lack of maintenance, highlighted by the presence of corroded and exposed reinforcements, crushed brick units and loss of painting. Furthermore, as these buildings had undergone several structural modifications carried out over time, it is likely that their severe damage is the result of a load path alteration due to construction of additional floors, creation of large openings at the ground level. Typical damage observed for these building, such as the displacement of the prefabricated panels of the prefabricated concrete slabs, yielding of the corrected reinforcements and shear failures in the load bearing walls, are shown in Fig. 21.
Albania has a rich history and a large presence of built heritage around the country. During the field mission, the castles of Krujë, Prezë and Durrës were inspected to investigate the seismic impact of the 26th November 2019 earthquake. Severe damage and partial collapse were observed on these castles and, in the following sections, the vulnerability of their structural components and the causes of the observed damage patterns are identified through the evaluation of their structural conditions before and after the earthquake.
The Castle of Durrës was built in the first century BCE and acquired its final form in the fifth century with the Byzantine emperor Anastasius I Dicorus. The fortification walls were devastated in an earthquake in 1273 and had to be extensively repaired (https://en.wikipedia.org/wiki/Durr%C3%ABs_Castle). Under the Republic of Venice, the castle was reinforced with several guard towers and the walls were reinforced during the Ottoman Empire. Signs of local reconstruction of the masonry walls and repair of the corners with visible metallic elements on the exterior sides are visible on the towers of Durrës in Fig. 22.
In Fig. 23a, the view of the tower after the earthquake shows its seismic response. Based on observation, it is possible to identify that the different masonry types adopted for the wall repair are possibly associated to several interventions performed over time. The North façade (N) shows a stone masonry base course supporting brick masonry walls, changing from brick to stone masonry towards the top. On the contrary, the West façade (W) appears to be made of quite uniform stone masonry until the merlon, which shows a different type of stone masonry. Local repair of stones with bricks is also visible on the wall. The stone masonry at the merlons appears to be in poor condition, with evident decay and loss of mortar. The failure on the N façade is showing that the walls made of different types of masonry in the corner rely on a poor interlocking, as confirmed by the vertical and regular cracks on the wall edges pointing out an out-of-plane mechanism.
Furthermore, it was observed that masonry had failed in blocks, with only a small amount of material resulting disintegrated. This is due to the presence of an external layer added to the existing masonry walls without transversal connectors, strengthening techniques practiced during the Ottoman period (Nicolle 2010). This failure type of masonry walls collapsing in blocks is not common in masonry constructions and can be explained by the presence of mortar joints which are thicker than the single masonry brick units (see Fig. 23b). It is unknown whether the brick masonry is part of the original wall or it is a more recent alteration. However, this type of construction with thick mortar joints is quite common in the Byzantine brick masonries, in construction located in seismic areas and/or on subsidizing soils (Binda et al. 1999). Although it was not possible to access to the interior of the tower, it was possible to observe an extensive presence of vegetation growing on the interior side of the walls. The presence of vegetation is sign of both lack of roofing and lack of maintenance, which could be triggers for developing weakness in the masonry walls leading to the observed failure.
The Castle of Krujë is located on a hilltop overlooking Krujë town and is the symbol of Skanderbeg’s rebellion against the Ottoman Empire and considered the most significant expressions of medieval constructions in Albania. The castle is surrounded by fortification walls and sits on a rocky substructure formed by different blocks of rocks fallen off the mountain that constitute the base of the castle. The concerns related to the geological stability of the rocky hill (IRPP/SAAH 2006) led to the installation of anchors (their locations is pointed out in Fig. 24a) used to reinforce the hill’s side near the clock tower.
The clock tower is an unreinforced stone masonry structure dating from the twelfth century, with a quasi-pyramidal shape at the bottom and a squared geometry at the top. Access to the tower is understood to be on the front and rear sides from two doors located at different levels, connected through a timber staircase allowing the entry to the top floor. The clock tower underwent a series of interventions: the first intervention took place in the 1920–1930s and consisted on the reconstruction of the masonry of the tower in its middle part; a second intervention took place in the 1970s to reconstruct the top of the tower and included the introduction of a RC rigid floor, possibly with concrete ring beams, and columns with tuff cladding supporting the timber roof. The November earthquake caused the formation of some new cracks in the ground around the clock tower. Figure 24b shows a view of the tower before the earthquake, while Fig. 24c, d show two different views of after the earthquake. No damage is observed on the belfry, at the top of the tower, which proofs the high rigidity of the RC rigid floor introduced with the 1970s intervention. However, such heavy structural-oriented intervention is considered incompatible with the seismic performance of historic masonry structures (Colapietro et al. 2015), being often source of additional seismic damage. The unsuitability of this intervention of consolidation is confirmed by the global failure mechanism of the tower, highlighted by the presence of the vertical cracks showing an out of plane displacement of a portion of the front and rear tower walls.
