Keywords

1 Seismic Vulnerability of Existing Masonry Buildings

Architectural heritage is particularly prone to damage in seismic areas, due to construction defects or limitations, which combine, together with the effects of lack of maintenance and deterioration exposure of materials. Common vulnerabilities associated with building construction, often detectable in existing masonry buildings, involve one or more of the following aspects: (i) the poor quality of the masonry; (ii) the scarce connections among components (walls, floors and roof); (iii) the structural irregularities; (iv) the inadequate stiffness of horizontal components (floors and roof); (v) the existence of thrusting structures (e.g., arches and vaults, but also pounding elements in floors and roofs). These deficiencies result in partial or even overall collapses (‘mode 1’ mechanisms), mainly due to overturning and out-of-plane bending of walls and assemblages (e.g., corners), which can occur even at low-medium earthquake magnitudes [12, 3, 20, 16]. In addition, poor-quality masonry, mainly due to the lack of transverse shear elements in multi-leaf walls, irregular texture and composition, low-grade properties of the constituent materials (mortar and bricks/stones), deterioration of binding power in ancient mortars [6, 7], may even involve anticipatory brittle collapse (‘mode 0’) due to disaggregation of constituent materials, which may occur under even lower seismic magnitudes and before any other sort of mechanism can be activated.

Such brittle behavior (modes 0 and 1) can be inhibited by better quality of materials and more adequate construction systems and details, so that more extensive collaboration among parts can be activated and exploited (‘box-like’ behavior). This results in overall pseudo-ductile behavior (adequate displacement capacity without collapse), which can concentrate dissipative shear cracks in lintel elements, thus preserving as much as possible the integrity of piers (‘mode 2’ mechanisms), provided that their continuity in height is ensured by the regular lay-out of openings.

All these behaviors were observed as the effects of a series of seismic events in central Italy from August to October 2016 and are still active, i.e., in the highlands encompassed by Macerata (N), L’Aquila (S), Ascoli Piceno (E) and Perugia (W), known as Sibillini Mountains Park. These areas and their surroundings are some of the most active seismic areas (0.255g expected PGA according to the seismic hazard map; more than 0.7g actual PGA recorded in 2016) [15, 19]. High seismicity in this area has always meant that local population are scattered in many villages and ‘castles’, so that the very concept of ‘cultural landscape’ is embodied in them, with the result that they have been built and rebuilt over the centuries after every earthquake. They are the product of traditional building techniques, which gave rise both to their vulnerability and to attempts at avoiding it.

Figure 1 shows the location of the centers studied here, in the area struck by the 2016 sequence: Castelsantangelo sul Nera (province of Macerata: MC), Campi Alto di Norcia and Castelluccio di Norcia (Perugia: PG). They are all fortified villages built in the late Middle Age to keep control of fertile lands and roads between the borders of Norcia and Visso, the two largest, at the time of their expansion.

Fig. 1
figure 1

Shake map of October 30 2016 earthquake (not to scale). Case studies are located within the red area, which includes the Sibillini Park [15]

Full size image

Figures 2 and 3 show the typical damage, detected in central Italy after the 2016 earthquakes, encompassing damage modes 0 and 1 (unfavorable conditions), and mode 2 (favorable conditions), due to constructive aspects.

Fig. 2
figure 2

Unfavorable behavior of masonry structures under seismic actions. Out-of-plane collapse (modes 0 and 1) due to poor-quality construction: a collapse due to poor-quality masonry (Gualdo, near Castelsantangelo sul Nera); b local collapse due to lack of connections in multi-layer masonry at corners (Castelsantangelo sul Nera); c pounding due to irregular interaction between house and contiguous bell-tower (Castelsantangelo sul Nera); d facade overturning in terraced house with original timber floors parallel to façade (Castelsantangelo sul Nera); e effect of thrusting roof resting on two-leaf masonry wall (Castelsantangelo sul Nera)

Full size image
Fig. 3
figure 3

Favorable behavior of masonry structures under seismic actions. In-plane damage limited to repairable elements without collapse (mode 2): shear failure mainly extensive in masonry piers a (Castelsantangelo sul Nera) and, more favorably, on lintels b (Pieve Torina, MC)

Full size image

Unfortunately, ‘mode 0’ failure is a major problem. The supply of building materials near it greatly influences the quality of masonry, which ranges from including very poor sand and clay mortar with almost spherical sandstone rocks (Fig. 2b) to better-dressed limestone with pure lime mortar (Fig. 2a).

