A Numerical Study for Assessing the Risk Reduction Using an Emergency Vehicle Equipped with a Micronized Water System for Contrasting the Fire Growth Phase in Road Tunnels

Abstract

We performed Computational Fluid Dynamics (CFD) modeling, and simulated a people evacuation process from a tunnel in the event of a fire, for evaluating the potentialities of using, as a temporary safety measure, an emergency vehicle equipped with a micronized water system for contrasting the fire growth phase. The structure investigated is a one-way road tunnel with only natural ventilation, and with a length less than 1000 m. The tunnel is assumed at present to be affected by refurbishment works for making it comply with the minimum safety requirements of the European Directive 2004/54/EC. In particular, it is considered that it has not yet been provided with hydrants, and with the sidewalks and the emergency exit which are still under construction. This means that users are forced to use the road carriageway for escaping from the tunnel if a fire occurs. The CFD findings have shown that the use of the micronized water system might lead to a significant improvement in the environmental conditions along the escape route since the tenability limits of temperature, radiant heat flux, CO and CO₂ concentration were found to be better satisfied. Additionally, the visibility distance was shown to increase, even though it was found to be higher than the acceptable threshold value only in a few cases. However, the quantitative risk analysis based on a probabilistic approach, which was combined with a method currently used in Europe for assessing the risk due to the transit of only dangerous goods, shows that the final cumulative F-N curves related to the micronized water system always lie below those without the mentioned system, and in addition, they are always contained within the limits of the ALARP region. It is to be stressed that our paper might represent a reference in showing the effectiveness of the micronized water system as a temporary safety measure. However, it is desirable that the Tunnel Management Agencies accelerate the refurbishment works for making road tunnels definitively safer for users in a short period of time.

1. Introduction

After catastrophic events due to fires occurred in some road tunnels in Europe (i.e., Mont Blanc tunnel, Tauern tunnel, and Saint Gotthard tunnel), the European Parliament and Council adopted the Directive 2004/54/EC [1]. This Directive introduced the minimum safety requirements, for tunnels belonging to the Trans-European Road Network with lengths of over 500 m, with the purpose of ensuring an appropriate safety level to tunnel users. The minimum safety requirements regard both structural measures and equipment which are applicable for both the design of new tunnels and the existing ones. In the case of tunnels already open to traffic, the compliance of the existing safety measures with the requirements of the Directive must be assessed, and in the event of a negative response, refurbishment works must be made for adapting the tunnel to the provisions mentioned that are mandatory or recommended. The Italian Ministry of Infrastructure and Transports adopted the mentioned Directive in 2006 [2], and fixed, for the existing tunnels, the date of 30 April 2019 as the deadline within which the refurbishment works were to be completed. At the present time, due to the very high density of tunnels in Italy, most of these structures still present works in progress. For this reason, the Italian Tunnels Permanent Commission has recently imposed on the Tunnel Management Agencies (TMAs) the implementation of certain temporary safety measures for tunnels that have not yet achieved the minimum safety requirements. Among these temporary measures, a crucial role for people safety is played—more especially if hydrants and/or extinguishers inside the tunnel have not yet been installed—the presence of rescue teams stationed in the proximity of the tunnel portals for emergency services, operating 24 h a day and seven days a week. The emergency service team is foreseen to be trained by qualified personal for fighting a fire, and appropriately equipped for containing the fire growth. The intervention time of the rescue team for this purpose, for example in the case of a tunnel open to traffic with a length < 1000 m and without any emergency exit (except the entrance and exit portals of vehicular traffic), is required to be less than 10 min from the fire’s start.

Indeed, the effectiveness of the equipment generally used by the cited emergency staff for contrasting a fire in a tunnel is not hitherto clearly known. This represents a gap of knowledge that the present paper intends to fill.

The emergency procedure involves, in general, the use of a four-wheeled vehicle (e.g., a pick-up van) equipped with a fire-extinguishing technology based on a micronized water system; moreover, this vehicle is often provided also with certain portable extinguishers. However, the main issues that appear still to be unsolved are: (i) fire size on which the mentioned micronized water system may be effective; (ii) time within which to act for contrasting the fire; (iii) consequent environmental conditions in the tunnel (i.e., temperatures, radiant heat fluxes, visibility distances, and CO and CO₂ concentrations); (iv) effects on tunnel users during their evacuation process in terms of toxic gases and heat exposure; (v) relationship between the annual cumulative frequency (F) and the number of potential fatalities (N) that might be expected; (vi) risk acceptability.

The mentioned lacuna of knowledge is the major driving motivation for the authors of the present paper in investigating in greater depth the potentialities of using a micronized water system, as a temporary safety measure, for containing the fire growth phase in road tunnels before the arrival of the fire brigade.

The article is organized as follows: the next section contains a literature review regarding studies on the use of water mist systems for fighting a fire. Then, the Computational Fluid Dynamics (CFD) modeling is presented and implemented. Consequently, the results are presented and discussed, and comparisons are made between with and without the use of a micronized water system for contrasting fires. Subsequently, a Quantitative Risk Analysis (QRA) based on a probabilistic approach is set up for building the corresponding F-N curves, and the risk reduction in terms of potential fatalities is quantified. Finally, conclusions and further developments of research are addressed.

2. Literature Review

In the last few years, several studies have been made on the evaluation of the effectiveness of fire-fighting safety measures based on the use of water mist systems (i.e., a protection system that controls fire by projecting ultrafine droplets) in road tunnels.

Rein et al. [3], for example, developed an analytical model to predict the trajectories of water mist droplets in tunnels. They found that, unlike the conventional sprinklers, water mist systems are highly affected by the ventilation velocity inside the tunnel due to the much smaller size of droplets. Under current ventilation in the event of a fire, with the nozzles located at the ceiling, the droplets may reach the fire location covering a distance even longer than 50 m, while a very higher ventilation velocity might carry the droplets beyond the fire location. The results suggested that additional research concerning the interaction between the water mist system and tunnel ventilation was required.

Chen et al. [4] conducted a series of small-scale experiments for investigating the effectiveness of water mist systems in the event of a fire in a tunnel under different longitudinal ventilation velocities. According to the results obtained, a suitable design regarding both the water mist system and longitudinal ventilation in the tunnel might significantly increase the success of fire suppression.