On the other side of the castle, another building was found in extremely poor condition at the time of the survey. This is the Tekke of Dollma, shown in Fig. 25, which is a religious construction dated back to 1789 and proclaimed Monument of Culture in 1973 (IRPP/SAAH 2006). The building’s square shape turns into hexagonal in the upper part to envelope the central dome. The dome is sitting on squinches which are in turn resting on the stone masonry walls. Figure 25a–c show the front and rear view before the earthquake, while Fig. 25d, e show the situation after the November earthquake.
When comparing its view before and after the earthquake in Fig. 25c, d, it is possible to observe that in-plane diagonal cracks on the spandrels of the openings of the masonry wall were present at the same locations before the November earthquake. The original causes of the cracking could be associated to the ground instability observed in the area, as well as previous earthquakes (IRPP/SAAH 2006). Therefore, the November earthquake appears to have caused a progression of the existing damage, significantly increasing the size of the pre-existing crack pattern. The failure of the corners is likely to be the cause of the cracking observed on the dome, visible from Fig. 25e. From a glimpse of the inside, the dome presents cracks along the radial arches, with no cracking shown at the crown and regular cracking like ‘slices’ in the lower part. This type of failure due to excessive lateral loads is caused by movements in the supports of the dome and increase of the span which develop the vertical cracks observed in Fig. 25e (Heyman 2014).
The Prezë Castle, built in the fifteenth century, was designed to follow the conformation of the hilltop, resulting in an irregular pentagonal shape, with towers in each corner connected with fortification walls. All towers have a circular shape except for one that was reshaped into a 14.5 m high clock tower in 1852. The clock tower has a rectangular shape of 4.2 × 4.2 m and has two storeys accessible through an internal staircase. The tower has lost its original Ottoman style, which is visible from the photos in Fig. 26a, b capturing the damage suffered by the towers in different periods, most likely due to earthquakes.
Information on the condition of the tower before the earthquake is provided by Mustafaraj and Yardim (2014), who conducted a structural assessment of the tower in 2014. The results of the assessment identified the need of an intervention of retrofitting as the masonry of the tower was found with surface degradation and structural cracks propagating throughout the entire height of the tower, with the most significant cracks observed on the east side. Over the summer in 2019, works have been undertaken on the tower which included the strengthening of the masonry walls with the introduction of metallic ties to prevent overturning failures and the addition of an internal metallic structure to support a spiral staircase to access the 2nd floor.
A view of the tower after the earthquake is provided in Fig. 26c, d. It is possible to observe that the clock tower completely lost the roof and the columns framing the top floor. Figure 26e, f show the opposite east side of the tower, where the damage is more extended and includes the loss of the masonry of the upper third of the tower, below the top floor. Where the masonry was reinforced with anchors, the walls resisted the out of plane forces by activating the parallel wall and/or perhaps the internal steel structure and engaging the shear capacity of the transverse walls. In this part, the masonry shows vertical cracks between ties, a few of them could have been present since before the November earthquake.
The North tower in Fig. 27 was also damaged by the November earthquake. This tower has a circular shape and is made of stone masonry walls. Visible alteration to the original configuration of the tower observed during the visit are the introduction of a RC floor to create a terrace on the roof of the tower (Fig. 27a) and the construction of a building on the side of the fortification walls and adjacent to one of the sides of the tower (Fig. 27b). The damage occurred on the outer side of the tower, facing the cliff. This side is the most vulnerable due to the lack of confinement provided by adjacent structures. In this part, the masonry collapsed for out of plane mechanism, whereas the stiff slab remained undamaged (Fig. 27c). The presence of the new heavy RC floor activated a larger seismic mass and therefore resulted in a higher demand for overturning capacity on the wall.