A Numerical Evaluation for Estimating the Consequences on Users and Rescue Teams Due to the Fire of an Electric Bus in a Road Tunnel

Abstract

E-mobility is progressively penetrating the European market with the ever-increasing registration of new battery electric vehicles (BEVs). Although BEVs can significantly contribute to achieving the goal of sustainable road transport, they pose new challenges related to the complexity of managing battery fire events, especially in confined spaces such as road tunnels. In this regard, while fires of BEVs with small-sized batteries (i.e., cars and vans) have been widely studied, the consequences of fires involving battery electric buses (BEBs), which are equipped with larger-capacity traction batteries, have not yet been sufficiently investigated. In this context, 3D computational fluid dynamics (CFD) simulations were performed to quantitatively assess the threat that a BEB might pose to the safety of users and rescue teams when it catches fire in a bi-directional road tunnel. In this respect, a comparison was also carried out with the consequences of the fire of a similar internal combustion engine bus (ICEB). Since the environmental conditions inside a tunnel, apart from its geometry, depend on both the traffic flow and type of ventilation, the safety of the users and rescue teams in the event of a BEB fire was evaluated by considering the tunnel under conditions of congested traffic, as well as natural or longitudinal mechanical ventilation. The results showed that the fire of the BEB, compared to that of its ICEB counterpart, worsened the environmental conditions inside the tunnel, especially in terms of toxic gas concentrations. This caused a reduction in the safety level of the users when considering the scenarios of both the naturally and mechanically ventilated tunnel. Moreover, in the case of natural ventilation, the BEB fire was found to cause a higher reduction in the safety level of the rescue teams.

1. Introduction

The European Green Deal, adopted by the European Commission in 2019, is a package of policy initiatives aimed at reducing greenhouse gas emissions in Europe by at least 50–55% by 2030 compared to 1990 levels, with the final objective of reaching climate neutrality in the European Union (EU) by 2050 [1]. In this context, since the road transport industry has been found to be one of the largest producers of greenhouse gases in Europe, a reduction in CO₂ emissions in this sector is crucial to achieving the targets set by the EU [2]. Due to the well-known environmental benefits of decreasing CO₂ emissions and enhancing air quality on the roads, e-mobility has become a possible solution for reaching sustainable road transport in the future.

E-mobility, which includes Plug-in Hybrid Electric Vehicles (PHEVs) and battery electric vehicles (BEVs), is progressively penetrating the EU market, especially with the ever-increasing new registration of cars and buses. In 2023, the EU electric car market closed with a 13.9% growth compared to the previous year, achieving a full-year volume of 10.5 million units [3]. The registrations of new electric buses in 2023 increased by 19.4% compared to 2022, with a total of 32,593 units [4]. Obviously, the fleet of electric vehicles (EVs) on the roads is destined to grow further in the coming years to meet the current policy of greenhouse gas reduction. Moreover, since the sales growth rate of BEVs has mostly been quicker than that of PHEVs [2], BEVs are expected to become the prevalent type of EVs in the future [5].

Although BEVs can significantly contribute to reaching the goal of sustainable road transport, they pose new challenges in the context of fire safety engineering. These challenges are mainly associated with batteries and the complexity of addressing a fire event involving a BEV. Currently, battery electric vehicles are mostly powered by Lithium-Ion Batteries (LIBs), whose thermal runaway (i.e., the phenomena whereby an LIB goes into a state of rapid and uncontrollable self-heating) has been found to be the main cause of BEV fires [5,6]. During the burning of an LIB, the combustion of the electrolyte results in the release of Hydrogen Fluoride (HF), which is a colorless but highly poisonous gas that may pose a serious threat to human health [7]. Apart from the HF emissions, which are absent or reduced in the event of a fire involving conventional internal combustion engine vehicles (ICEVs), the potential reignition of the battery fire, the possible explosion of one or more battery cells with the projectile risk, and the need for an excessive quantity of suppression agent to cool the burning battery make the extinguishment of BEV fires more difficult to address than their ICEV counterparts [5]. In this respect, even more severe consequences and challenging firefighting operations are expected in the event of a BEV fire in confined spaces such as road tunnels, in which the combustion products (e.g., hot gases, toxic substances, and smoke) can quickly fill up the entire structure, making the environmental conditions within it extremely dangerous for the safety of the users and rescue teams engaged in extinguishing the EV fire and/or assisting the tunnel occupants.

In recent years, many studies have carried out experiments and/or numerical simulations to assess the consequences of fires involving EVs powered by LIBs and/or the efficiency of some firefighting measures to address their extinguishment, while also investigating fires of similar ICEVs for comparative purposes.

Lecocq et al. [8] performed full-scale fire tests in an experimental tunnel facility to compare the fire consequences of analogous Battery Electric Cars (BECs) and internal combustion engine cars (ICECs). They found that the two types of vehicles showed similar values of the maximum Heat Release Rate (HRR_max) as well as CO₂ and CO concentrations, with a significant difference in the production of Hydrogen Fluoride (HF), a higher release of which was observed for BECs due to the combustion of their LIB pack. They also pointed out that the results of these tests might be used in numerical simulations of BEC fires in confined spaces such as road tunnels, underground car parks, or other underground facilities.

Watanabe et al. [9] carried out real-scale fire tests in a fire room to compare the risk associated with the burning of a BEC rather than an equivalent ICEC. They found that the maximum magnitudes of both the HRR and heat flux for the BEC were higher than those of its ICEC counterpart. Moreover, the fire risk posed by the BEC was greater than that corresponding to the similar ICEC.

Blum and Long [10] conducted a series of full-scale fire suppression tests in an open-space environment on a BEC model. They found that BEC fires need much longer active suppression operations to counter LIB reignitions and considerably larger amounts of water than ICEC fires. Moreover, due to the long-expected cooling duration of burning BEVs, firefighter protocols should consider the potential need to equip rescue teams with multiple self-contained breathing apparatus tanks.

Lam et al. [11] performed a series of experiments in a full-scale fire test facility to compare the heat flux and HRR of similar BECs and ICECs subjected to external fire conditions simulating a fuel spill fire. Although the hazard did not change with the propulsion system, the heat flux levels and HRR_max measured during BEC fires were a bit lower than those recorded in the corresponding ICEC fires. However, the authors remarked that the fire response of BEVs might vary with the design and State of Charge (SOC) of the battery, as well as the vehicle model.

Truchot et al. [12] conducted two series of fire tests in an experimental tunnel facility. The first involved the individual combustible components of a vehicle (i.e., tires, plastics, gasoil, and electrical compounds), while the second focused on detailed smoke analysis concerning full-scale tests on similar BECs and ICECs. They observed that the HRR was not modified by the battery, as well as that the toxic gases released by the two types of vehicles were comparable. However, the total amount of HF released by the BEC was found to be 1.8 times (in kg units) higher than that of the corresponding ICEC.

Willstrand et al. [13] performed full-scale fire tests inside a fire hall to investigate the gases and heat released during the burning of analogous battery electric and internal combustion engine full-size vans. They found that the kind of propulsion system did not significantly affect the HRR_max and total heat release. Instead, the production of HF was found to be the main difference between BEV and ICEV fires. Finally, using the full-scale fire test results as input data, they also developed computational fluid dynamics (CFD) models, for the resolution of which the Fire Dynamics Simulator (FDS) code was applied.