Nmira et al. [5] developed a numerical model for assessing the capability of a water mist system in mitigating thermoplastic fires in a tunnel. They carried out a parametric study by varying the water flow rate under longitudinal ventilation velocities inside the tunnel between 1.5 and 2 m/s. The results showed that the efficiency of the water mist system in mitigating the fire in the tunnel depends on fire characteristics, ventilation velocity, nozzle location, water flow rate, and droplet size.

Qin and Chow [6] reported some experimental data on water mist systems in suppressing fires in confined spaces. High-pressure water mist systems extinguished fires more effectively: in particular, the time required to extinguish a 1.6 MW diesel oil spray fire was found to be 40 s for an operating pressure of 6 MPa.

Blanchard et al. [7], by performing CFD simulations, reproduced a fire test with the scope to show the interaction between a water mist system and hot gases in a tunnel with longitudinal ventilation. A water mist system consisting of six nozzles with an operating pressure of 90 bars and a water flow rate of 5.5 litres/min was implemented. The CFD code accurately predicted the reduction of gas temperatures and heat fluxes due to the water mist system activation.

Beard [8] developed a theoretical model for assessing the effectiveness of a water mist system on major fire spread in a longitudinally ventilated tunnel. The author showed that the efficiency of the suppression system depends on several factors, such as nozzle design and activation method, water discharge rate density, the spatial distribution of droplets, and ventilation velocity within the tunnel. However, he stressed that if a fire is relatively small when the water mist system is activated, it may, maybe, be extinguished or never reach a critical value of the Heat Release Rate (HRR).

Liang et al. [9] proposed a system that combines water mist screens and transverse ventilation in the event of a tunnel fire. The water mist screens are used, in particular, to confine smoke from spreading widely. The effectiveness of the proposed system was evaluated using the Fire Dynamics Simulator (FDS). The results showed that the hot smoke due to a fire can be effectively confined in the region between the water mist screens, and as a result, a safer environment for user evacuation is expected.

Yang et al. [10] conducted a series of reduced-scale tunnel fire experiments for investigating the influence of the water mist particle size, in the event of a fire in a tunnel, on smoke layer, temperature, and CO concentration under natural or mechanical ventilation conditions. They found that the particle size has a significant influence on the thickness of smoke layer and distribution of temperature, but it has less effect on CO concentration.

Klaffenböck and Gertl [11] carried out CFD simulations for assessing the effects, expressed in terms of Fractional Effective Dose (FED), of a high-pressure water mist system on users during a tunnel fire. They found that the FED at breathing height (2 m) corresponding to the high-pressure water mist system is less than 1 (FED = 0.3), while it exceeds the unit in the absence of an automatic fire-fighting system.

Li et al. [12] presented the results of reduced-scale tunnel experiments for showing the effectiveness of a water mist system in blocking fire-induced smoke and heat. Tests were carried out for different HRRs, nozzle numbers, water pressures, and longitudinal ventilation velocities. A method for qualitatively analysing the effects of the water mist system was also proposed. The authors suggested that a water mist system should not be used in combination with longitudinal mechanical ventilation, but to limit its application to naturally ventilated tunnels.

Sarvari and Mazinani [13] developed a new system, different from the conventional water mist one, for fighting tunnel fires. This is based on the use of camera images to determine the fire coordinates, and jet fans with nozzles to generate water mist. The performance of this system was assessed by means of FDS simulations. According to these authors, the proposed system allows one to reduce the fire suppression time and maximum HRR when compared to the conventional water mist system, as well as to improve the field of vision.

The above chronological literature review shows that the scientific articles have focused prevalently on evaluating the effectiveness of certain water mist systems installed inside tunnels for fighting a fire. Therefore, as far as the authors of the present manuscript are aware, studies that investigate the effects of emergency vehicles (e.g., a pick-up van), stationed in the proximity of the tunnel portals, and equipped with a micronized water system for controlling the fire growth phase in road tunnels have hardly ever been investigated. This is a gap in our knowledge that the present paper intends to bridge.

3. Materials and Methods

3.1. Description of Tunnel Characteristics

A two-lane unidirectional motorway tunnel was investigated. It is less than 1000 m (i.e., 850 m) long, straight, flat, with two sidewalks, and without emergency lanes. The tunnel is assumed to be open to traffic, with the refurbishment works that are still in progress for making it comply with the minimum safety requirements of the Directive [1]. In particular, the hydrants inside the tunnel for fighting a fire have not yet been provided, which implies the need to verify the potentiality of another fire-fighting system that should be temporarily provided (e.g., the presence of rescue personnel at the tunnel portals with an emergency vehicle equipped for contrasting fires). The two sidewalks, which will be used for the people evacuation process from the tunnel in the event of a fire, have not yet been completed, and thereby at this stage of works they are not yet available as escape routes for tunnel users. The emergency exit, located in the middle of the tunnel length, is also supposed to be temporarily not available being still under construction. The circumstances mentioned may often be encountered with reference to Italian motorway tunnels that have not yet achieved the minimum safety requirements; more especially if it is not possible to close the one-way tunnel tube by using the adjacent tube for bi-directional traffic (e.g., this might happen if the other tube is also affected by works) and/or because there are no alternative traffic itineraries.

Figure 1 shows the cross-section of the investigated tunnel, with the presence of safety barriers for protecting men at work from traffic conflicts.

The tunnel is not equipped with a mechanical ventilation system but is only naturally ventilated. In order to simulate, in our numerical study, worse fire scenarios in the tunnel investigated, we almost neglected the influence of natural ventilation due to the piston effect of vehicles in motion. In fact, only a pressure difference of 0.5 Pa was assumed to be applied between the entrance portal (portal A) and the exit one (portal B).

The tunnel walls are made of currently used concrete and with a thickness of 0.4 m. Since the thermal properties of the concrete employed were not known, these were taken from the literature. In this respect, Schrefler et al. [14] consider, for common concrete such as in our case, that the thermal conductivity is 1.67 W/m/K, the specific heat is 0.94 kJ/kg/K, the density is 2585 kg/m³, and the emissivity is 0.9.

