The National Institute for the Study and Treatment of Cancer (IRCCS) “Giovanni Pascale Foundation” in Naples is composed of five independent buildings whose total plan extension, also comprising the access zones and car parks, amounts to about 60000 m². A satellite view of the entire complex and a general plan with the position of the different buildings are provided in Fig. (1), where:

• Edifice 1 is the “Scientific Building” (hereinafter stated as [SB]);
• Edifice 2 is the “Hospitalization Building” (hereinafter stated as [HB]);
• Edifice 3 is the “Day-Hospital Building” (hereinafter stated as [DH]);
• Edifice 4 is the “Administrative Building” (hereinafter stated as [AB]);
• Edifice 5 is the “Nun Building” (hereinafter stated as [NB]).

It is worth mentioning that Edifice 1 [SB] has not been involved in the assessment process, and Edifice 2 [HB] consists of several structural units, as better shown in the following section.

The predominant construction materials of the structures of the “G. Pascale” Institute are reinforced concrete and steel. In particular, [HB] and [AB] buildings were entirely realized in reinforced concrete, while [DH] presents beams and columns in structural steel with r.c. cores and [NB] consists of two different portions made of masonry and r.c. Some photos of the investigated buildings are shown in Figs. (2-5).

The activities started with a wide research aimed at finding all the historical information necessary to achieve a deep knowledge of the investigated structures. The history of the hospital complex has been reconstructed in detail. In particular, the realization of the first building, i.e., [SB], began in March 1934. Probably, it is hypothesized that in the same period, the masonry portion of the [NB] was erected, too. Between 1960 and 1985, several enlarging projects of the hospital were drawn up, which have deep transformed its organization and have led to the complex architectural and structural system that nowadays distinguishes the entire hospital plexus.

Fig. (1). Satellite view of the Giovanni Pascale Foundation (left) and general plan of the hospital facility (right).

Fig. (2). Photos of the [HB] building.

Fig. (3). Photos of the [DH] building.

Fig. (4). Photos of the [AB] building.

Fig. (5). Photos of the [NB] building.

As far as the construction time of each building is concerned, by means of the obtained documents and some oral information, it has been possible to determine the historical evolution of the “G. Pascale Foundation” as follows:

• The first enlargement of the original hospital (intended as the “Scientific Building”) dates back to February-August 1961 and consisted of realizing a small portion of the unit A of the [HB] building (the first two stories);
• Between June and November in 1968, the [HB] building was significantly expanded by adding two additional stories and a staircase attached to unit A and by erecting the units B, C and D;
• The following year, between February and June, a portion of the units F and G of the [HB] building were built;
• In the period between November 1975 and December 1980, all the units of the [HB] building were completed;
• The realization of the [AB] and [DH] dates back to the period between June 1977 and August 1982.

As far as the [NB] is concerned, it can be assumed that the erection of the masonry portion was completed in the same period of the [SB], while the r.c. structure dates back after 1980, despite precise information have not been found.

The structural analysis of the existing buildings has been carried out according to the Italian and European technical codes, thus following a “Performance approach” [16-18]. In particular, coefficients to reduce material strengths (namely confidence factors, FC) based on the achieved level of knowledge (KL) have been used. Since a wide preliminary investigation campaign involving both geometries and material tests has been performed, as well as original design documents were prevalently available, the highest level of knowledge (i.e., KL3), to which corresponds a confidence factor FC=1, was adopted for the cases under consideration. In this section, a general description of the destructive and non-destructive tests conducted with the aim to mechanically and geometrically characterize the investigated structural elements, as well as the main outcomes, is provided.

In particular, the wide investigation campaign foresaw: i. compressive tests of cylindrical concrete samples; ii. tensile tests of reinforcement steel bars; iii. tensile tests of steel structural elements; iv. durometer tests of steel structural elements; v. Vickers hardness tests on steel bolts; vi. flat jack tests of masonry elements; vii. pachometer tests or visual inspections on constructional details. A summary of the number of executed tests for each building and story is provided in Tables 1-4.

Fig. (15). Carpentry plan (2nd story) of [AB] buildings.

Fig. (16). Carpentry plan (2nd story) of [NB] buildings.

The structural capacity of investigated buildings, for both seismic forces and vertical actions, has been analysed by means of numerical models implemented in Sismicad, which is commercial software for the structural calculation based on the Finite Element Method theory.