Król and Król [14] presented CFD simulations, which were performed with the FDS code, to evaluate the temperature distribution and HF spread during a BEC fire in an underground garage, using as input the HRR and HF emissions data obtained experimentally by Lecocq et al. [8]. They found that even the fire of a small BEV in an enclosed space, such as an underground garage, released an amount of HF that might pose a threat to the health of people in the structure, even in the early phase of the fire.

Dorsz and Lewandowski [15] set up CFD models, for the resolution of which the FDS code was used, to compare the consequences on user safety and property due to a fire involving similar BECs and ICECs in an underground garage, using as input the smoke and heat data computed by Lecocq et al. [8]. They observed that the burning of the two types of vehicles had similar impacts in terms of visibility and temperature. Moreover, the fire of BECs with small-capacity traction batteries was found not to cause—compared to their ICEC counterparts—a greater risk in terms of power and energy released. However, given the lack of available data, they emphasize the importance of conducting research on vehicles with larger-capacity traction batteries.

Sturm et al. [16] carried out full-scale fire tests in a road tunnel to evaluate the heat and toxic gases released during the burning of vehicles (i.e., vans and cars) with different energy storage technologies (i.e., LIB and diesel), while also investigating the efficiency of certain firefighting methods. They observed that the HRR_max of BEVs was higher than that of similar ICEVs. Moreover, the most critical combustion product of a BEV fire was found to be the HF, whose concentration at the breathing height did not exceed the corresponding threshold value for human health. They also stressed that an efficient firefighting system to extinguish a BEV fire consisted of using a fire lance to inject water directly into the battery, while the use of a fire blanket failed once the battery was implicated in the fire.

Hynynen et al. [17] extended the study of Willstrand et al. [13] by performing further full-scale fire tests inside a fire hall to investigate the effectiveness of a sprinkler system in mitigating the fire consequences of a BEC and a similar ICEC. Overall, the results obtained in this work, expressed in terms of HRR, toxic gas emissions, and smoke composition, confirmed those found by Willstrand et al. [13] (i.e., the HF constituted the largest difference between BEV and ICEV fires).

Kang et al. [18] conducted a series of real-scale fire tests in an open-space environment to assess the hazards caused by the fire of a BEC and a similar ICEC. They found that the propulsion system did not significantly change the HRR_max, total heat released, and average effective heat of combustion. However, BEV fires are particularly dangerous for first responders due to the thermal runaway of the LIB packs. This increased risk stems from the delayed detection of such fires and the rapid escalation once the batteries catch fire.

Raza and Li [19] developed CFD models, which were run with the FDS tool, to investigate the impact of the fire of a battery electric bus (BEB) and a similar internal combustion engine bus (ICEB) in a short road tunnel (i.e., 300 m long), considering both cases of natural and mechanical ventilation. Compared to ICEBs, BEB fires were found to produce higher levels of toxicants, which might make the evacuation process more difficult for some vulnerable evacuees. Moreover, in the absence of a mechanical ventilation system, the results showed that some users might be exposed to irritant concentrations above the threshold values for human health during the burning of both types of vehicles.

Wang et al. [6] carried out a review of studies aimed at evaluating the impact of BEVs (i.e., cars and light commercial vehicles) on ventilation design for road tunnels. They observed that (i) the average HRR developed by BEVs and ICEVs in the event of a fire was comparable; (ii) the production of HF was higher when BEVs caught flames; (iii) water was still suitable for addressing the burning of BEVs. Overall, since the risk after the fire of BEVs and their ICEV counterparts was found to be equivalent, the authors concluded that tunnel ventilation systems designed in accordance with current guidelines do not need to be adapted to cope with the increasingly widespread use of BEVs. However, they pointed out that it had not yet been verified whether these conclusions are still valid in the event of a fire involving larger BEVs such as buses.

Gao et al. [20] conducted FDS simulations on fires involving new energy electric vehicles to analyze the effects of different blocking ratios and water mist spray angles on smoke propagation in road tunnels. The results showed that the smoke back-layering length in the upstream direction increased with the blocking ratio and that setting an appropriate spray angle significantly reduced CO concentrations while also improving tunnel visibility, which might improve the safety of escaping people.

Bai et al. [21] developed a numerical model, which was run using the FDS code, to assess the safety of LIBs for electric cars in the event of a fire in a naturally ventilated road tunnel. Simulations were carried out with different HRRs up to 15 MW, and the results in terms of temperature, CO, and CO₂ concentrations were compared with those from fuel vehicle fires. They found that BEVs produce higher temperatures, greater smoke volumes, and higher CO emissions than ICEVs, which might hinder people’s evacuation and firefighter operations.

Hodges et al. [22], by extrapolating existing data from full-scale fire tests performed on relatively small BEVs and ICEVs (e.g., cars), predicted the HRR curve over time of larger BEVs and ICEVs (e.g., large SUVs). They found that the total energy released by BEVs was consistently greater than that of the corresponding ICEVs, especially for larger vehicles. No significant difference in the HRR_max was measured between BEVs and ICEVs of similar size.

Zhao et al. [23] performed full-scale fire tests in an open-space environment to evaluate the suppression ability of compressed air foam, water spray, and fire blankets on BEC fires in the initial phases. They found that the flame was rapidly reduced when compressed air foam was used, but the battery kept releasing white gas smoke, thus requiring the application of water for a long period of time. Water spray contributed to reducing the HRR and temperature, but its low flow rate could neither extinguish fires in the passenger compartment nor stop the battery thermal runaway. Finally, fire blankets effectively extinguished the fire and controlled its propagation, but they had little effect on arresting the thermal runaway of the LIB.

Yao et al. [24], using the FDS code, analyzed the temperatures and CO concentrations inside an electric bus generated by the burning of only its LIB (i.e., the other components of the bus, such as the seats and curtains, are not assumed to be combustible), considering different wind speeds, ventilation areas, and window opening positions. They found that in the event of a BEB fire, the risk for its occupants can be minimized by opening 100% of the windows and opening the rear window first.

The results emerging from the above chronological literature review are evidently conflicting. It is still not completely clear whether BEV fires in confined spaces such as road tunnels can pose a more serious threat to the safety of the users and rescue teams when compared to the fire of similar ICEVs, thus indicating a gap in knowledge.

Moreover, it is worth pointing out that most of the above-mentioned studies have compared the fire consequences of BEVs and ICEVs of relatively small sizes (i.e., cars and vans), while very few authors have investigated larger vehicles such as buses. Since battery electric buses (BEBs) are generally powered by larger-capacity traction batteries, their fire is expected to pose a greater hazard to human health than that due to the burning of BEVs such as cars and vans. This is a further issue that this paper intends to examine. A comparison with the consequences of the fire of a similar internal combustion engine bus (ICEB) also needs to be carried out.

The aforementioned literature review also shows that the firefighting operations to extinguish a burning BEV are generally more challenging for rescue teams than those needed to address the fire of a similar ICEV, as the fire brigade (i) is forced to get closer to the burning vehicle to inject the suppression agent directly on the battery, thus preventing its reignition, and (ii) is exposed for a longer period of time to the high temperatures and toxic gas concentrations resulting from the fire, having to use a greater amount of suppression agent to deal with the BEV in flames. However, there has been limited research on quantitatively evaluating the safety of firefighters during a BEV fire in road tunnels. Even less research has been carried out on battery electric buses, thus representing an additional gap in knowledge.