3.2. Traffic

Different peak traffic flows, expressed in terms of hourly volumes per lane, that can affect the investigated tunnel were considered. In particular, the following three peak hour volumes (VHPs) were assumed: 1000, 1500, and 2000 vehicles/hour per lane. The maximum VHP is slightly less than the value of capacity (C = 2400 vehicles/hour per lane) provided by the Highway Capacity Manual [15] for similar roads (i.e., freeways); therefore, the maximum ratio VHP/C is 0.83. The percentage of Heavy Goods Vehicles (HGVs) was assumed to be equal to 25% in all cases investigated.

3.3. Fire Scenarios

Five fire scenarios were considered in the present paper. They involve two cars, a van, a bus, and two different types of HGVs, respectively. Each fire source is assumed to be positioned in five different points along the tunnel length: 145 m, 280 m, 420 m, 570 m, and 710 m from the entrance portal (i.e., portal A). Fire scenarios for the cited vehicles, which are geometrically schematized as parallelepipeds in our numerical analysis, are described in terms of both maximum Heat Release Rate (HRR_max) and time to reach the maximum HRR (t_max). In particular, we considered fires with HRR_max of 8, 15, 30, 50, and 100 MW for the two cars, the van, the bus, and the two types of HGVs, respectively; and the corresponding t_max are: 5, 7, 9, 10, and 10 min. It is to be stressed that for each fire curve a linear law is considered for the fire growth phase (e.g., HRR_max = 8 MW is achieved according to a linear increase after t_max = 5 min from the fire’s start, HRR_max = 15 and 30 MW are achieved according to linear increases after t_max = 7 and 9 min from the fire’s start, respectively, etc.); it is then followed by a constant HRR_max phase. Details regarding the assumptions made for the yields of combustion products are reported in Caliendo et al. [16,17].

3.4. Emergency Vehicle Equipped with a Micronized Water System

Since the hydrants inside the investigated tunnel are assumed not to have yet been installed, certain temporary measures play a fundamental role for people’s safety in the event of a fire with the tunnel open to traffic. In this respect, in Italy, the presence of a rescue team is generally provided for stationed in proximity to the tunnel portals for emergency services, operating 24 h a day and 7 days a week. The emergency staff is expected to be trained by qualified personnel and appropriately equipped for contrasting fires.

In this regard, in our paper, we made provisions for the use of a four-wheeled vehicle (e.g., a pick-up van) equipped with a technology based on a micronized water system (nominal tank capacity equal to about 500 litres) for contrasting the fire growth phase, and, more especially, the system sprays high-pressure micron-sized water droplets. Foaming agents may also be added to the water for improving the fire-fighting capability. The use of a micronized water system may lead to a high evaporation rate with effective cooling and suffocation of the fire growth phase. By means of the micronized water system, a smaller amount of water may be needed to control a given fire compared to conventional water mist systems. Figure 2 shows a typical emergency vehicle equipped with a micronized water system. One can note that the emergency vehicle is provided also with 4 portable extinguishers for increasing the probability of success in fighting a fire if it were necessary. In our study, we referred to this type of vehicle.

3.5. Modified Fire Curves

Given the reduced capacity (i.e., 500 L) of the tank containing the micronized water, it is logical to assume that the effectiveness of the mentioned system, even if it were supported by the use of the four portable extinguishers, may interest prevalently small and/or medium-sized fires such as those characterized by HRR_max of 8, 15, or 30 MW. The effectiveness of the system on bigger and/or very large-sized fires such as those with HRR_max of 50 and/or 100 MW is, instead, assumed conservatively to be negligible.

However, the main function of the micronized water system is to contrast the fire growth phase by trying to prevent the achievement of the maximum value of HRR that is expected for a given fire. In general, it is assumed that this system may be effective when it is activated before that the fire growth achieves values of HRR more than 8 MW.

In the light of the above considerations, we investigated the cited fires characterized by HRR_max of 8, 15, and 30 MW, and assumed for each of them that the micronized water system is activated before the corresponding fire growth phase achieves 8 MW. In particular, the micronized water system is assumed to modify the original fire curves (i.e., those of HRR_max equal to 8, 15, and 30 MW) by intercepting the linear law of the fire growth phase at 8, 6, or 4 MW. Figure 3, for example, with reference to the original fire curve of HRR_max of 30 and 15 MW, respectively, shows how each of them is modified by the micronized system activated at 8, 6, and 4 MW (i.e., points A, B, and C). One can note that each modified fire curve is still linear till the mentioned points, then is constant over time. The coordinates of these points are: A (HRR = 8 MW, t = 2.5 min), B (HRR = 6 MW, t = 2 min), C (HRR = 4 MW, t = 1.5 min) with reference to the case of HRR_max = 30 MW; A (HRR = 8 MW, t = 4 min), B (HRR = 6 MW, t = 3 min), C (HRR = 4 MW, t = 2 min) relatively to HRR_max = 15 MW; finally was found: A (HRR = 8 MW, t = t_max = 5 min), B (HRR = 6 MW, t = 4 min), C (HRR = 4 MW, t = 2.5 min) for the HRR_max = 8 MW. In other words, the time within which the micronized water system has to be activated from the fire’s start should be less than 5 min for all investigated cases.

In the aforementioned figures, the estimated time is also indicated for the fire brigade to get to the tunnel investigated, to organize the fire extinguishing procedure, and to start the definitive suppression of the fire.

3.6. Research Framework

The present paper is set in the field of research on the effectiveness of temporary safety measures based on the use of an emergency vehicle equipped with a micronized water system in the event of tunnel fires, but extends the state-of-the-art by means of the development of a numerical fluid dynamics analysis for assessing the risk reduction in the function of the modified fire curves, and for contributing to the advancement of a quantitative risk method aimed at assessing the number of potential fatalities. Therefore, the present paper can serve, given the gap in our knowledge of micronized water systems, as a possible reference for the tunnel operators during refurbishment works, and to increase our experience in the context of fire safety engineering.

The methodology applied is briefly reported in Figure 4.

4. CFD Modeling

4.1. Fluid-Dynamics Analysis Description

CFD modeling was carried out in our study for simulating tunnel fire scenarios, certain applications of which can also be found more especially in Caliendo et al. [18,19,20], and Khattri et al. [21]. CFD modeling is based on the finite volume method for discretizing the tunnel in elementary cells, each cell is then solved by using the fundamental equations (i.e., conservation of mass, momentum, and energy) coupled with physical models describing the combustion, turbulence, and thermal radiation phenomena. The accuracy of results, expressed in terms of temperature, smoke spread, and toxic gas concentration, depends mainly on the characteristics of fire, physical models employed, and mesh optimization.