Many houses show signs of earthquake damage and also the basic systems used to prevent major damage, ascribed to mode 1: buttressing, tie-bars and special systems, such as the detachment of barrel vaults from the façade to preserve the former when the latter collapses, as already observed in Campi Alto di Norcia [8, 4].

Lastly, the preferred mode 2 occurred in a very few cases and mostly in more recent buildings, dating to the 19th century, when some rules for seismic-proof practices (after the 1859 earthquake that struck the region) were issued. However, these regulations apply only to large towns (Visso, Pieve Torina, Norcia), while traditional systems survived in villages.

More commonly, since 1979 (year of the Valnerina earthquake, Mw 5.9) and until the early 2000s (1997 Colfiorito earthquake, Mw 5.8), various retrofitting techniques have been applied to repair damage, mainly substitution of components (especially floors and roofs) and the addition of heavy, incompatible structures (e.g., reinforced concrete).

1.1 Procedures for Vulnerability and Damage Evaluation

The earthquakes of the early 1980s stimulated the development of simplified tools and procedures to evaluate both the damage and vulnerability of ordinary masonry buildings, mainly through survey forms. Of these, ‘GNDTFootnote 1 II level’ [2, 13, 11] and AeDESFootnote 2 forms [1, 10] are the most used.

The GNDT form evaluates a normalized index (Iv) with three levels of vulnerability, which can be related to verbal assessment of the overall vulnerability of a building (from ‘very high’ to ‘very low’), as shown in Table 1 [9].

Table 1 Correlation between GNDT form scores and vulnerability classes according to EMS 98 (from [9])
Full size table

The AeDES form dates back to the 1997–98 seismic events which struck the Umbria and Marche regions and which has been extensively validated in many post-earthquake contexts (Molise 2002, L’Aquila 2009, Emilia 2012). It relates the grading of damage levels of five main components, according to the European Macroseismic Scale 1998 (EMS 98) [14], also taking into account the extent of damage throughout the building in question (Fig. 4). Evaluation requires simpler information than the GNDT form, because it was designed for on-site use, just after seismic events.

Fig. 4
figure 4

Damage evaluation module in AeDES form [18]

Full size image

These tools have been used in vulnerability and damage evaluations since the early 2000s for case studies throughout Italy’s central regions. However, recent events have shown that they would need updating, in order to represent better the peculiar actual conditions of built heritage all over Italy, seriously affected since the adoption of r.c.-based techniques.

2 Aim and actual effect of interventions

Interventions in masonry buildings in seismic areas mainly focus on rehabilitating collaboration among parts, at both local (cross-section, texture, connections) and more extensive building scales (strengthening of structural components and assemblages, overall working). Table 2 lists intervention techniques aimed at reducing the possible vulnerability of masonry buildings and their structural components.

Table 2 Intervention techniques to reduce vulnerability in masonry buildings
Full size table

The proper design of interventions according to the vulnerability in question, the adoption of compatible materials and techniques, and the suitable installation are all essential requisites, if effective results in enhancing structural behavior are to be attained [5, 17, 21].

Nonetheless, analysis of historic city centers struck by earthquakes in the last 40 years in Italy has revealed severe damage and collapse attributable to additional vulnerability triggered by heavy retrofitting techniques: although they were ideally conceived to improve mutual collaboration among the structures, in actual fact they entail more complex and hybrid behaviors (Fig. 5). It is the case of the use of r.c. to substitute floors and roofs (aimed at increasing the in-plane stiffness), to strengthen vaults (to prevent collapse), or applied as ring beams at floor and roof levels (to connect walls against overturning). This practice was commonly adopted in masonry buildings from the 1980s onwards, according to the knowledge and recommendations available then. However, even more recent techniques, thought to be more compatible with ancient fabrics, e.g., horizontal steel trusses as ties, have proved to be inadequate, mainly due to the lack of connections to masonry.