In light of these considerations, the scope of this paper is to contribute to filling the aforementioned gap of knowledge by quantitatively assessing the potential additional threat that a BEB—compared to a similar ICEB—might pose to the safety of the evacuees and firefighters when it catches fire in a road tunnel, which is also essential to understand if the current firefighter protocols in road tunnels are still valid or need to be adjusted to cope with the increasingly widespread use of BEBs. To achieve these objectives, the environmental situations in a tunnel in the event of a fire of a BEB or its ICEB counterpart were reproduced by setting up a 3D CFD modeling solved with the FDS tool version 6.7.3 [25]. The people evacuation process toward a safe place (i.e., emergency exits and tunnel portals) was simulated using the Evac code [26]. Furthermore, given that the environmental conditions inside road tunnels, apart from their geometry, depend on both the traffic flow and type of ventilation, the safety of the occupants and fire brigade in the event of a BEB fire was evaluated by considering the tunnel under conditions of congested traffic, as well as natural and mechanical ventilation.

The scientific contribution of this study lies in the advancement of knowledge regarding the effects on the evacuees and firefighters due to a fire involving large battery electric vehicles, such as buses, in road tunnels. By providing accurate details on the fire-induced threats to the users of a naturally or mechanically ventilated road tunnel, this study can serve as a reference for tunnel management agencies. Specifically, it offers a better understanding of the BEB fire-related issue from a fire safety engineering perspective, assesses whether current firefighter protocols in road tunnels need adaptation, and supports more informed decision-making regarding the implementation of additional mitigation measures and/or traffic control strategies.

The structure of the paper is as follows: the next paragraph presents a summary of the literature review. This is followed by a description of the geometry, ventilation conditions, material properties, and traffic characteristics of the road tunnel under investigation, as well as the fire source and escape scenario. After this, the proposed 3D CFD modeling and its calibration process are presented, together with the corresponding modeling of the evacuation process. Subsequently, the results obtained for both BEB and ICEB fires are shown and commented on, and appropriate comparisons are made to point out whether an electric bus powered by an LIB may worsen the environmental situations in the investigated tunnel.

2. Summary of the Literature Review

This paragraph provides a short summary of the study methods used by several authors to analyze the consequences associated with BEV fires and the scope of the investigation.

Table 1. Summary of the literature review.

Year	Ref.	Study Methods	Scope
2012	[8]	Full-scale fire tests in an experimental tunnel facility	To compare the fire consequences of equivalent BECs and ICECs
2012	[9]	Full-scale fire tests in a fire room	To compare the risk associated with the fire of a BEC rather than an analogous ICEC
2015	[10]	Full-scale fire suppression tests in an open-space environment	To investigate whether tactical changes related to emergency response procedures are required to address BEC fires
2016	[11]	Full-scale fire tests in a burn hall	To compare the heat flux and HRR of similar BECs and ICECs subjected to external fire conditions
2018	[12]	Full-scale fire tests in an experimental tunnel facility	To compare the fire consequences of a BEC and its ICEC counterpart
2020	[13]	Full-scale fire tests inside a fire hall and numerical simulations in an enclosed garage	To evaluate the gases and heat released by the fire of similar battery electric and internal combustion engine full-size vans
2021	[14]	Numerical simulations in an underground garage	To assess the temperature distribution and the HF spread during a BEC fire
2022	[15]	Numerical simulations in an underground garage	To compare the consequences on user safety and property due to a fire involving a BEC and a similar ICEC
2022	[16]	Full-scale fire tests in a road tunnel	To evaluate the heat and toxic gases released by the fire of equivalent BEVs and ICEVs (i.e., vans and cars), while also investigating the efficiency of certain firefighting methods
2023	[17]	Full-scale fire tests inside a fire hall	To examine the effects on the fire consequences of a BEC and an analogous ICEC because of a sprinkler system
2023	[18]	Full-scale fire tests in an open-space environment	To assess the hazards due to the fire of a BEC and its ICEC counterpart
2023	[19]	Numerical simulations in a short road tunnel	To investigate the fire consequences of a BEB and a similar ICEB
2023	[6]	Review of studies	To evaluate the impact of BEVs (i.e., cars and light commercial vehicles) on ventilation design for road tunnels
2023	[20]	Numerical modeling	To analyze the impact of different blocking ratios and water mist spray angles in the event of electric vehicle fires in road tunnels
2024	[21]	Numerical modeling	To assess the safety of LIBs for electric cars in the event of a fire in a naturally ventilated road tunnel
2024	[22]	Prediction modeling	To predict the HRR curve over time of large BEVs and ICEVs (e.g., large SUVs) by extrapolating existing data from full-scale fire tests carried out on relatively small BEVs and ICEVs (e.g., cars)
2024	[23]	Full-scale fire tests in an open-space environment	To evaluate the suppression ability of compressed air foam, water spray, and fire blankets on BEC fires
2024	[24]	Numerical modeling	To analyze the temperatures and CO concentrations inside an electric bus generated by the fire of only its LIB

illustrates how a small number of studies have conducted numerical analyses to quantitatively evaluate the threat that an electric bus powered by an LIB can represent for the safety of the users and firefighters when it catches fire in a road tunnel. In the present article, this current lack of knowledge is addressed by performing 3D CFD modeling.

3. Materials and Methods

3.1. Tunnel Geometry

A bi-directional road tunnel complying with the minimum safety requirements defined by the European Directive 2004/54/EC [27] is investigated. It is 850 m long, straight, and flat, with an emergency exit situated halfway along its length. The geometric features of the road tunnel under consideration are reported in detail in Figure 1.

Figure 1 also shows (i) the thicknesses of the road pavement, ceiling, and lateral walls; (ii) the position of the burning vehicle, which is a bus, halfway along the tunnel length; (iii) the transverse location of the jet fans that are assumed to be fixed beneath the ceiling when, as an alternative to the case of natural ventilation only, the tunnel is assumed to be equipped with a longitudinal mechanical ventilation system; (iv) the queued vehicles both downstream and upstream of the burning bus.

3.2. Ventilation Scenarios

3.2.1. General Description

The European Directive 2004/54/EC [27] requires that tunnels with a length exceeding 1000 m must be equipped with a mechanical ventilation system. This implies that when a tunnel is characterized by a length ≤ 1000 m, a quantitative risk analysis is to be performed to understand whether the natural ventilation alone can guarantee an adequate safety level when a fire occurs inside the structure. Given that the tunnel under investigation has a length of 850 m, the fire simulations were performed considering two alternative ventilation scenarios to make a comparison: (i) tunnel under natural ventilation only and (ii) tunnel equipped with a longitudinal mechanical ventilation system.

3.2.2. Natural Ventilation

The piston effect caused by the movement of vehicles passing through the tunnel is assumed to generate a natural air flow along it, which is reproduced in the simulations by means of a positive pressure difference (ΔP) set between the two portals of the tunnel (Figure 1). The ΔP is imposed to be equal to only +0.5 Pa (i.e., between Portal A and Portal B) [28] because, being the examined tunnel characterized by bi-directional traffic (i.e., traffic flows transiting in the two opposite traveling directions), the piston effect is expected to be limited.