With reference to r.c. elements, frames have been modelled as mono-dimensional elements (hereinafter stated as beams, for both horizontal and vertical elements), while the presence of walls has been introduced in the models by means of 4-node shell elements. As far as structural steel elements are concerned, again, beams have been adopted, taking care to release the extremity bending moment in the horizontal elements to account for the hinge connections. Finally, for the masonry structure, a typical equivalent frame modelling has been used, with beam elements for piers and spandrels and rigid offset simulating their connections. In Fig. (18), the pictures of the implemented numerical models are shown.

Fig. (18). Views of the numerical models: [HB] (top-left), [DH] (top-right), [AB] (bottom-left) and [NB] (bottom-right).

Since the condition of a concrete slab thicker than 5 cm was always guaranteed, the horizontal structures have been considered as rigid diaphragms, after sensitive analyses that returned similar outcomes in both deformable and undeformable conditions. The stiffness contributes of partitions and infill walls have been neglected, while their masses have been accounted for by means of external loads. Finally, the foundation systems, given the sufficient stiffness and the geotechnical characteristics of the underlying grounds, have been always simulated by introducing total constraint conditions at the base nodes of the models.

As far as the applied loads are concerned, they have been defined according to the Italian and European code provisions. In particular, other than structural and non-structural permanent actions, superimposed live loads, depending on the use of the various structures, and snow have been considered. Hence, in Table 7 a summary of the adopted values of vertical live loads are provided for each investigated unit, where are also mentioned the corresponding categories as defined in the Italian Standards for the constructions.

Fig. (19). Response spectra with CU=2 (left) and CU=1 (right).

The vertical loads have been combined according to the so-called fundamental combination, which has been adopted for checking the structures in the Ultimate Limit State conditions and foresees the amplification of the loads with partial safety factors and combination coefficients.

With regards to the seismic action, it has been defined as a function of the site, the soil characteristics, the exceeding probability depending on the considered limit state (P_VR), and the use coefficient of the structures according to their strategic importance (C_U), as well as the nominal life V_N, which has been considered equal to 50 years for all buildings. Hence, it is worth noticing that the investigated structures being almost all considered as strategic, a C_U=2 have been always adopted except for [NB], to which C_U=1 has been assigned. Furthermore, for the latter cases, the solely Life Safety (SLV) and Damage Limit States (SLD) have been considered (P_VR=10% and P_VR=63%, respectively), while for the strategic units also the attainment of the Operatively (SLO) one (P_VR=81%) has been investigated. Graphs of the adopted response spectra are plotted in Fig. (19).

The vertical checks have been executed based on linear analysis. On the other hand, the numerical analyses and the relative seismic checks to define the risk index (I_R) have been carried out by means of non-linear static analyses (Pushover), assuming both mass and modal proportional distribution of loads applied in the centre of the masses of each story. The risk indexes have been expressed in terms of both peak ground acceleration (PGA) and return period (T_R). Hence, the checks have consisted in comparing the displacement capacity of each structure or unit, obtained assuming the structure behaviour as a system having a Single Degree of Freedom (SDOF) and thus bi-linearizing the relative force-displacement curve, with the displacement demand arisen from the elastic displacement response spectra related to each specific limit state, opportunely reduced to account for the ductility of the structures.

In particular, the bilinear curves have been obtained from the numerical force-displacement curves, by assuming: i. a stiffness of the initial branch equal to those corresponding to the 60% of the maximum lateral force; ii. an ultimate displacement corresponding to those for which a decreasing up to 20% of the maximum lateral force occurs; iii. a maximum force obtained by equalling the areas under the real and the bilinear curves.

Hence, once that the elastic response spectrum (Spectral elastic Acceleration S_aeversus period T) has been defined, it is converted and plotted in the Spectral elastic Acceleration versus Displacements (S_de) plan (Acceleration Displacement Response Spectra –ADRS) by means of Eq. (4).

(4)

Then, by adopting the so-called N2 method implemented by Fajfar [22, 23], the demand spectrum (S_aversus S_d) is evaluated according to Eq. (5):

(5)

where, μ is the ductility factor defined as the ratio between the maximum and yielding displacement of the bi-linearized curve and R_μ is the reductive factor due to the ductility of the SDOF system, which could be calculated according to Eq.(6) [24]:

(6)

Hence, the demanded displacement is evaluated by estimating the so-called performance point, which is obtained by the intersection between the capacity spectra (i.e., the bi-linearized Acceleration versus Displacement curve) and the reduced response one. On the other hand, the capacity is evaluated by checking the attainment of the relative limit state, depending on the material of the structures, as better explained in the following. It is worth specifying that the seismic-resistant structure of the [DH] building consists of concrete cores. Indeed, being the steel beams connected by means of hinge nodes to the columns, this latter assumes an inverted-pendulum behaviour, which is characterized by very low lateral stiffness with respect to concrete elements. Moreover, regarding [NB] building, the operatively limited state has not been considered, given its non-strategic function. Thus, each limit state for r.c. structures have been considered attained if the following conditions have occurred:

• Operatively Limit State (SLO): drift (relative displacement between two consecutive floors) higher than 2/3·5.0‰·h (story height);
• Damage Limit State (SLD): drift higher than 5.0‰·h;
• Life Safety Limit State (SLV): i. for ductile collapse mechanisms, total chord rotation θ of a floor higher than 3/4·θ_u (ultimate chord rotation); ii. for brittle collapse mechanisms, shear failure or collapse of unconfined nodes.

Such conditions remain valid for [NB] concrete portion, while for the masonry one, the attainment of the limit states has been considered under the following conditions:

• Damage Limit State (SLD): lower displacement between those corresponding to the maximum force and those corresponding to a drift higher than 2.0‰·h;
• Life Safety Limit State (SLV): displacement equal to 75% of those at the Collapse Limit State (SLC). The SLC displacement is defined as the lower between those corresponding to a residual base shear equal to 80% of the maximum shear and those corresponding to the attainment of the maximum angular displacement in each masonry pier relevant for the global stability of the structure.

Finally, it is worth mentioning that the post elastic behaviour of r.c. and steel elements is accounted for by means of fiber elements. In particular, the stress-deformation state of a generic section of an element is obtained by integrating the uniaxial law of each fiber composing the section. On the other hand, the masonry elements have been considered with the typical elastic perfectly-plastic behaviour until the attainment of limit drifts depending on the type of crisis (flexural or shear).

In this section, the main results obtained by means of numerical analyses are summarised. Given the high number of performed analyses, for the sake of brevity here, the attention is focused on the outcomes arisen from only one of the investigated structures. Indeed, in addition to structural checks against vertical loads, 32 analyses have been performed for each considered unit in order to evaluate their seismic capacity. In particular, for each distribution of horizontal loads (mass and modal proportional), the two main directions have been combined with the relative eccentricities (even in this case assuming both signs) in order to account for the uncertainties related to the real position of the centre of masses. Hence, in this section, unit C of the [HB] is deeply analysed, since its dimensions and construction features are considered representative of the other units, as well as of major interest. Nevertheless, the results obtained for the other cases are also provided.

As far as the vertical loads are concerned, the numerical analyses showed, in the whole, a good capacity of the all investigated structures. Indeed, the structural checks always returned positive outcomes, despite in some cases safety indexes close to one, defined as capacity-to-demand ratios, have been obtained. In Fig. (20) the safety indexes of the unit C of [HB] due to vertical loads (i.e. fundamental combination) are plotted with a contour view.

It is possible to observe that in almost all beams, the work rates resulted close to the unit due to shear stresses. Nevertheless, these results are surely acceptable in order to guarantee operatively the structures in case of standard load conditions. In addition, the obtained results have been further corroborated by means of six load tests on slabs, which always showed a good structural behaviour.

In the present study, the risk indexes have been expressed in terms of both PGA and T_R. In the first case, the comparisons between demand and capacity are made directly by a graphical method in the ADRS plane, and the respective risk index I_R(PGA) is calculated by operating a simple ratio between the two values, namely capacity-to-demand ratio. Nevertheless, the current 2018 Italian code proposes indexes in terms of the return period, which is considered a more representative risk parameter. Indeed, with the new approach, the scale of risk assessment results very different from the one commonly used previously in terms of PGA, mainly due to the concave shapes of the response spectrum in terms of accelerations (or pseudo-accelerations) versus periods. Hence, despite it is possible to associate a return period to each level of acceleration, the risk index in terms of the return period, can be estimated according to Eq.(7):

(7)

Where T_R is the return period related to the attainment of a specific limit state (i.e., the capacity of the structure), T_REF is the return period with the exceeding probability typical of the considered limit state (i.e., the demand) and a is a coefficient adopted to obtain a scale of risk compared with the previous one. According to a statistical analysis of the national hazard curves, the coefficient a could be assumed equal to 0.41 [25].

As has been aforementioned, the results obtained for unit C of the [HB] are deeply analysed in this section. In Fig. (21), the dynamical characteristics of the model arisen from a modal analysis, which has been adopted for the modal proportional distribution of loads, are shown.

In Fig.(22), the checks in the ADRS plane related to the SLV conditions are plotted, while in Table 8, the obtained seismic risk indexes for each collapse (or damage) mechanism of unit C are summarised.