Among CFD codes, we used the commonly employed FDS version 6.7.3 that is an open-source software developed by the National Institute of Standards and Technology (NIST), the characteristics of which with the corresponding physical models implemented are more especially discussed in the FDS user’s guide [22]. Tunnel geometry, location and size of fire, HRR growth law, number and distance between the vehicles in the queue upstream of the fire, yields of combustion products, and the pressure difference between tunnel portals for simulating the ventilation within the tunnel, are some of the main input data to implement the FDS code.

4.2. Validation and Mesh Fineness

In the present study, the FDS code was preliminarily validated by comparing the results of simulations with the experimental data in terms of temperature contained in Xue et al. [23]. In this regard, we found a good level of conformity between the experimental and predicted temperatures (i.e., an error of no more than 5%).

Once the FDS tool was validated, a grid sensitivity analysis was carried out in order to identify an adequate mesh fineness. The results of this analysis, detailed in Caliendo et al. [16,17], showed that the cubic cells of 0.4 m side were appropriate for correctly discretizing the investigated tunnel; in particular, 1,095,255 cells were used.

5. Evacuation Modeling

5.1. Simulation Tool

The people evacuation process was simulated using the agent-based egress calculation module of FDS, known as Evac [24]. This code treats each agent as an individual user with specific characteristics (e.g., walking speed) and escape strategy (e.g., choice of exit and escape route). Evac permits evacuees to move in a 2D space and considers the fact that environmental conditions resulting from a fire (i.e., concentration of toxic substances) can affect user behaviour such as walking speed. The code estimates the number of potential victims based on their exposure to toxic gases only. The initial position of tunnel users when the fire starts (i.e., t = 0), the number of users within each queued vehicle, walking speed, pre-movement time, and escape direction are the main input data to implement the Evac code.

It is to be stressed that Evac does not take into account the effects of temperature and radiation on evacuees, which were computed in our study according to the procedure reported in DiNenno et al. [25].

5.2. Queue Formation and Evacuation Process

Evac simulations were carried out for each position occupied by the five investigated burning vehicles (characterized by a HRR_max of 8, 15, 30, 50, and 100 MW) along the tunnel length (145, 280, 420, 570, and 710 m from portal A). The same was done with reference to the aforementioned modified fire curves of the HRR_max of 30, 15, and 8 MW in order to show the effects of using the micronized water system. However, in all the investigated cases, the number of vehicles queued upstream of the fire source was computed according to the following assumptions: (i) queued vehicles stop without going beyond the burning vehicle; (ii) vehicles queue up filling both lanes; (iii) the distance between the burning vehicle and the first queued vehicle is 10 m; (iv) the distance between the queued vehicles is 1 m.

The frequency per time unit with which vehicles enter the tunnel was assumed to be stationary. This means that the arrival time of the first vehicle upstream of the fire source is computed to be 3.6, 2.4, and 1.8 s from the fire’s start for the three investigated peak hour volumes (VHPs) of 1000, 1500, and 2000 vehicles/hour per lane, respectively. Therefore, the last user who leaves the tunnel in the event of a fire, walking towards a safe place, will be affected by a delay of 1.2 or 1.8 s when the VHP is equal to 1000 vehicles/hour per lane compared to the case in which VHP is of 1500 or 2000 vehicles/hour per lane, respectively.

The users’ pre-movement time (i.e., the sum of the detection and reaction time) for leaving their own queued vehicle upstream of the fire was assumed to be 1.5 min when HRR_max equal to 8, 15, 50, and 100 MW; while it increased by 1 min (i.e., 2.5 min in total) in the event of a fire with HRR_max equal to 30 MW by considering the additional time required to get off the bus [26].

The analysis of the consequences on users was made after 10 min from the fire’s start, which means that after this period the longitudinal part of the tunnel upstream of the fire is full of vehicles in all the aforementioned three conditions of traffic volumes (i.e., VHPs), and with a distance between the vehicles queued of 1 m only.

According to the mentioned assumptions, the number of vehicles in the queue upstream of the burning vehicle is the same with reference to the three VHPs and is equal to 36, 74, 114, 158, and 198 when the fire source is at 145, 280, 420, 570, and 710 m from portal A, respectively. The corresponding number of users potentially at risk was computed by considering that on average there are 2 people in each equivalent vehicle in the queue (for more details see Caliendo et al. [16], in which we assumed that the occupancy rate of cars, HGVs, and buses were equal to 1.7, 1, and 30 people, respectively). As a result, the number of people escaping upstream of the fire is 72, 148, 228, 316, and 396 when the fire source is at 145, 280, 420, 570, and 710 m from portal A, respectively.

Figure 5 shows a schematic layout of vehicles in the queue upstream of the fire source for the investigated unidirectional road tunnel. It is to be remembered that the two sidewalks and the emergency exit are assumed to be still under construction. Consequently, each user, initially positioned next to his/her own vehicle, is forced to walk on the road carriageway by using the physical space between the two queues of stopped vehicles or the space between each queue of stopped vehicles and the safety barrier for reaching a safe place, which, in the circumstances examined, is represented by the traffic entrance portal only (i.e., portal A). The presence of these queued vehicles acts as an obstacle for the movement of evacuees, reducing their walking speed to a value assumed to be on average of 0.3 m/s.

6. Analysis of Results

6.1. Longitudinal Profiles

Longitudinal profiles of temperature, radiant heat flux, toxic gases concentration, and visibility distance at breathing height (2 m) along the escape path on the road carriageway are compared with the corresponding acceptability criteria in order to verify the tenability conditions for the survival of users during a fire in the tunnel investigated both without and with the presence of the emergency vehicle equipped with a micronized water system for contrasting the growth phase of certain fires (i.e., HRR_max of 8, 15, or 30 MW).