In this respect, a preliminary analysis showed that the ΔP of +0.5 Pa set between the two tunnel portals produces a natural air flow along the structure having an average velocity of approximately 0.35 m/s.

3.2.3. Longitudinal Mechanical Ventilation

When the tunnel is assumed to be mechanically ventilated, the induced air flow along it is due not only to the piston effect caused by moving vehicles but also to a longitudinal mechanical ventilation system made up of eight pairs of jet fans fixed at the ceiling (for more details, see Figure 1). All the jet fans are considered to come into operation with the activation of the fire alarm system at the time t_alarm = 150 s [29] after the fire start (i.e., t₀ = 0).

A preliminary simulation showed that the combined effect of the natural and longitudinal mechanical ventilation generates an air flow moving from Portal A to Portal B of the tunnel with an average velocity of about 3.2 m/s, which was found to be large enough to avoid the occurrence of the back-layering phenomenon.

3.3. Material Characterization

The road pavement of the tunnel, which includes the two shoulders and the two traffic lanes, consists of an asphalt mixture, while the tunnel structure (i.e., the ceiling and the two lateral walls) is made of cement concrete. The specific heat, apparent density, thermal conductivity, and emissivity coefficient of asphalt mixture [30] and cement concrete [31] are shown in

Table 2. Material characterization: specific heat, apparent density, thermal conductivity, and emissivity coefficient.

Material	Specific Heat [kJ/(kgK)]	Apparent Density [kg/m3]	Thermal Conductivity [W/(mK)]	Emissivity Coefficient [−]
Asphalt mixture	0.88	2275	0.56	0.91
Cement concrete	0.94	2585	1.67	0.90

.

3.4. Hourly Traffic Volume

In this study, the worst-case traffic scenario, in which the tunnel is assumed to be under conditions of congested traffic, was considered (i.e., the hourly traffic volume is supposed to be equal to the capacity of a bi-directional road according to the HCM [32]: 1600 vehicles/h per each travel direction). Moreover, the traffic composition was considered as follows: 75% cars, 23% heavy goods vehicles (HGVs), and 2% buses.

3.5. Fire and Evacuation Scenarios

3.5.1. Type, Location, and Geometry of the Burning Vehicles

Two different fire sources consisting of a battery electric bus (BEB) powered by a Lithium-Ion Battery (LIB) and a conventional internal combustion engine bus (ICEB) are investigated.

Each burning vehicle is assumed to be in the middle of the tunnel length and close to the sidewalk β.

The buses in flames are modeled in the simulations as parallelepipeds of 12 × 2.9 × 2.5 m³ (length × height × width) located at 0.2 m from the road surface. It is worth noting that the BEB and ICEB under consideration are assumed to be analogs to each other (i.e., the same vehicle model from the same manufacturer), which allows for an appropriate comparison between the two propulsion systems.

3.5.2. Heat Release Rate of the Two Buses on Fire Investigated

According to NFPA [33], the maximum Heat Release Rate (HRR_max) representative of an ICEB fire is of the order of 30 MW, which is typically due to the combustion of different materials including bus interior shells and attachment, seats, tires, and fuel [34]. With specific reference to the propulsion system (i.e., gasoline/diesel for ICEBs or battery for BEBs), it is to be stressed that while its contribution to the HRR_max decreases in the event of a fire as the amount of gasoline/diesel in the burning ICEB decreases with the distance traveled, in the case of a BEB fire, its contribution to the HRR_max remains more or less constant since the impact of the battery combustion does not vary significantly with its State of Charge (SOC) [5]. Therefore, by assuming that the bus fire occurs after a long journey, the HRR_max is expected to be higher in the event of a BEB fire compared to that of an ICEB.

By using Equation (1) [5],

$$∆{\mathrm{H}\mathrm{R}\mathrm{R}}_{\mathrm{m}\mathrm{a}\mathrm{x}}=2{\mathrm{E}}_{\mathrm{B}}^{0.6},$$

(1)

where the ∆HRR_max, expressed in MW, is the additional contribution to the HRR_max due to the combustion of the only battery, and E_B is the battery energy capacity in terms of Wh, and by assuming E_B = 300,000 Wh for the bus [35], the ∆HRR_max associated with the combustion of the only LIB of the BEB was estimated to be about 4 MW. Therefore, the HRR_max considered for the BEB fire is 34 MW in contrast to 30 MW of the burning ICEB.

3.5.3. Fire Curves

For both of the aforementioned types of buses, the HRR was assumed to grow according to a linear law until it reaches its maximum value (i.e., 34 MW and 30 MW for BEB and ICEB fire, respectively) after t_max = 9 min since the bus catches fire, and then it remains constant and fully developed until the arrival of the rescue teams to extinguish the fire and/or assist the tunnel occupants, which was considered to happen after t = 15 min from the fire start. In this paper, all the numerical simulations were performed for 15 min (i.e., the instant in which the firefighting operations start).

3.5.4. Hydrogen Fluoride of the Two Burning Buses

According to the literature review, the difference between a fire involving a BEV and an ICEV lies more in the amount of Hydrogen Fluoride (HF) released, which is found to be higher for a burning BEV. However, the results of the experimental tests are currently available only for small vehicles (i.e., cars and vans), while any data regarding the production of HF during the fire of larger vehicles, such as buses, have not yet been reported. Since the amount of HF released by a BEV fire is mainly associated with the combustion of its LIB and is expected to increase as the battery energy capacity increases [36], by linearly extrapolating [13,17,36,37] the HF data (i.e., the amount of HF emitted in the unit time of the experiment, which is hereafter denoted as HF release rate) obtained from fire tests performed on small BEVs such as cars and vans [8,13,16], we found that the HF release rate in the event of a fire involving an electric bus with a battery energy capacity of 300,000 Wh might be about 3 g/s.

Regarding the production of HF during an ICEV fire, for example, Lecocq et al. [8] and Truchot et al. [12] found that the amount of HF released by an internal combustion engine car was about half that measured for a similar BEV.

Therefore, given the lack of experimental data concerning the amount of HF produced by the fire of a BEB, and since there are contrasting results about the amount of HF emitted by an ICEV fire, the HF release rate for the BEB fire was assumed to be equal to the aforementioned value of 3 g/s, while the burning of the ICEB was considered to be characterized by an HF release rate of 1.5 g/s.

3.5.5. Combustion Products

With reference to the heat of combustion and the production of both CO and soot, the above-mentioned literature review showed that they do not change significantly with the propulsion system. On this basis, a heat of combustion of 26.2 MJ/kg [25], a CO yield of 0.1 kg/kg [38], and a soot yield of 0.05 kg/kg [39] were assumed for both types of buses under investigation (i.e., battery electric bus and internal combustion engine bus).

3.5.6. Vehicles in the Queue

When the bus starts to catch fire (i.e., t₀ = 0), as mentioned above, the examined bi-directional road tunnel was considered to be full of vehicles stopped in the queue, the number of which was computed assuming that (i) vehicles entering the tunnel form the queue without passing the burning bus; (ii) all the vehicles in the queue (i.e., cars, HGVs, and buses) are converted into equivalent cars, each of which is modeled in the simulations as a rectangular block of 6 × 1.5 × 1.8 m³ (length × height × width) placed 0.2 m from the asphalt wearing course; (iii) the equivalent car at the front of the queue (i.e., the one nearest the burning bus) on both traffic lanes stops 16 m away from the fire center; and (iv) each vehicle entering the tunnel lines up one after another, keeping a safe distance of 2 m.