The results put into clear evidence the high seismic vulnerability of the investigated structure, being the minimum seismic risk index in terms of return periods equal to 0.197. More in detail, as far as the Operatively and Damage limit states are concerned, the structural checks are not satisfied, even if risk indexes quite close to the unit have been obtained. On the contrary, with reference to the structural checks at SLV conditions, the structure exhibited significant deficiencies, whatever the collapse mechanism was considered. In order to better understand the evolution of the considered collapse mechanisms, in Fig (23) the seismic risk indexes for each structural element and related to the different collapse criteria at the SLV are plotted in a contour view.

Fig. (20). Work rates of unit C of [HB]: minimum between bending moment and shear (left), bending moment (centre) and shear (right).

Fig. (21). Main results of the modal analysis: (left) mode 1 (T=2.06s, Mx=62.68%, My=0.18%), (centre) mode 2 (T=1.29s, Mx=0.59%, My=27.88%) and (right) mode 3 (T=0.98s, Mx=0.00%, My=38.21%).

Fig. (22). Check in the ADRS plane for the SLV conditions: shear failures (left), node collapse (center) and flexural failure (right).

Fig. (23). Seismic risk indexes at the SLV for shear failure (left), node collapse (centre) and flexural failure (right).

In particular, among the collapse mechanisms of SLV, it is interesting to note that the mechanism corresponding to the lower index is the shear failure of the external beams of 5^th and 6^th stories. Moreover, columns are also affected by a relevant shear deficiency, especially the ones close to the staircases. Similarly, with reference to node collapse, the main seismic issues have been obtained in the staircase proximity, despite a lack of capacities that have been widely obtained. Finally, with regard to ductile (i.e., flexural) collapses, although some columns present a bad capacity with respect to the demand, the largest portion of elements resulted verified against the design seismic loads.

It is worth noting that the structural problems affecting unit C of [HB] are typical for buildings erected between ’50s and ’60s of the last century. This aspect appears even more relevant considering that the first Italian rigorous standards (i.e., based on scientific knowledge rather than empirical criteria) in the field of seismic engineering dates back to 1974 [26].

Indeed, most of the buildings analysed in this study, returned a significant seismic vulnerability, as it is shown in Table 9 and Fig. (24), where the seismic risk indexes are summarised for each considered limit state. For convenience of the reader, only the most penalizing collapse mechanism has been considered for the SLV.

Fig. (24). Seismic risk indexes in terms of both PGA and TR obtained with the numerical analysis for each unit.

In the whole, the results reveal a lack of seismic capacity with respect to the demand of the investigated structures. More precisely, even in the case of less significant seismic actions (i.e., in SLO and SLD conditions), the seismic risk indexes resulted in most cases less than 1. As far as the SLV conditions are concerned, the so-defined strategic buildings, all returned a seismic risk index less than 0.26, in terms of both PGA and return period. Nevertheless, [NB] also exhibited a significant vulnerability, especially for the r.c. portion, despite less demand due to the ordinary use of the construction.

In addition, with reference to units of [HB], possible pounding phenomena between adjacent structures could occur under seismic loads, as well as the loss of supports in case of Gerber connections. Moreover, for each investigated building, the so-called unquantifiable vulnerabilities, both structural and non-structural, have also been detected. In particular, such problems are mainly related to the state of conservations of the buildings, as largely advanced degrade phenomena of the r.c. structures, i.e., oxidation of steel bars and consequent detaching of the concrete coverage, have been detected. Other similar problems, which could lead to early significant damage or unusability, are represented by the anchorage of both plant systems and shelves, double-leaf infills without transversal connections, plaster degrades, and humidity issues, (Fig. 25).

The last phase of the study provided the definition of possible strengthening interventions to guarantee the functionality of the structures also in case of seismic events. It is worth noting that the possible strengthening measures proposed in this phase were mainly aimed at preliminary estimating the retrofitting costs, rather than designing the specific interventions, for which an additional and extensive detailed activity would need. In particular, based on the outcomes obtained by the structural assessment, strengthening interventions aimed at improving the seismic capacity of the structures have been hypothesized in order to increase the seismic risk indexes to a value in the range of 0.80-0.85, which represents a minimum recommendable threshold for strategic buildings, like the ones considered in this study. To this purpose, as a basic retrofitting strategy, supplementary structural elements have been introduced in the structural system, which should be able to counteract about 60-65% of the total seismic story shear forces. Moreover, interventions aimed at reducing the so-called unquantifiable vulnerability have also been suggested.