6.1.1. Temperatures

Figure 6 shows the longitudinal temperature profiles at a height of 2 m along the escape route on the road carriageway after 10 min from the start of the fire when the burning vehicle is at 420 m from portal A. Considering a walking speed of 0.3 m/s and a pre-movement time both of 1.5 min (i.e., for the two cars, the van, and the HGVs fire scenarios; black user) and 2.5 min (i.e., for the bus fire scenario; violet user), and assuming that VHP = 1000 vehicles/hour per lane (i.e., the worst condition for the last escaping user given that he/she is affected by a delay of 1.2 or 1.8 s compared to the case of VHP equal to 1500 or 2000 vehicle/hour per lane, respectively), Figure 6 reports that the position occupied by the last user escaping from the tunnel towards portal A in the event of a fire is: 250 m from portal A for the HRR_max of 8, 15, 50, and 100 MW, as well as 270 m for the HRR_max of 30 MW. From Figure 6 one can see that the mentioned user is always exposed to a temperature below the tenability limit of 60 °C [27].

6.1.2. Radiant Heat Flux

Figure 7 shows the corresponding longitudinal radiant heat flux profiles at a height of 2 m along the escape route on the road carriageway after 10 min from the fire’s start when the burning vehicle is at 420 m from portal A. Independently of the fire scenario considered, from Figure 7 one may note that the last evacuating user is exposed to a radiant heat flux lower than the tenability limit of 2 kW/m² [27].

6.1.3. CO₂ Concentration

Figure 8 shows the longitudinal profiles of CO₂ concentration at a height of 2 m along the escape route on the road carriageway after 10 min from the fire’s start when the burning vehicle is at 420 m from portal A. From Figure 8 it is possible to observe that the last escaping user is exposed to a CO₂ concentration lower than the tenability limit of 40,000 ppm [28] in the event of 8, 15, 30, and 50 MW fires, while he/she (black user) is exposed to a CO₂ concentration slightly higher than 40,000 ppm in the case of a 100 MW fire.

6.1.4. CO Concentration

Figure 9 shows the longitudinal profiles of CO concentration at a height of 2 m along the escape route on the road carriageway after 10 min from the fire’s start when the burning vehicle is at 420 m from portal A. From Figure 9 one may note that the last evacuating user is exposed to a CO concentration below the tenability limit of 1200 ppm [28] in the event of 8 and 15 MW fires, while he/she is exposed to unsafe conditions for 30, 50, and 100 MW fires.

6.1.5. Visibility Distance

Figure 10 shows the longitudinal visibility distance profiles at a height of 2 m along the escape route on the road carriageway after 10 min from the fire’s start when the burning vehicle is at 420 m from portal A. From Figure 10 one can see that the mentioned user always has visibility lower than the tenability limit of 10 m [27]. As a result, he/she may be exposed to unsafe conditions due to his/her incapacity to distinguish evacuation signs present along the escape route.

Summing up, in the event of a fire within a tunnel open to traffic without both hydrants and the availability of sidewalks and emergency exits due to them still being under construction, not equipped with a micronized water system as a temporary measure for contrasting fires, we found for the last evacuating user that: (i) the CO₂ concentration is slightly higher than the tenability limit of 40,000 ppm in the case of HRR_max equal to 100 MW; (ii) the CO concentration exceeds 1200 ppm in the event of a fire with HRR_max of 30, 50, and 100 MW; (iii) the visibility distance is lower than the acceptable limit of 10 m in all the fire scenarios investigated.

It is to be stressed that similar evacuation conditions were also found when the burning vehicle is assumed to be located at 145, 280, 570, and 710 m from portal A. However, in order to save space, the corresponding results are not reported in the present paper.

In light of the above results, the temporary use of an emergency vehicle equipped with a micronized water system might improve safety conditions for tunnel users in the event of a fire.

6.2. Effects of the Micronized Water System

It is to be remembered that the micronized water system was assumed to modify the original fire curves characterized by HRR_max of 8, 15, and 30 MW, by intercepting the linear law of the fire growth phase at 8, 6, or 4 MW (i.e., points A, B, and C of Figure 3). For assessing the effectiveness of the cited temporary measure in reducing the consequences on tunnel users due to a fire, the corresponding longitudinal profiles of temperature, radiant heat flux, toxic gases, and visibility distance at breathing height (2 m) along the escape route on the road carriageway were computed by means of the FDS code. The results are reported in the subsequent paragraph and compared with those obtained for the corresponding unmodified fire curves.

Temperature, Radiant Heat Flux, CO₂ Concentration, CO Concentration, and Visibility Distance

Figure 11, Figure 12, Figure 13, Figure 14 and Figure 15 show, respectively, the longitudinal profiles of temperature, radiant heat flux, CO₂ concentration, CO concentration, and visibility distance at a height of 2 m along the escape route on the road carriageway after 10 min from the fire’s start when the burning vehicle is at 420 m from portal A, under the hypothesis that the micronized water system is activated. These figures show a significant reduction in temperatures, radiant heat fluxes, CO₂ and CO concentrations, and an important increase in the visibility distance for escaping users in the event of fire compared to the results of the original fire curves characterized by HRR_max of 8, 15, and 30 MW, respectively.

In particular, it is to be stressed that the tenability limits of temperature, radiant heat flux, and CO₂ concentration, are better verified in all cases investigated.

When the micronized water system is activated, in contrast with the results of the previous Section 6.1.4, the CO concentration is within the tenability limit of 1200 ppm also with reference to the fire characterized by HRR_max of 30 MW (i.e., Figure 14).

With reference to the visibility distance, even if significant increases regarding the values of this parameter are found, the last user escaping from the tunnel might still find a visibility distance lower than the threshold value of 10 m, except for the case in which the original fire curve with HRR_max equal to 8 MW is modified at 4 MW by the micronized water system (Figure 15).

Given that the results obtained appear to show that the visibility distance would not be more than 10 m in all investigated cases, it is necessary to indicate a criterion by which to concatenate all the results obtained showing the global effectiveness of the micronized water system. In this respect, the use of the number of potential fatalities and a comparison with the corresponding one in the absence of the mentioned system seems to be suitable for this aim.