Considering the above assumptions, the total number of equivalent cars in the two queues (i.e., upstream and downstream of the burning bus) was estimated to be 102.

3.5.7. Escaping Users

The number of users inside the tunnel when the bus fire occurs depends not only on the number of cars, HGVs, and buses in the queue but also on the number of people present in each type of queued vehicle. In this respect, the occupancy rate of the vehicles in the queue was assumed to be 1.7, 1, and 30 people for cars, HGVs, and buses, respectively, and according to the traffic composition assumed in [28], the total number of users in the equivalent cars stopped in the two queues (i.e., upstream and downstream of the burning bus) was estimated to be 214.

To these 214 users in the equivalent cars in the queue should be added the 30 people occupying the bus in flames, for a total of 244 users potentially at risk in the tunnel when the fire occurs. In this regard, it is to be said that, except in the case of a serious traffic collision, the bus occupants can self-evacuate the tunnel when it starts to burn.

3.6. Research Framework

This paper is set in the field of research concerning New Energy Carriers (NECs) aimed at reducing greenhouse gas emissions, including more especially the battery electric buses (BEBs), whose numbers on road networks are expected to increase significantly in the coming years. Since the technological advantages of these vehicles might have, when compared to internal combustion engine buses (ICEBs), additional negative effects in enclosed spaces such as road tunnels in the event of a fire, this paper extends the state of knowledge by performing 3D CFD simulations to quantitatively assess the safety of the occupants and firefighters in the event of a BEB fire in the bi-directional road tunnel investigated.

Considering the conflicting results of the literature, this study could contribute to a better understanding of the issue under the aspects of fire safety engineering and if the current firefighter protocols need to be adapted.

The methodology followed in this study is briefly illustrated in Figure 2.

4. Computational Fluid Dynamics Modeling

4.1. Overview

CFD codes are suitable tools for modeling the interactions of fluid stream, heat exchange, and combustion inside confined spaces such as tunnels. By using numerical methods to solve the governing equations of fluid dynamics and heat transfer, CFD allows researchers and engineers to gain valuable knowledge about the behavior of fires in tunnels and the safety of the users involved.

CFD simulations allow for the prediction of temperature profiles, toxic gas concentrations, smoke spread, and air flow velocity distributions in the event of a tunnel fire. These predictions are fundamental for understanding the evolution and propagation of fires in confined spaces such as tunnels. Numerous factors contribute to the reliability and accuracy of CFD predictions, including, for example, the fineness of the mesh employed to discretize the computational domain, as well as the reliability of both methods for solving the governing equations and models for simulating thermal radiation, turbulence, and combustion.

Some studies in which CFD tools have been applied to simulate fires in road tunnels are, for example, those carried out by Caliendo et al. [29,40,41]. Nevertheless, it is worth pointing out that our previous research works did not deal with BEV fires in road tunnels.

4.2. Fire Dynamics Simulator Code

The fire scenarios under consideration were simulated using version 6.7.3 of the Fire Dynamics Simulator (FDS) code [25]. The FDS is a CFD tool developed by the collaboration between the VTT Technical Research Centre of Finland and the National Institute of Standards and Technology (NIST). It is widely used for reproducing the dynamics of fire development and smoke diffusion inside confined spaces such as road tunnels. The FDS code uses numerical methods for solving the governing equations of fluid flow, heat transfer, species transport, and combustion in 3D spaces.

Besides the models to describe the thermal radiation, turbulence, and combustion phenomena, the main input data to be defined in the FDS code to reproduce a tunnel fire scenario involve (i) the geometry of the domain of interest (i.e., the entire tunnel in our case); (ii) the dimensions of the 3D cells used to discretize the computational volume; (iii) the position and geometric characteristics of the fire source and vehicles in the queue; (iv) the HRR curve over time and the yields of combustion products of the burning vehicle; (v) the properties and thicknesses of all materials; and (vi) the natural and longitudinal mechanical ventilation as described above.

4.3. Thermal Radiation, Turbulence, and Combustion Models

Among the different physical models contained in the FDS tool for predicting the fundamental processes that control the development of smoke and fire, the modeling approaches used in this paper are those set as defaults in the code [25]. To incorporate the radiative heat transfer into the model, the FDS tool solves the radiation transport equation for a gray gas employing a method similar to that of finite volumes used for convective transport. The turbulence was modeled using the very large eddy simulation approach, while the wall function method was applied for the near-wall regions to account for the turbulent boundary close to solid objects. The mixing-controlled method was used for the combustion modeling.

4.4. FDS Performing

4.4.1. Calibration

The calibration process of the FDS code against fire scenarios involving battery electric vehicles (BEVs), powered by Lithium-Ion Batteries (LIBs), in confined spaces was performed by reproducing the study of Willstrand et al. [13]. These authors investigated a BEV fire in an enclosed garage consisting of 65 parking spaces distributed over an area of 1750 m², with a height of 3 m. The BEV fire, located close to the permanently open garage door, was characterized by an HRR_max of 5 MW and an HF release rate of 0.22 g/s. Using the default thermal radiation, turbulence, and combustion models of the FDS code [25], Figure 3 shows a good agreement between the HF concentrations predicted by the FDS code and those provided by Willstrand et al. [13] in the range of human breathing heights (i.e., 1.2–2.1 m from the walking surface) after a time (t) of 15 min from the fire start. In this respect, an error not exceeding 5% was found, thus proving the ability of the FDS code to simulate BEV fires.

Considering this, the FDS pre-defined models of thermal radiation, turbulence, and combustion were also set in the 3D CFD modeling of the battery electric bus (BEB) fire in the full-scale road tunnel investigated in this paper.

4.4.2. Grid Sensitivity Analysis

The mesh resolution that constitutes an acceptable compromise between the calculation time and the accuracy of the numerical simulation results for the investigation being undertaken was identified by performing a grid sensitivity analysis. In this respect, the optimal grid size for a given problem is achieved when the ratio between the characteristic fire diameter $D^{\ast }$ [m] and the nominal size of a mesh cell ${\delta }_x$ [m] is within the range of 4–16 [25], with $D^{\ast }$ calculated as follows:

$$D^{\ast }={\left({\displaystyle \frac{Q}{{{\rho }_{\infty }c}_p{T}_{\infty }\sqrt{g}}}\right)}^{{\displaystyle \frac{2}{5}}},$$

(2)

where $Q$ is the HRR (i.e., 34,000 kW for the BEB and 30,000 kW for the ICEB), ${\rho }_{\infty }$ is the density of air (i.e., 1.204 kg/m³), $c_p$ is the specific heat of air (i.e., 1.005 J/(kgK)), $T_{\infty }$ is the ambient temperature (i.e., 293 K), and $g$ is the gravity acceleration (i.e., 9.81 m/s²). For the problem under consideration, $D^{\ast }$ was found to be equal to 3.93 and 3.74 for the fire scenario involving the BEB and ICEB, respectively. This means that the optimal cell size should range from 0.25 m to 0.98 for the BEB fire and from 0.23 m to 0.93 m for the burning ICEB. It is also worth noting how the FDS code only allows for the use of parallelepiped cells, better if cubic.