Fig. 5
figure 5

Damage caused by unfavorable interaction of existing masonry structures with retrofitting techniques: a corner overturning due to r.c. ring beams (Campi Alto di Norcia); b damage at interface between r.c. beams and masonry (Castelsantangelo sul Nera); c disaggregation of ancient masonry together with new masonry (Gualdo, near Castelsantangelo sul Nera); d unreinforced first-floor overturning under reinforced upper storeys (Campi Alto di Norcia); e collapse of jacketed multilayer masonry (Nocelleto, near Castelsantangelo sul Nera); f failure due to irregular distribution of r.c. jacketing (Castelluccio di Norcia); g shear failure in precast r.c. beams and consequent overall collapse due to lack of horizontal floor diaphragms (Campi Alto di Norcia); h partial collapse due to poor-quality masonry under heavy r.c. floors (Pretare, near Arquata del Tronto, AP; i overall (Castelluccio di Norcia) and j partial (Castelsantangelo sul Nera) collapse under heavy r.c. floors and roofs; k V-shaped overturning under r.c. ring beam, even in presence of anchoring rebars (Castelsantangelo sul Nera)

Full size image

3 Vulnerability Maps of Historic Centers

Due to extensive observation of unfavorable effects of modern interventions after the 2016 seismic events in central Italy, a new form has been proposed to flank the assessment procedures mentioned above (i.e., GNDT II level and AeDES forms). This form collects information on interventions and modern alterations of heritage buildings as part of a purpose-designed method able to implement data collected on entire (small) historic towns in GIS [19].

Among others, the three villages studied here (Campi Alto di Norcia, Castelluccio di Norcia, Castelsantangelo sul Nera) offer an interesting overview of damage mechanisms proper to ‘original’ buildings, i.e. with only traditional interventions, as well as ‘altered’ buildings affected by invasive techniques.

Campi Alto di Norcia and Castelsantangelo sul Nera have a similar layout: an egg-shaped boundary wall built on slope with terraced houses on the lower half, and a tower and church at the top. Instead, Castelluccio di Norcia is built on a small cliff overlooking the surrounding plain with a curtain wall at its base. Over the centuries, these ‘castles’ have been abandoned because of the development of extra moenia suburbs, closer to the road systems, like for Castelluccio di Norcia and Castelsangelo sul Nera. Campi Alto di Norcia, being further from the valley road, is now almost completely abandoned (Fig. 6).

Fig. 6
figure 6

View of three towns surveyed after 2016 seismic sequences: a Campi Alto di Norcia (PG); b Castelluccio di Norcia (PG); c Castelsantangelo sul Nera (MC)

Full size image

The new procedure collects data on interventions, both architectural (e.g., changes in plans’ layouts of openings distribution) and structural, with reference to vulnerability assessments and damage states available over the last 40 years (i.e., from 1979 onwards), according to building types existing in the three centers (Fig. 7).

Fig. 7
figure 7

Building types in a Campi Alto di Norcia (PG), b Castelluccio di Norcia (PG), c Castelsantangelo sul Nera (MC)

Full size image

In particular, comparisons between the conditions and the damage scenarios surveyed after the 1979 and 2016 seismic events provide significant results in terms of evaluation of intervention effects. Figure 8 shows a considerable shift towards higher damage states in Campi Alto di Norcia: no cases of collapses can be found in 1979 and ‘medium’ damage (i.e., D3 category, according to EMS 98) involves much of the building stock. In 2016, damage state surveys show the opposite behavior, as most buildings are listed either in EMS 98 classes D1-D2 (low or almost no damage) or D5 (collapse).

