5. A Case Study Application: Vittorio Veneto Square in Matera, Italy

The selected case study to apply behavioural design methods for terrorist risk assessment and mitigation concerns the outdoor Open Area (OA) of Vittorio Veneto Square in Matera, Italy. Vittorio Veneto Square is a public square located within the actual perimetration of the historic district of Matera, a city in the Basilicata region in the south of Italy. Its development is related to the urban expansion from the thirteenth to the seventeenth centuries, when the actual buildings were built, while its physical transformations and maintenance rely on its uses. It represents the nodal point for tourist access to the Sassi, the complex underground system of ancient dwellings of Matera listed as a UNESCO site [1].

Vittorio Veneto Square is configured as a flat terrace overlooking the Sassi, albeit, in the transformation process, direct exposure to the Sassi has been restricted to some points due to the construction of buildings along its perimeter (Fig. 5.1). Two main un-built elements allow to connect the square to Sassi: the balcony, for the landscape view of Sassi, created within a Loggia (“Belvedere Luigi Guerricchio”) in the central part of the western-built profile, and stairs to physically connect the square to Sassi. Another physical discontinuity of the floor is represented by the presence of the “Palombaro”, ancient water cisterns dug in 1882 under the square and connected with the other cisterns in the Sassi to serve the city. They fell into disuse with the activation of the urban aqueduct in 1991 but today they are uncovered to be observed from the square as engineered masterpieces, protecting their perimeter with fences [1]. Similarly, a second archaeological dug interrupts the pavement of Piazza Vittorio Veneto, allowing tourists to observe lower rooms while fences limited the perimeter. The Loggia and the “Palombaro” are both placed, in the northern part of the OA.

An illustration exhibits a map inside a circle. The locations of the Plazza dell Annunziata, Bank Pascarelli Palace, prefecture, San Domenico Church, Loggia, and Palombaro are indicated on the map. The photographs of the highlighted places are provided around the map.

The built profile reflects the historical evolution of the square, presenting private palaces, with 2–3 floors, and public buildings. Most of the ground floors of private dwellings host commercial activities (bars, pubs, restaurants, shops), while public buildings urban services. Specifically for these, three main public uses can be identified:

Moreover, the square presents also a religious fabric, the San Domenico Church.

The case study OA is characterized by a complex morphology, merging a main trapezoidal shape in the northern part of the square to a rectangular one in the southern. Concerning the square connection with the surrounding urban built environment, six access streets can be recognized, mainly pedestrian (Fig. 5.2), limiting the vehicular traffic within the same pedestrian square. However, three possible vehicular access streets can be identified: Via Roma, Via del Corso, Via Luigi Lavista (Fig. 5.2). Here, the geometric dimension of entrances and the urban vehicular traffic regulations allow to move close to the square, while fixed bollards limit the vehicular accessibility. In view of the above, it is worth noting that the assessed OA is also characterized by multiple uses, combining the public and touristic services and attractions that make Vittorio Veneto Square a sensitive soft target. In fact, the symbolism of the place determined by the presence of the UNESCO site and the presence of high touristic (cultural) flux determine a high potential level of proneness. This can be summarized as the local relevance of the place by local users which usually serve public places and buildings for daily activities. Moreover, the tourist inflow has increased during the last years due to the international relevance of Matera as the “2019 European Capital of Culture” which enhances the relevance of the place for tourist attractiveness and symbolism [2].

A map highlights a location in the middle and indicates the access via Roma, via San Biagio, stairs for Sassi, via delle Beccherie, via del Corso, via Luigi Lavista, and via Giuseppe de Blasiis. The places with limited vehicular access, walkable access, and stairs are indicated with different triangles.