Fig. (25). Examples of degradation of concrete elements (left) and inadequate anchorage of plant systems (centre) and shelves (right) in [HB] building.

Some of the main suggested structural interventions for the examined buildings are listed in the following:

• Insertion of dissipative steel diagonals or V-bracings for r.c. structures in combination with retrofitting of beam-to-column nodes and beams (with FRP) and columns (with steel jacketing);
• Insertion of dissipative steel diagonals or V-bracings for steel structures with retrofitting of beam-to-column nodes, beams and columns;
• Enlarging of existing r.c. walls;
• Realization of new r.c. walls;
• Enlarging of existing r.c. beams;
• Retrofitting of existing r.c. beams by means of FRP applications;
• Retrofitting of existing r.c. columns by means of jacketing of steel elements;
• Demolition and reconstruction of existing joints between adjacent buildings;
• Realization of new steel beams;
• Realization of new r.c. frames;
• Retrofitting of connections between steel beams and r.c. cores;
• Realization of new tuff masonry walls;
• Realization of reinforced plaster on both sides of masonry walls.

In Table 10, a summary of the structural interventions applied for each structural unit is provided.

The suggested structural interventions were also aimed at reducing the so-called unquantifiable vulnerabilities, namely those not detectable by means of numerical models. The most relevant of these problems was surely represented by the double-leaf infills, which usually exhibit out-of-plane collapse mechanisms in case of seismic events. To overcome this issue, the insertion of transversal drills consisting of masonry bricks, as well as the realization of anti-overturning systems by means of reinforcement strips (e.g. GFRP), has been considered. In addition, updates of the existing installation systems were also prescribed by means of suitable and flexible connections in order to guarantee their functionality also under seismic load conditions. Similarly, appropriate anchorages of the shelves to the walls were planned. Moreover, widespread interventions aimed at preserving and reducing the degradation of the existing r.c. structures were also foreseen.

The strengthening interventions qualitatively described in the previous section allowed the estimation of costs for upgrading the existing structures. Such a phase is significantly relevant for decision-making processes since, by means of a cost-benefit analysis, a complete reconstruction could often result more convenient than diffused interventions for ancient buildings such as the ones under consideration. Hence, a comparison between the retrofitting costs and those corresponding to the realization of new buildings is also provided by considering parametric costs typical of the investigated structures according to official reports and documents in terms of cost incidence per square meter of story.

In Table 11, a summary of the estimated costs for retrofitting the existing structures in order to achieve seismic risk indexes, as defined by means of Eq.(7), of at least 0.80 is provided for each unit.

Fig. (26). Comparison between retrofitting and reconstruction costs.

The so-estimated costs include both structural and non-structural interventions, while disregard the labour cost, as well as design and tax amounts. Moreover, a weighted average has been considered for [HB] buildings, which leads to define a total retrofitting cost of 3’1855’000 € and a cost per square meter of story of 932 €/m². It can be stated that, according to this estimation, the Hospitalization Building, due to its structural complexity and strategic functionality, revealed the need of the most expensive interventions.

Hence, the costs expressed in terms of incidence per square meter have been compared to parametric costs available in official documents, which are related to the realization of new buildings and, again, disregard additional costs. In particular, for [HB] and [DH] reference to [27] has been made, which is a relevant report including theoretical construction costs for hospitals. In particular, by assuming a mean value of the values proposed in this document, a cost in the range of 1800-2500 €/m² for the realization of new buildings has been obtained. This value resulted comparable to those obtained for subunits D and F of [HB]. In this regard, it is worth mentioning that these units present a low surface extension, as well as they should need relevant interventions aimed at avoiding interactions with other structures, which significantly engrave on the total cost. On the contrary, by comparing such a value with the ones returning by the entire units [HB] and [DH], can be noted that it corresponds to about two and three times, respectively, with respect to upgrading costs of the existing structures.

As regards [AB], reference to [28] has been made, in which is assumed that the cost of realization of a new building with a rectangular plan for offices can be estimated ranging between 1300 €/m² and 1450 €/m². Hence such a value results strongly higher than those evaluated for upgrading the existing structure. Finally, [NB] has been considered on a par with residential buildings. Hence, again according to [28], the price for the realization of a new building can be estimated in the range of 800÷1000 €/m², which resulted higher than the evaluated retrofitting cost.

In the whole, the reliability of the planned possible interventions has been widely confirmed, being the reconstruction costs always higher than retrofitting ranges. A summary of the obtained results is provided in the histogram of Fig. (26), where the retrofitting and reconstruction costs per square meter are compared.