In light of this, by assuming cubic elements with sides of 0.25, 0.4, 0.5, or 0.8 m, Figure 4 shows the predictions of temperatures and HF concentrations, as a function of the cell size, at points A, B, and C after a time (t) of 15 min from the start of the BEB fire (i.e., HRR_max of 34 MW and HF release rate of 3 g/s), considering the scenario of longitudinal mechanical ventilation. The points A, B, and C have the same distance z of 2 m (i.e., about the breathing height) from the road surface, while their distance x from the wall γ is 0.6 m (i.e., along the sidewalk α centerline), 5 m (i.e., along the tunnel centerline), and 9.4 m (i.e., along the sidewalk β centerline), respectively. The positions of these three points, all of which refer to the cross-section located at 10 m downstream of the fire center, were selected to highlight the effects of grid size in a sufficiently large domain on the predicted temperatures and HF concentrations. It is to be stressed that given the transverse location of the pairs of jet fans (i.e., centrally beneath the tunnel ceiling, see Figure 1), the velocity of the air flow generated by the longitudinal mechanical ventilation system was found to be higher along the tunnel centerline (i.e., point B) than in the proximity of the tunnel walls (i.e., points A and C). Since the effects of longitudinal ventilation to push the combustion products (e.g., hot gases, toxic substances, and smoke) toward Portal B increase as its velocity increases, the peak values of the temperatures and HF concentrations along the tunnel centerline are expected to be reached farther downstream of the fire than near the tunnel walls. This explains why the values of the temperatures and HF concentrations detected at point B are lower than at points A and C.

Figure 4 shows that using cubic cells with side sizes smaller than 0.4 m did not lead to substantial differences in the predictions of either temperatures or HF levels. In fact, the percentage differences between the results obtained using cubic elements having sides of 0.4 m rather than 0.25 m were found to be less than 5%.

It is worth pointing out that a grid sensitivity analysis was also performed for the scenario involving the BEB fire in the naturally ventilated tunnel. However, since the results of this grid sensitivity analysis led to the same conclusions commented on for the mechanically ventilated tunnel, they have not been reported here to save space.

On this basis, the entire tunnel volume was discretized into 975,375 cubic cells of 0.4 m side in all the simulations performed in this study (i.e., those for reproducing the fire scenario involving the BEB or ICEB, in both cases of ventilation investigated).

5. Escape Process Modeling

5.1. Evac Code

The people evacuation process from the tunnel during the bus fire was simulated using the Evac tool [26], which is the egress module of the FDS code. Being an agent-based simulator, Evac allows each user to be treated as a separate entity, or an agent, characterized by specific properties (e.g., pre-movement time and walking speed) and evacuation strategies (e.g., selection of the escape route and exit). In this code, the algorithm governing the exit movement of agents solves, for each of them, an equation of motion in a continuous 2D space and time.

The Evac simulator employs the Fractional Effective Dose due to toxic gases (FED_{toxic gases}) parameter to assess the number of potential victims in the event of a fire. In this respect, the concentrations of toxic gases (e.g., HF, CO, and CO₂) are computed by the FDS tool and then automatically implemented in the Evac code, which also accounts for the effects of these substances on the reduction in walking speed. The Evac tool calculates the FED_{toxic gases} for each agent during their entire evacuation process as follows:

$$\mathrm{F}\mathrm{E}{\mathrm{D}}_{\mathrm{t}\mathrm{o}\mathrm{x}\mathrm{i}\mathrm{c} \mathrm{g}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{s}}=\mathrm{F}\mathrm{E}{\mathrm{D}}_{\mathrm{O}2}+\left(\mathrm{F}\mathrm{E}{\mathrm{D}}_{\mathrm{C}\mathrm{O}}+\mathrm{F}\mathrm{E}{\mathrm{D}}_{\mathrm{N}\mathrm{O}\mathrm{x}}+\mathrm{F}\mathrm{E}{\mathrm{D}}_{\mathrm{C}\mathrm{N}}+{\mathrm{F}\mathrm{L}\mathrm{D}}_{\mathrm{i}\mathrm{r}\mathrm{r}}\right)\times {\mathrm{H}\mathrm{V}}_{\mathrm{C}\mathrm{O}2},$$

(3)

in which ${\mathrm{F}\mathrm{L}\mathrm{D}}_{\mathrm{i}\mathrm{r}\mathrm{r}}$ is the fraction lethal dose of irritants (e.g., HF), ${\mathrm{H}\mathrm{V}}_{\mathrm{C}\mathrm{O}2}$ is the hyperventilation factor (i.e., condition in which a person tends to breathe quicker and deeper than normal) due to CO₂, while $\mathrm{F}\mathrm{E}{\mathrm{D}}_{\mathrm{O}2}$, $\mathrm{F}\mathrm{E}{\mathrm{D}}_{\mathrm{C}\mathrm{O}}$, $\mathrm{F}\mathrm{E}{\mathrm{D}}_{\mathrm{N}\mathrm{O}\mathrm{x}}$, and $\mathrm{F}\mathrm{E}{\mathrm{D}}_{\mathrm{C}\mathrm{N}}$ are the fraction of an incapacitating dose of low O₂ hypoxia (i.e., oxygen deficiency at the level of tissues), CO, NO_x, and CN, respectively.

In this context, a user is considered incapable of self-evacuating the tunnel when the corresponding FED_{toxic gases} exceeds 0.1. This threshold value was chosen to include even the most vulnerable or sensitive categories of people present in the tunnel when the fire occurs [42].

The main input data to be implemented in the Evac code to simulate the people’s escape process from the tunnel when the bus catches fire concern both the properties (i.e., the unimpeded walking speed and the pre-movement time) and the evacuation strategies (i.e., the initial position along the tunnel when the fire starts, the exit, and the escape direction) of each user or group of them.

5.2. Users’ Properties

The effective walking speed of each evacuee is automatically computed by the Evac code as a function of the corresponding unimpeded walking speed (i.e., the walking speed with which a user would move if there were no obstacles, such as queueing vehicles, high concentrations of toxic gases, and reduced visibility distances due to smoke, along their escape path), which was set at 0.7 m/s.

Once the fire occurs, people start to self-evacuate the tunnel after a certain period defined as pre-movement time, which is the sum of the detection and reaction time. It was assumed that each user in the queued equivalent cars detects the dangerous situation thanks to the activation of the fire alarm system at t_alarm = 150 s from when the fire starts and, consequently, reacts by leaving their own vehicle within the next 30 s. Hence, the pre-movement time of people in the queued equivalent cars was set at 180 s, while it was increased by 60 s (i.e., 240 s in total) for the occupants of the burning bus to consider the time needed for all of them to leave the vehicle when it catches fire [43].

Furthermore, given that the average frequency per unit time that vehicles enter the tunnel is 2.25 s, the structure is already full of vehicles in the queue when the fire alarm system is activated (t_alarm = 150 s from when the fire occurs). This means that it was not necessary to assign an extra pre-movement time to any user to account for their arrival time in the tunnel.