Methods and tools provided in Chap. 4 are then used to create scenarios and evaluate risk due to terrorist acts in the case study, using the same phase order. In particular, the risk assessment methods devoted to provide possible attack points are applied in Sect. 5.2, while the time-dependent assessment of the users’ exposure and vulnerability is defined in Sect. 5.3. According to the outcomes of these methodological steps, simulations are finally performed to investigate risk levels in different pre- and post-retrofit scenarios, as shown in Sect. 5.4, using behavioural and simulation-based key performance indicators (KPIs) to compare evacuation issues under the given boundary conditions. To this end, selected mitigation strategies implemented in post-retrofit scenarios refer to emergency management and planning, thus ensuring the best compatibility with the heritage features of Vittorio Veneto Square, without altering the layout or the identity of the places.

Starting from the terrorism risk assessment formulation in OAs defined in Chap. 4, Sect. 4.​2 and based on the approach of previous works of the authors [3], all the required data are gathered and qualified. Considering the building intended uses presented in the previous section, their spaces of relevance (SoRs) are calculated and modelled in the plan (Fig. 5.3).

A map highlights the locations of 17 different places in different shades. The color codes represent Dehors, cultural sites, bars and restaurants, worship places, government administrative buildings, theatres and public libraries, office buildings, and shops.

The buildings and attractive places considered in the analysis include special and public buildings, tourist attractive places such as “Palombaro” and Loggia on the Sassi, public commercial activities such as bars, their covered terraces, and stores. Specifically, for buildings, the SoRs have been calculated considering their commercial extension and the number of floors occupied for the use [m²], which is also expressed by GS_i (compare with Sect. 5.3); while for all the gathering elements and parts of the squares, such as covered or uncovered bar terraces, their SoR extension [m²] is equal to the same element. Then, all the physical elements within the analysed built environment have been located and qualified in terms of their typologies (Fig. 5.4). A summary of details is presented in Table 5.1.

A map highlights the locations of different objects in different shades. The color codes represent the covered bar traces, stairs, monuments, flower pots, bollards, trees, lampposts, and fences.

The results of risk calculations for each area and SoR are processed considering the attack types recognized for OAs, T2—armed assault and T3—bombing attack with a vehicle, summarized in Tables 5.2 and 5.3 and outlined in Figs. 5.5 and 5.6.

A map highlights the locations of 17 different places in different shades. The color codes at the bottom represent the areas with negligible risk, low risk, medium risk, and high risk.

A map highlights the locations of 17 different places in different shades. The color codes at the bottom represent the areas with negligible risk, low risk, medium risk, and high risk. A maximum number of places are highlighted under low risk.

As a critical analysis of results, two main aspects can be discussed:

This analytic reading of the phenomenon through qualitative and quantitative data allows to interpret the OA and to determine two possible attack points for the T2 type while neglecting the T3 one, as schematized in Fig. 5.7.

A map represents the locations of different places. Two points are highlighted in the map in different shades and represent attack scenario 1 A S 1 and attack scenario 2 A S 2.

Specifically, the determined attack points describe two possible significant scenarios coherently with the terroristic strategies and efficacy of violent acts:

As discussed in Chap. 4, the identification of strategies to prevent and mitigate the effects of a terroristic attack is strictly linked to the attack type and modus operandi. The identification of possible scenarios of violent acts to be carried out through cold arms for the case study of Vittorio Veneto Square requires to be merged with the possible effective strategies and their efficacy. However, due to its inherent features, the strategies for the T2 attack type have a prevalent tactical dimension. In fact, even if literature, guidelines, and previous experiences have highlighted the relevance of physical obstacles in determining possible temporary secure areas during the violent act, such effectiveness requires to be combined with coherent actions of education of users.

Starting from the Risk Mitigation and Reduction Strategies (RMRSs) classification (Chap. 2, Sect. 2.​3), strategies related to security personnel deployment, emergency management, and wayfinding in emergency scenarios have been selected in this research, since they can essentially support users and define design solutions within the built environment layout without altering the physical aspect of such places. On the other side, these strategies can be also consistent with the supposed “modus operandi” in the T2 attack, which does not imply damage to the buildings, thus limiting the need for interventions on physical elements, and which contrarily implies a dynamic perpetrator behaviour according to the prey-predator modelling criteria (see Chap. 3, Sect. 3.​4). Considering the peculiarities of the assessed case study, moreover, the nearby presence of the local prefecture may support this kind of action. Near to that, the study of possible evacuation scenarios involving different conditions of evacuation with respect to the accesses and the LEA’s position can support the study and the expectation of emergency management.