6.3. FED Results by Using the Micronized Water System

The number of potential users exposed at risk of fatalities and/or injuries in the event of a fire in the investigated tunnel was evaluated by means of the Fractional Effective Dose (FED) concept, for which an escaping user is considered to be incapacitated and unable to reach a safe place when the FED due to the exposure of toxic gas concentration (FED_{toxic gases}) and/or due to the heat and radiant heat flux (FED_heat) is more than 1 (the analytic expressions both of the FED_{toxic gases} and FED_heat are reported in Caliendo et al. [16]).

Table 1. Number of potential fatalities computed by using the FDS + Evac with and without a micronized water system.

Distance from Portal A [m]	Fire Scenarios HRRmax [MW]	Number of Potential Fatalities Computed by Using the FDS + Evac
		Without a Micronized Water System	With a Micronized Water System. Fire Growth Curves Controlled at:
			8 MW	6 MW	4 MW
145	8	0	-	0	0
	15	2	0	0	0
	30	10	4	2	0
270	8	2	-	0	0
	15	7	3	0	0
	30	28	8	4	2
420	8	8	-	4	0
	15	16	8	6	3
	30	50	17	12	9
570	8	12	-	6	3
	15	25	11	7	5
	30	59	21	15	12
710	8	14	-	7	5
	15	36	13	9	7
	30	67	24	21	17

shows the number of potential fatalities computed by using the aforementioned FED_{toxic gases} (given that FED_heat was less than 1) for each longitudinal position within the tunnel (145, 280, 420, 570, and 710 m from portal A) assumed for the burning vehicle that is characterized by HRR_max of 8, 15, and 30 MW, respectively. From Table 1 one may note that the number of potential fatalities is significantly reduced when the micronized water system is activated. In fact, the number of potential fatalities might be between 0 and 24 against the range of 0–67 without the mentioned water system.

7. Quantitative Risk Analysis

7.1. General Framework

Although the findings of the FED_{toxic gases} based on the outcomes of FDS + Evac show certain benefits for the users along the evacuation route on the road carriageway with the use of a micronized water system in the event of a fire in the investigated tunnel, the consequences in terms of probability (or cumulative frequency) of potential fatalities are not yet quantified by also considering the remaining two fire scenarios (i.e., HRR_max of 50 and 100 MW). Moreover, the additional risk due to the transit of vehicles carrying dangerous goods (DGVs) through the tunnel is not yet investigated.

7.2. Probabilistic Approach

A Quantitative Risk Analysis (QRA) based on a probabilistic approach is the best tool for assessing the overall risk reduction in road tunnels due to the use of a micronized water system. The probabilistic approach involves the following steps: choice of objective, characterization of the tunnel system, individualization of what can be dangerous and its probability (or frequency) of occurrence, assessment of damages more especially on human health, implementation of safety measures, quantification of the risk level as the sum of probabilities multiplied by damage, definition of a criterion of acceptable risk, verification that the risk level is within the tenability limit. As a necessary part of this process, there is the event and fault tree, as well as models for evaluating consequences.

The main outcome of a QRA method is the so-called societal risk (i.e., potential number of fatalities in a tunnel per year) that is generally expressed in graphic form by means of F-N curves, where F is the annual cumulative frequency of having a given number of fatalities N. The F-N curve computed for a given tunnel might be used to show the effectiveness of the implementation of safety measures (i.e., the micronized water system in our case) and/or to compare it with the well-known ALARP (As Low As Reasonably Practicable) region, which singles out the threshold values of tolerable and intolerable risk in order to judge whether the tunnel is safe. However, there is no common agreement about the values of the ALARP thresholds. In this respect, the European Directive states that the thresholds of the ALARP region must be defined by each Member State. In Italy, the unacceptable and acceptable risk limits are defined, respectively, by the following function: F = 10⁻¹ × N⁻¹ and F = 10⁻⁴ × N⁻¹, for N ≥ 1.

7.3. Event Tree

The definition of each hazard (i.e., fire with HRR_max of 8, 15, 30, 50, or 100 MW) and the determination of the probability of occurrence of each of them derives from the event tree. It is to be remembered that the event tree is drawn in a graphic form for representing different chronological series of consecutive events beginning from an initial event. In the event tree, the probability (p) is reported of the occurrence of each event given the probability of the initial event. In this respect, we used the event tree reported in ANAS [29]. The initial event is represented by the annual frequency of traffic accidents in tunnels, which is assumed to be characterized as follows: traffic collisions (p = 94.9%) and fires (p = 5%). Fires are considered to involve two groups of vehicles: light vehicles (p = 70%) or heavy vehicles (p = 30%); a fire is considered to be relevant (p = 70% for light vehicles, or p = 20% for heavy vehicles). The relevant fires of light vehicles might involve a HRR_max of 8 MW (p = 2.5%), whereas those of heavy vehicles might entail a HRR_max of 15 (p = 81.5%), 30 (p = 14.5%), 50 (p = 2.5%), and 100 (p = 1.5%) MW.

However, it is to be stressed that the aforementioned event tree does not contain the hazards related to vehicles transporting dangerous goods (i.e., DGVs) that will be subsequently dealt with in the present paper.

7.4. Annual Frequency of Traffic Accidents

Concerning the initial event of traffic accidents, which is mentioned in the event tree, it is to be remembered that we made a parametric analysis with reference to traffic in the investigated tunnel by considering three different peak hourly traffic volumes (i.e., VHP = 1000, 1500, or 2000 vehicles/hour per lane). Since in the Quantitative Risk Analysis (QRA) the use of traffic is required in terms of Annual Average Daily Traffic (AADT), we used the relationship reported in the Highway Capacity Manual [15] for this purpose: AADT = VHP/k, where k is the factor that varies between 0.09 and 0.13 for roads with characteristics similar with that investigated (i.e., freeways); we used k = 0.10 in our case. As a result, we obtained the corresponding AADTs per lane equal to 10,000, 15,000, and 20,000 vehicles/day for the mentioned VHP of 1000, 1500, and 2000 vehicles/hour per lane, respectively.

Given that the average accident rate (i.e., the average number of accidents per 100 million vehicle-kilometres) provided by the Tunnel Management Agency (TMA) for the mentioned tunnel was equal to 43.5 accidents/10⁸ vehicles·km, by using the mentioned AADTs of 10,000, 15,000, and 20,000 vehicles/day per lane, we estimated that the corresponding average annual frequency of traffic accidents (i.e., the average number of accidents per year) was equal to 2.7, 4.05, and 5.4, respectively. Each of these three values was implemented as the initial event in the mentioned event tree.