Fig. 8
figure 8

Campi Alto di Norcia, distribution of EMS 98 damage states in 1979 and 2016: a absolute frequencies; b correlation between them

Full size image

GIS mapping clearly shows that, for instance (Fig. 9), the buildings severely damaged in Campi Alto di Norcia in 1979 are now not damaged, and vice versa. On one hand, this means that retrofit interventions have been effective, whereas buildings with generic structural updating became more vulnerable than the previous states. On the other hand, effective interventions imply the complete transformation of the original architectural qualities of a building, because all components (wall, floors, roof) are involved in the strengthening process.

Fig. 9
figure 9

Campi Alto di Norcia, comparison between EMS 98 damage level distributions after 1979 a [19] and 2016 b earthquakes in retrofitted buildings

Full size image

Also in Castelluccio di Norcia and Castelsantangelo sul Nera, for which no information about damage states after previous earthquakes is available, a certain shift towards higher EMS 98 damage states (D4-D5 compared to D2-D3) can be observed (Fig. 10). This trend may be an effect of the high number of partial ‘mode 1’ failures caused by interventions (mainly installations of r.c. ring beams and floors).

Fig. 10
figure 10

EMS 98 damage level distributions in a Castelluccio di Norcia and b  Castelsantangelo sul Nera after 2016 earthquakes

Full size image

Lastly, a comparison between actual damage states and existing vulnerability assessment procedures (GNDT II level and AeDES forms) has been proposed. To this aim, the AeDES damage module assessment (Fig. 4) has been transformed into a normalized index (Id) with assigned scores of 9 for the worst damage state and 0 for ‘no damage’, and calculation of the weighted average according to the values listed in Table 3.

Table 3 Proposed values to be used in AeDES damage index evaluations
Full size table

Comparisons among the resulting Id to damage states according to EMS 98 show a significant correlation between indexes (Fig. 11). Instead, when Id is compared with a vulnerability estimator such as the GNDT index (Iv), almost no correlation is found (Fig. 12): that is, vulnerability factors identified by the procedure were either wrongly judged or ill-defined, again presumably because of structural interventions.

Fig. 11
figure 11

Distribution of EMS 98 damage states in 2016: Campi Alto di Norcia (a) and Castelsantangelo sul Nera (b)

Full size image
Fig. 12
figure 12

Campi Alto di Norcia, correlation between vulnerability index (GNDT II level method) and EMS 98 damage state (a) and AeDES damage index (b), both showing very little correlation. Iv scale is limited to 0.5, to distinguish among points (max Iv = 1)

Full size image

4 Conclusions

The latest earthquakes in Italy in 2016 confirm the essential role played by masonry quality and connections among components for proper functioning of improvement techniques commonly proposed to strengthen horizontal structural components. This aspect is particularly important when heavy retrofitting interventions (substitution of floors and roof, or strengthening of vaults with r.c.-based slabs) are applied to existing multi-leaf masonry structures with few or no inner and/or outer connections.

It is well known that these techniques, widely used in the past after previous earthquakes, greatly influence the seismic behavior of masonry buildings. In particular, with reference to the three city centers analyzed here, research showed that retrofitted buildings suffered severe damage or even collapse. According to EMS 98 damage classification (D3: ‘average’ and D4-D5 ‘high’ damage states), Castelsantangelo sul Nera, Castelluccio di Norcia and Campi Alto di Norcia, showed the following percentages of damaged buildings: 45, 41 and 10% of buildings in D4-D5 and 15, 12 and 4% of buildings in D3.

In addition, building alterations after such large-scale interventions also affected the ‘representativeness’ of vulnerability assessment methods. The application of common expeditious and simplified procedures (GNDT II level and AeDES forms) to the three city centers repeatedly struck by earthquakes over the past 40 years, with subsequent retrofitting measures, showed high scatter of the vulnerability judgments with respect to the actual damage observed.

Further research will focus on developing calibrated procedures for vulnerability assessment, based on the observations of the effects of intervention techniques on masonry buildings in other historical city centers in the area.