Due to that, post-retrofit scenarios are outlined merging the configuration of the real investigated OA and the outlined attack points, preparing the simulation scenarios for the qualification of single strategies when combined (Figs. 5.8, 5.9, and 5.10). Specifically, for Scenario 1 (AS1), the strategy involves LEA in the northern part of Vittorio Veneto Square (AS1-ST1), taking advantage of the Prefecture (Fig. 5.8); in Scenario 2 (AS1) the first strategy considers closing minor exits (in the southern part) to the advantage of more wide accesses in the northern part (AS2-ST1) (Fig. 5.9), while the second adds the LEA in the nearest of the Prefecture as a temporary secure area to be reached (AS2-ST2) (Fig. 5.10).

A map represents the locations of different places. The LEA positions and attack points are marked with different circles. The possible exits are denoted with triangles. The map is marked as Strategy 1 A S 1 S T 1.

A map represents the locations of different places. The denied exits are highlighted with the crosses around the attack point. The possible exits are denoted with triangles. The map is marked as Strategy 1 A S 2 S T 1.

A map represents the locations of different places. The denied exits are highlighted with the crosses around the attack point. The LEA position is highlighted with a circle. The possible exits are denoted with 2 triangles. The map is labeled Strategy 2 A S 2 S T 2.

The time-dependent assessment of user-related factors has been performed according to Chap. 4, Sect. 4.​3 methods, using remote analysis and quick (standard) input data to focus on the rapid application of the methodology and the related capability demonstration.

The main areas connected to outdoor and indoor intended uses which can generate overcrowding are shown in Fig. 5.3, and selected according to the classification of Chap. 4, Table 4.​3. The main exits/access streets to the OA considered in the evacuation process and the main obstacles outdoors (that are greeneries, fountains, outdoor walls, and street furniture) are shown in Fig. 5.4.

Intended uses are then associated with their related main features in terms of surface, users’ typologies (only outdoor users—OO, prevalent outdoor users—PO, non-residents—NR), and users’ exposure over time. From a methodological perspective, the intended uses are associated with the OA via Google Maps/Street, the available gross surface GS_i [m²] is calculated by Calcmaps, and standard occupant loads and online timetables are assumed to pursue the rapid applicability of the proposed methodology, thus avoiding time-consuming on-site survey. For these reasons, uncertainties due to specific conditions of the square and to the relationship of the OA with the whole urban fabric (e.g., in terms of visitors’ flows) can exist, but the whole capabilities of the methodology are not affected by such simplified assumptions. The users’ vulnerabilities assessment by users’ age and gender are also performed according to a quick approach, thanks to the statistics of the National ISTAT annual reports,¹ assuming a homogeneous municipality-related distribution of data for the sake of simplicity.

In view of the, Table 5.4 traces the summary of the main features of the intended uses open to the public, thus excluding residential areas as in the rationale of the methodology in view of the attack attraction towards soft targets [4, 5]. Particular attention is paid to NR associated with special buildings having a symbolic value or that are widely interested in visitors’ presence over the day (marked with * in Table 5.4), since their position can be associated with the immediate outdoor areas of the buildings, thus maximizing exposure [6] (compare Chap. 3, Sect. 3.​4 on the simulation model). Therefore, the total number of users per intended use NU_t.i [persons] is also shown in Table 5.4. Figure 5.11 shows the trends of the main user-related KPIs defined in Chap. 4, Table 4.​4, focusing on occupant density (Fig. 5.11a), normalized KPIs on overall exposure and exposure referring to the outdoors (Fig. 5.11b), percentages of users by use behaviours (Fig. 5.11c) and age (Fig. 5.11d). In this sense, according to Table 5.1, working days seem to be related to higher users’ exposure levels due to the higher crowding. Thus, the data shown in Fig. 5.11 refers to typical working days.