7.5. Annual Fire Frequency

The annual frequency of each fire scenario simulated using FDS + Evac (i.e., HRR_max of 8, 15, 30, 50, or 100 MW) was calculated by multiplying the mentioned annual frequency of traffic accidents (i.e., 2.7, 4.05, and 5.4 accidents/year for an AADT of 10,000, 15,000, and 20,000 vehicles/day per lane, respectively) by all the intermediate probabilities between the initial event and the final one of the event tree (i.e., with reference to the 15 MW fire and an AADT per lane equal to 10,000 vehicles/day, we computed: 2.7 × 0.05 × 0.3 × 0.2 × 0.815 = 6.60 × 10⁻³).

Table 2. Annual frequency of occurrence of fire scenarios simulated using FDS + Evac.

AADT Per Lane [Vehicles/Day]	Annual Frequency of Occurrence of Fire Scenarios [1/Year]
	8 MW	15 MW	30 MW	50 MW	100 MW
10,000	1.65 × 10−3	6.60 × 10−3	1.17 × 10−3	2.02 × 10−4	1.22 × 10−4
15,000	2.48 × 10−3	9.90 × 10−3	1.76 × 10−3	3.04 × 10−4	1.82 × 10−4
20,000	3.31 × 10−3	1.32 × 10−2	2.35 × 10−3	4.05 × 10−4	2.45 × 10−4

shows the annual fire frequency for the different values of AADT per lane and HRR_max.

It is to be said that to each of the five locations occupied by the burning vehicle along the tunnel length (i.e., 145, 280, 420, 570, and 710 m from portal A) was assigned the same probability of fire occurrence, thus equal to 0.2. As a result, the annual frequency of occurrence of a given fire in each position occupied by the burning vehicle was estimated by multiplying the values reported in Table 2 by 0.2.

7.6. F-N Curves

The outcomes of the FDS + Evac concerning the exposure of evacuating people on the road carriageway of the investigated tunnel (upstream of the fire in direction of portal A) to toxic gases expressed by the FED_{toxic gases}, and the exposure to heat and radiant heat flux expressed by FED_heat according to DiNenno et al. [25], are generally utilized for showing when the tenability limits are exceeded (i.e., FED_{toxic gases} ≥ 1 and/or FED_heat ≥ 1). The risk analysis is carried out taking into account both the aforementioned FEDs of all users and their entire evacuation process (i.e., between t = 0 and t = 10 min after the fire’s start). This is made for each fire (i.e., characterized by HRR_max of 8, 15, 30, 50, and 100 MW; in particular with reference to the HRR_max of 8, 15, and 30 MW both with and without the activation of the micronized water system) and for the five different longitudinal positions of each burning vehicle in the tunnel (i.e., 145, 280, 420, 570, and 710 m from portal A). However, it is to be remembered that for certain combinations of HRR_max and positions of fire, we found that the FED_{toxic gases} is more than 1, while FED_heat is always << 1. As a consequence, we computed the number of potential fatalities that might be caused by the exposure to toxic gases only.

The number of potential fatalities, related to each fire scenario (i.e., HRR_max of 8, 15, 30, 50, and 100 MW fire; and in particular with reference to the HRR_max of 8, 15, and 30 MW with and without the activation of the water mist system) and its location within the tunnel (i.e., 145, 280, 420, 570, and 710 m from portal A), was then associated with the corresponding annual frequency of fire occurrence. Then, we computed the individual cumulative frequency of each fire, and by combining the individual cumulative frequency, we calculated the final cumulative frequency with reference to the people escaping upstream of the fire from the tunnel investigated.

7.7. Dangerous Goods Vehicles (DGVs)

The global risk level in the tunnel is obtained by combining the F-N curves related to DGVs with the F-N curves computed in the previous section related to non-dangerous goods and characterized by HRRmax of 8, 15, 30, 50, and 100 MW, respectively.

In the light of the above considerations, the Dangerous Goods Quantitative Risk Assessment Model (DG-QRAM) version 4.04 [30], developed jointly by PIARC and the Organization for Economic Co-operation and Development (OECD), was employed in the present paper for evaluating the consequences due to accident scenarios associated with DGVs, some applications of which can be found in Caliendo et al. [31,32,33,34]. We used the DG-QRAM in the present paper because in Europe it is considered as one of the most suitable for dealing with dangerous goods. The percentage of DGVs was assumed to be equal to 6% of HGVs.

It is to be said that the code analyses 13 accident scenarios including two scenarios related to non-dangerous goods (i.e., 20 and 100 MW fires), which were not set in the DG-QRAM tool since they were already included in the CFD simulations. Therefore, we implemented 11 accident scenarios of the DG-QRAM that might be grouped in: Boiling Liquid Expanding Vapor Explosion (BLEVE); motor spirit pool fire; Vapor Cloud Explosion (VCE); propane torch fire; chlorine, acrolein, and ammonia releases. However, we extended the mentioned standard scenarios analysed by the DG-QRAM with 3 additional accident scenarios (i.e., fireball, VCE, and flash fire) related to vehicles for the transport of liquid hydrogen (LH₂TVs) through the tunnel (for more in-depth knowledge about the transport of hydrogen through tunnels, see Caliendo and Genovese [35]).

It is to be mentioned that the use of the DG-QRAM permitted expressing the results related to risk level due to the transit of DGVs through the investigated tunnel directly in terms of the F-N curve (i.e., the F-N curve is an output of this software).

7.8. Cumulative F-N Curves

The combination of each F-N curve developed using FDS + Evac with the F-N curve computed for the transport of dangerous goods (which includes also the transport of liquid hydrogen) permitted obtaining the final cumulative F-N curve for estimating the overall risk level of the investigated tunnel. In particular, for each of the three values of the aforementioned AADT per lane (i.e., 10,000, 15,000, and 20,000 vehicles/day), we computed four final cumulative F-N curves for making an appropriate comparison: (i) one F-N curve obtained without modifying the original fire growth phase; (ii) three F-N curves based on the use of the micronized water system provided by the emergency vehicle and activated at 8, 6, and 4 MW (i.e., point A, B, and C of Figure 3) of the fire growth phase characterized in origin by HRR_max of 8, 15, or 30 MW, respectively.