5.3. Evacuation Strategies

As far as the evacuation strategies are concerned, all the evacuees are considered to be in the vicinity of their own vehicle when the fire starts. The occupants of the equivalent cars in the queue are assumed to leave the structure using the nearest sidewalk toward the closest tunnel portal: (i) people upstream of the bus in flames self-evacuate the structure using the sidewalk β in the direction of Portal A; (ii) the users downstream of the fire leave the tunnel through Portal B moving along the sidewalk α (Figure 5). The occupants of the burning bus are considered to exit the structure using the nearest sidewalk β toward Portal A because they would find more sustainable environmental conditions upstream rather than downstream of the fire, with the combustion products (e.g., hot gases, toxic substances, and smoke) being pushed by the longitudinal ventilation in the direction of Portal B. Moreover, it is to be said that the emergency exit was assumed to be inaccessible to the users due to its close proximity to the bus in flames located halfway along the tunnel length.

6. Analysis and Discussion of the Results

6.1. Environmental Situations in the Tunnel to Which the Users Downstream of the Burning Bus Are Exposed (after t_max = 9 min from the Fire Start)

Given that the fire-induced threats (e.g., hot gases, toxic substances, and smoke) generated by the burning vehicle are pushed by the longitudinal ventilation, especially when it is mechanical, in the direction of Portal B for the case studied, the environmental conditions in the bi-directional road tunnel under investigation are foreseen to be more severe for evacuees downstream of the burning bus. On this basis, to evaluate if the occupants of the examined road tunnel can safely self-evacuate when the bus catches fire, the spatial distributions downstream of the fire concerning HF, CO, and CO₂ levels, as well as the values of the temperature, radiant heat flux, visibility distance, and FED_{toxic gases}, at the height of breathing along the escape route toward Portal B are reported for both BEB (continuous lines) and ICEB (dashed lines) fires, under conditions of natural (red lines) or mechanical (green lines) ventilation.

The position at t_max = 9 min of the last escaping user (i.e., the one closest to the bus in flames) leaving the tunnel through Portal B is also reported in the following figures.

Table 3. Distance downstream of the fire center of the last user escaping toward Portal B after t_max = 9 min from the fire start computed with reference to the burning of both the BEB and the ICEB, in both cases of ventilation investigated. The average effective walking speed of this user is reported in brackets.

Burning Vehicle	HRRmax [MW]	HF [g/s]	Natural Ventilation	Mechanical Ventilation
			Distance Downstream of the Fire Center of the Last User Escaping toward Portal B after tmax = 9 min from the Fire Start (Their Average Effective Walking Speed):
BEB	34	3	150 m (0.417 m/s)	63 m (0.175 m/s)
ICEB	30	1.5	157 m (0.436 m/s)	69 m (0.192 m/s)

shows the distance of the mentioned user from the fire center, as well as their average effective walking speed, during the burning of both the BEB and the ICEB. It is to be said that the effective walking speed of this user, and consequently their distance from the fire center, decreases as the toxic gas concentrations to which they are exposed while evacuating from the tunnel increase.

6.1.1. HF Amount

Figure 6 shows that the burning of the BEB causes significantly higher HF concentrations downstream of the bus in flames than the ICEB fire, in both cases of ventilation investigated. It is to be noted that although the input data on HF were given to the FDS code in g/s, the software automatically provides as output the temporal and spatial profiles of the HF levels in terms of ppm, which is a common unit of measurement used to express the concentration of a given substance in a mixture and to make a comparison with the tenability limits for human health.

By considering the scenario of the naturally ventilated tunnel, from Figure 6, it can be seen that during the BEB fire, assumed to be characterized by an HRR_max of 34 MW and an HF release rate of 3 g/s, the spatial profile (after t_max = 9 min) of the HF level starts to be above the tenability limit of 30 ppm [7] at a distance of about 80 m from the fire center; then, it remains above the mentioned limit up to around 400 m from the fire center, and it decreases at Portal B where there are conditions of fresh air, with a peak value of approximately 60 ppm measured at about 150 m from the electric bus in flames. In contrast, in the case of the burning ICEB, assumed to be characterized by an HRR_max of 30 MW and an HF release rate of 1.5 g/s, the spatial profile of the HF concentration is always below 30 ppm, reaching a peak value of about 28 ppm at approximately 205 m downstream of the fire center. With reference to the last user escaping toward Portal B, it is also possible to note that, under the condition of natural ventilation, they are exposed to HF levels above the threshold value only in the case of the BEB fire. This means that the burning BEB may cause a reduction in the safety level of tunnel users.

Regarding the case of longitudinal mechanical ventilation, Figure 6 shows that with reference to the BEB fire, the spatial profile of the HF level exceeds 30 ppm only in the proximity of the burning bus, with a peak value of almost 80 ppm measured at a distance of about 20 m away from the fire center; then, it decreases, achieving a value below 30 ppm at a distance of about 30 m downstream of the fire center; thus, it remains below the mentioned limit downstream of the fire up to Portal B. This shows the efficiency of the longitudinal mechanical ventilation, which, by pushing toxic substances out of the tunnel through Portal B, also reduces the HF level inside the structure. In the event of the ICEB fire, the spatial profile of the HF level is always below both that of the burning BEB and the value of 30 ppm, with a peak value of only 34 ppm at about 20 m from the fire center. From Figure 6, it can also be noted that the last user exiting the tunnel through Portal B is exposed to HF levels below the tenability limit during the fire of both types of buses.

6.1.2. CO Level

Figure 7 shows that the BEB fire was found to cause higher CO levels downstream of the burning bus compared to the ICEB, in both cases of ventilation investigated.

With reference to the natural ventilation, Figure 7 shows that during the BEB fire, the CO levels slightly exceed the threshold value of 1200 ppm [44] at a distance between 150 m and 175 m downstream of the fire center, achieving a peak value of approximately 1240 ppm at about 170 m away from the fire center. The CO concentrations produced by the burning ICEB never reach 1200 ppm, with a maximum value of about 1050 ppm detected at approximately 200 m downstream of the fire center. From Figure 7, it is also possible to observe that the last user exiting the tunnel through Portal B may encounter CO concentrations above the tenability limit only during the burning of the BEB, which might contribute to the worsening of their health conditions.

Regarding the case of the mechanically ventilated tunnel, Figure 7 shows how the fire involving both BEB and ICEB leads to CO concentrations above 1200 ppm only in the vicinity of the burning vehicle, with a peak value, measured at about 20 m away from the fire center, that was found to be much higher for the combustion of the BEB rather than the ICEB (i.e., approximatively 2615 ppm against about 1990 ppm). Figure 7 also shows how the last user evacuating toward Portal B is affected by CO levels below the acceptability limit during the fire of both the BEB and the ICEB.

6.1.3. CO Quantity

Figure 8 shows that the burning of the BEB causes higher CO₂ concentrations downstream of the bus in flames than the ICEB fire, in both cases of ventilation investigated. However, for both ventilation conditions examined, the CO₂ levels are always below the tenability limit of 40,000 ppm [44].