4 line graphs represent the trends of U O d, ratio, and percentage of users at different hours of the day. a. The lines represent exposed and overall. b. The lines represent Nunt and I E t. c. The lines represent not exposed, special buildings, O OP, P O p. d. The lines denote T p, P C p, Y A p, A p, and E p.

Considering the overall users’ outdoor density in outdoor at a given time t UOd_t [persons/m²] (Fig. 5.11a) case study, OA seems to be characterized by the highest exposure values during the late morning and the afternoon. The maximum exposed users’ density (about 0.37 persons/m²) is reached at about t = 18, but values are lower than those of the whole users’ density, also comprising users placed indoors and in possible protected areas (about 0.73 persons/m²) at the same given daytime. As expected, this hour of the day is affected by the highest users’ normalized number NUn_t [-], as shown by Fig. 5.11b. Nevertheless, the effects of users’ exposure in the outdoors seem to be more evident when the number of NR placed indoors is minimized, i.e. during the nighttime and in the late evening, as remarked by the KPI relating to the impact of an event in the OA on the whole population at a given time t, that is IE_t [-] (Fig. 5.11b). Therefore, the analysis of Fig. 5.11a, b should be jointly performed since the two panels and the related KPIs show different aspects of the exposure assessment in the OA. The most crowded scenario refers to t = 18, indeed, when NU_t is about 3900 persons (that is 0.73 persons/m² × 5390 m² of outdoor surface including pedestrian areas and other sites, according to Table 5.4) and the effective number of users placed outdoors, and thus exposed to the attack (summing OO, PO, NR considered as special buildings visitors as in Table 5.1), is about 1940 persons. On the contrary, at t = 23, IE_t = 0.87 but UOd_t = 0.32 persons/m², and thus, the overall number of effective users exposed to the attack is significantly lower.

In addition, Fig. 5.11c then traces the impact of OO, PO, and NR (also distinguishing between visitors of special buildings), so as to point out possible dynamics in the use behaviours. Finally, Fig. 5.11d outlines the percentage of users by age, that is toddlers T (0–4 years), parents-assisted children PA (5–14 years), young autonomous YA users (15–19 years), adult users AU (20–69 years) and elderly users EU (70 + years), showing that they are almost constant over the daytime. This result is essentially affected by the quick assessment approach in users’ vulnerability by age relying on homogeneous statistical data.

In view of the above, it is hence possible to conclude that:

Outcomes of Sects. 5.2, 5.3, and 5.4 have been then merged to provide emergency and evacuation scenarios to be assessed according to behavioural design simulations. Table 5.5 provides the full summary of these scenarios, which both include pre- (Sect. 5.2) and post-retrofit conditions (Sect. 5.3), and which are also characterized by different distribution of users at the start of the evacuation process, and two attack typologies (no attack conditions, thus implying simple OA evacuation, to have a baseline and reference scenario for comparisons, as discussed in Chap. 4, Sect. 4.​4; T2, according to relevance assessment discussed in Sects. 5.2 and 5.3). Table 5.5 also reports simulation codes then used in the following discussion of results.

Simulations have been performed according to the model defined in Chap. 3, Sect. 3.​4, and mainly using validated tools developed under the BE S²ECURe project [6]. In particular, the model has been implemented in Netlogo 6.2.0 [7]. Simulation results have been then analysed according to the behavioural KPIs reported in Chap. 4, Sect. 4.​4.

Figure 5.12 shows the evacuation curves for the simulated scenarios. In detail, Fig. 5.12 groups simulation curves by users’ initial position and thus attack points without effects of the attack (simple evacuation) in pre-retrofit conditions, while Fig. 5.12b, c compare pre- (see Fig. 5.8) and post-retrofit (Figs. 5.9 and 5.10) scenarios, with and without attack effects, depending on the users’ initial position and thus attack points.

3 line graphs represent an increasing trend of the users arriving at a safe area with the rise in time. a. The lines represent H N o pre, F N o pre, A S 2 N o pre, and A S 1 N o pre. b. The lines include H N o pre, A S 1 T 2 pre, and A S 1 t 2 S T 1. c. The lines include H N o pre, A S 2 N o pre, A S 2 T 2, A S 2 T 2 S T 1, and A S 2 T 2 S T 2.