Figure 16 shows the final cumulative F-N curves obtained, as well as the limits of both the unacceptable and acceptable risk levels according to the Italian ALARP criterion. One may note that: (i) in all cases investigated, the F-N curves with the use of the micronized water system always lie below those without the mentioned system; (ii) the risk reduction is greater when the micronized water system is activated at 4 MW of the fire growth phase compared to 6 and 8 MW, which means that for a lower activation time from the fire’s start, a greater number of benefits are expected; (iii) the F-N curves based on the activation of the micronized water system are always within the limits of the ALARP region, while the F-N curves in the absence of this system may reach the unacceptable limit in the case of an AADT per lane of 10,000 vehicle/day (Figure 16a) or may exceed it in the case of an AADT per lane of 15,000 and 20,000 vehicles/day (Figure 16b,c).

Therefore, Figure 16 graphically confirms that the use of a temporary safety measure involving an emergency vehicle equipped with a micronized water system might be effective in reducing the risk level in a tunnel open to traffic while refurbishment works are still in progress (i.e., without hydrants and with the sidewalks and emergency exit that are not available as escape routes because still under construction).

8. Conclusions and Discussion for Future Investigations

This research was mainly motivated by the need to perform a computational fluid dynamics modeling for assessing the effectiveness of a temporary safety measure, which is based on the use of an emergency vehicle equipped with a micronized water system, for contrasting the fire growth before the arrival of the fire brigade in road tunnels that are still involved in refurbishment works necessary for adapting these structures to the provisions of the European Directive [1].

The Fire Dynamics Simulator (FDS) code was used to reproduce the fire of two cars, a van, a bus, and two different types of HGVs (characterized by a HRR_max of 8, 15, 30, 50, and 100 MW, respectively). Each burning vehicle was located at five different points along the investigated tunnel (145, 280, 420, 570, and 710 m from portal A). The traffic flow, expressed in terms of peak hour volume (VHP), was assumed to be 1000, 1500, and 2000 vehicles/hour per lane, which corresponds to an Annual Average Daily Traffic (AADT) per lane of 10,000, 15,000, and 20,000 vehicles/day, respectively.

The one-way road tunnel investigated is naturally ventilated and was assumed to be affected by refurbishment works (i.e., without hydrants and with the sidewalks and emergency exit that are not available for escaping from the tunnel in the event of fire). The use of a micronized water system was assumed to modify the original fire curves characterized with HRR_max of 8, 15, and 30 MW by intercepting the corresponding linear law of the fire growth phase at 8, 6, or 4 MW. This involves that the maximum activation time of the micronized water system should be less than 5 min from the fire’s start.

With reference to the position of the last user evacuating from the tunnel along the escape route on the road carriageway after 10 min from the fire’s start, the FDS simulations showed that with the use of the micronized water system: (i) the tenability limits of temperature, radiant heat flux, and CO₂ concentration are much more verified in all investigated cases; (ii) the CO concentration is within the tenability limit of 1200 ppm also with reference to the fire characterized by HRR_max of 30 MW; (iii) the visibility distance increases, however, the last escaping user might still find a visibility distance lower than the threshold value of 10 m, except for the case in which the fire curve with HRR_max equal to 8 MW is modified at 4 MW.

The exposure to toxic gas concentration expressed in terms of FED_{toxic gases} showed that the number of potential fatalities was significantly reduced when the micronized water system was activated.

A Quantitative Risk Analysis (QRA), based on a probabilistic approach, was set up for building the F-N curves through which the risk reduction due to the use of a micronized water system can be quantified in terms of the cumulative frequency of fatalities. For this aim, first of all, the F-N curves developed by the application of the FDS and its evacuation module Evac (i.e., related to the fire of two cars, a van, a bus, and two different types of HGVs) were computed. Then, using the DG-QRAM method, the F-N curves associated with the transport of dangerous goods were calculated. By combining both the mentioned curves, we obtained the final cumulative F-N curves. It was found that: (i) the F-N curves with the use of the micronized water system always lie below those without the mentioned system; (ii) a greater risk reduction was found when the micronized water system is activated at 4 MW of the fire growth phase compared to 6 and 8 MW; (iii) by the activation of the micronized water system, the F-N curves were found to be always within the limits of the ALARP region, while in the absence of this system they may reach or exceed the unacceptable limit.

It is to be mentioned that the findings obtained might also be extended to other one-way road tunnels with similar characteristics, in which hydrants are not yet provided and sidewalks and emergency exits are temporarily unavailable as escape routes in the event of a fire.

Although the findings show the effectiveness as a temporary safety measure of the emergency vehicle stationed at the tunnel portals and equipped with a micronized water system, a possible extension of this study might also be to investigate the influence of droplets (i.e., size, air speed, and orientation) to make further progress in research.

In addition, it is necessary to underline the necessity that Tunnel Management Agencies accelerate the refurbishment works for making road tunnels definitively safer for users in a short space of time.

Generally speaking, it is to be said that the prevalent topics in the field of tunnel safety engineering are fire investigation and traffic crashes analysis; certain applications of the latter question can be found in more detail in Caliendo and Guida [36], and Caliendo et al. [37,38,39,40].

Without a doubt, many studies have been carried out on the two mentioned subjects; however, given the essential role of tunnels in the functionality of a transportation network, the resilience of these structures should be investigated in greater depth. In this respect, traffic simulation tools should be more applied especially. Some practical uses of the simulation tool can be found in Caliendo and Guida [41], Caliendo and De Guglielmo [42], Caliendo [43], and Astarita et al. [44]; however, these studies do not address the issue of tunnel resilience.

Low flammability of road pavements within tunnels might be another important question of research for limiting the release of toxic gases due to burning asphalt materials in the event of a fire. Pavements that employ materials inert to the combustion process, or in cement concrete (the behaviour of which under traffic can be seen, for example, in Caliendo and Parisi [45]), might be suitable for this scope.

Finally, it appears useful also to mention that it is worth investigating the risk level due to the transit of the Electric Vehicles (EVs) through tunnels for the significant releases of toxic gases in the event of battery fire.

The aforementioned lack of knowledge needs to be filled by future studies in order to make further progress in the field of tunnel safety.