The same rationale in comparisons is provided by Table 5.5, which shows the main KPIs provided in Chap. 4, Table 4.​6, that are: the normalized evacuation time at the 95^th percentile of arrived users—TN95 [-]; the normalized flows at the 95th percentile of arrived users—FN95 [-]; the normalized number of physical contacts among the users—PN [-]; the casualty ratio—CR [-]; and the not-arrived users’ ratio—NA [-]. Selected comparisons are provided to evaluate the impact of the users’ position on the overall evacuation process, and then to assess how the combination of attack typology and mitigation strategies can affect users’ safety.

In particular, in the scenarios without the attack effects on the crowd, and considering pre-retrofit conditions, the evacuation process seems to be quicker when users are placed in the southern part of the OA (AS1-No-Pre), since they are placed closer to square exits (Fig. 5.12a). As expected, when visitors are focused in front of the special buildings (F-No-Pre), TN95 increases up to + 25% with respect to the scenario with a homogeneous distribution of users, in view of the combination between crowd effects and path length. In this case, users should organize movement in overcrowding, thus also increasing FN95, which corresponds to a risk reduction since flows are far from being optimized.

When effects of the attack are present, and considering pre-retrofit conditions, CR and NA increase, as expected. The scenario characterized by the visitors’ main distribution in the northern part of the OA (AS1-T2-Pre) seems to be generally riskier than the one with users placed in the southern part (AS2-T2-Pre), essentially in view of the same issues on users’ paths and interactions which can be noticed in no attack effects conditions (see positive Percentage Variation PV [%] values related to TN95, FN95 and, mainly, CR in Table 5.5).

Nevertheless, physical contact is less relevant (see PV [%] related to CR in Table 5.5), and the number of users who can arrive to a safe area and do not stop the evacuation inside the square (e.g., nearby obstacles, or in temporarily protected areas) increases, too (see PV [%] reduction related to NA in Table 5.5). These phenomena could be linked to the wider area in which users are initially placed as well as to the effects of the obstacles in the square (compare Fig. 5.5). In AS1-T2-Pre, these conditions make users organize evacuation, while, in AS2-T2-Pre, physical contact among users is relevant at the starting of the process, and the obstacles placed near the Palombaro represent a protection area while users are moving towards the OA exits placed in the northern part of the OA itself.

Post-retrofit scenarios achieve a significant decrease in casualties in all the conditions, with respect to the related pre-retrofit conditions, as shown by CR decrease in Table 5.5. Similarly, NA decreases when a LEA’s point is implemented in the OA, since users are more attracted by it rather than by obstacle protection. The best improvement of safety relates to AS1-T2-ST1, essentially since the LEAs’ point is close to the attack area (see Table 5.5). As expected, evacuation times and curves are similar for the cases in which users are initially placed near the Loggia and the Palombaro, while they significantly vary in the AS2 scenarios (Fig. 5.10) where users are initially distributed in the southern part of the square. In particular, in AS2-T2-ST1 and AS2-T2-ST2, supporting main emergency guidance towards the northern part of the square increases both TN95 and FN95 (see Table 5.5). The related evacuation curve is composed of two main parts, indeed, as shown by Fig. 5.12c: the first one (from 0 to about 25 s) concerns the arrival of users placed near the southern exits of the OA, while the second one (from 25 s) concerns the arrival of users towards the northern OA exits. Nevertheless, introducing the LEA’s point in the square reduces the evacuation timing, as shown in Table 5.5 and Fig. 5.12c. Finally, it is worth noting that physical contact among users strictly depends on the initial position of the users when implementing mitigation strategies. In AS1-T2-ST1, PN increases, essentially in view of the great attractiveness of a unique point for the users, in the centre of the northern part of the OA. In both AS2-T2-ST1 and AS2-T2-ST2, on the contrary, PN decreases, essentially in view of the organized movement of users in a single direction, which reduces possible counterflow effects (Table 5.6).