6.1.4. Temperature Value

Figure 9 shows that when the tunnel is naturally ventilated, the temperatures may be close to the acceptability limit of 60 °C [45], but they do not exceed the mentioned limit during the burning of either the BEB or the ICEB. In contrast, for the mechanically ventilated tunnel, the values of the temperatures are more than 60 °C at distances from the fire center ranging from 5 to 250 m for both buses in flames, with a maximum value—which is reached at about 12.5 m away from the center of the fire—that was found to be higher for the combustion of the BEB rather than the ICEB (i.e., approximatively 330 °C versus about 275 °C). Figure 9 also shows how the last person escaping toward Portal B may be exposed to temperatures exceeding the tenability limit only when the tunnel is mechanically ventilated for both buses in flames. This confirms that, in a bi-directional road tunnel, mechanical ventilation, by spreading hot gases downstream of the fire, may expose users to temperatures higher than the tenable limit for human health.

6.1.5. Radiant Heat Flux Level

Figure 10 shows that, under natural ventilation, radiant heat fluxes are always below the threshold value of 2 kW/m² [45] for either the BEB or ICEB fire. However, this limit may be exceeded when the tunnel is mechanically ventilated, but only in the proximity of the burning BEB or ICEB (i.e., approximatively 12.5 m away from the fire center, with a peak of 4.5 kW/m² and 2.8 kW/m² for the BEB and the ICEB fire, respectively). From Figure 10, it is also possible to note that the last user escaping from the tunnel through Portal B is never subjected to radiant heat fluxes exceeding the mentioned threshold value.

6.1.6. Visibility Distance Measurement

From Figure 11, it is possible to observe that the BEB fire, in comparison to the ICEB fire, slightly worsens the environmental situations in terms of visibility conditions, in both cases of ventilation investigated. However, no significant differences were found among the scenarios examined because, in all of them, the visibility distance is <10 m (i.e., the threshold value [45]), except near Portal B. Consequently, as Figure 10 shows, the last user exiting the tunnel through Portal B never has acceptable visibility conditions.

6.1.7. FED_{toxic gases} Value

The above-mentioned results showed that the BEB fire worsens—compared to the burning ICEB—the environmental conditions along the escape route used by the tunnel users who are downstream of the bus in flames, especially in terms of concentrations of toxic gases (i.e., HF, CO, and CO₂) and, to a lesser degree, in terms of heat (i.e., temperatures and radiant heat fluxes) and visibility distances. Therefore, in this section, the FED_{toxic gases} parameter is also computed and used for a comparison with the tenability limit of 0.1 [42].

In both cases of ventilation investigated, from Figure 12, we can see how the BEB fire leads to much higher values of the FED_{toxic gases} parameter downstream of the bus in flames than the burning ICEB.

From Figure 12, considering the scenario of the naturally ventilated tunnel, it can also be seen that during the burning of the BEB, the values of the FED_{toxic gases} parameter exceed 0.1 at a distance from the fire center between 140 and 235 m, reaching a peak of approximately 0.15 at about 185 m downstream of the center of the bus in flames. In the event of the ICEB fire, the values of the FED_{toxic gases} parameter are above the tenability limit at a distance from the fire center in the range of 160–215 m, with a peak of about 0.12 detected at approximately 195 m away from the center of the burning bus. Nevertheless, the FED_{toxic gases} corresponding to the last user escaping from the tunnel through Portal B is >0.1 only in the case of the BEB fire.

With reference to the mechanical ventilation, Figure 12 shows that during the BEB fire, the values of the FED_{toxic gases} parameter are higher than the tenability limit both in the proximity of the bus in flames and at a distance from its center between 60 m and 145 m, with a peak of approximately 0.51 measured at about 20 m away from the fire center. In contrast, for the burning ICEB, the FED_{toxic gases} parameter exceeds 0.1 both in the vicinity of the burning bus and at a distance from its center between 75 m and 135 m, achieving a peak of about 0.37 at approximately 20 m downstream of the center of the bus in flames. Therefore, also in this case, the last person escaping from the tunnel through Portal B is subjected to FED_{toxic gases} > 0.1 only during the BEB fire.

In light of this, it can be concluded that in both cases of natural and longitudinal mechanical ventilation, the combustion of a BEB in a bi-directional road tunnel might cause—compared to an ICEB fire—a greater reduction in the safety level of users.

6.2. Environmental Situations in the Tunnel to Which the Firefighters are Exposed (after t = 15 min from the Fire Start)

In this section, to assess if the firefighters are exposed to sustainable environmental conditions when they enter the tunnel to extinguish the burning bus and/or help the tunnel users, the spatial distributions related to the HF, CO, and CO₂ levels, temperature, radiant heat flux, visibility distance, and FED_{toxic gases} at the breathing height along the escape route both upstream and downstream of the bus in flames are reported. It is to be stressed that firefighters are expected to enter the tunnel from Portal A (i.e., in the same direction as the air flow generated by natural or mechanical ventilation).

6.2.1. HF Concentration

Figure 13 shows that the burning BEB causes a substantial increase in the HF levels along the tunnel length compared to the ICEB fire, in both cases of ventilation investigated.

By considering the scenario of the naturally ventilated tunnel, Figure 13 also shows that when the BEB catches fire, the HF concentrations are, except in the proximity of the two tunnel portals, always above the threshold value of 30 ppm [7], achieving a maximum value of approximately 85 ppm and 80 ppm downstream and upstream of the fire, respectively. During the burning of the ICEB, the HF concentrations are above 30 ppm at a distance from the fire center in the range of 110–295 m downstream and 100–290 m upstream of the bus in flames, with a peak value of approximately 40 ppm both downstream and upstream of the fire. These results indicate that the natural ventilation alone is not enough to prevent the back-layering phenomenon.

With reference to the mechanical ventilation, from Figure 13, it can be noted that for both types of bus fires, the HF levels are above the tenability limit only immediately downstream of the fire, while they are almost equal to zero upstream of the bus in flames. These results prove that longitudinal mechanical ventilation can prevent the back-layering phenomenon, pushing toxic gases out of the tunnel through Portal B. It is also worth saying that, in this case, the maximum HF concentration measured for the burning BEB is much higher than that corresponding to the ICEB fire (i.e., approximately 80 ppm versus about 40 ppm).

6.2.2. CO Concentration

From Figure 14, it is possible to note that the CO levels corresponding to the BEB fire are always higher than those associated with the burning ICEB, in both cases of ventilation investigated.

By considering the scenario of natural ventilation, Figure 14 also shows that when the BEB catches fire, the CO concentrations are above the threshold value of 1200 ppm [44] at a distance from the fire center in the range of 40–365 m downstream and 30–360 m upstream of the bus in flames, with a peak value of about 2410 ppm and 2190 ppm downstream and upstream of the burning bus, respectively. In contrast, in the event of the ICEB fire, the CO concentrations are above 1200 ppm at a distance from the fire center in the range of 70–330 m downstream and 70–320 m upstream of the bus in flames, achieving a peak value of about 2015 ppm and 2000 ppm downstream and upstream of the burning bus, respectively. Therefore, these results confirm the previous ones concerning the inability of natural ventilation alone to prevent the back-layering phenomenon.

From Figure 14, it can also be seen that when the BEB or ICEB catches fire, the longitudinal mechanical ventilation efficiently pushes the CO concentrations toward Portal B, so that the CO levels are very low (almost zero) upstream of the bus in flames and higher downstream of the fire, reaching values exceeding the acceptability limit only immediately downstream of the burning bus. The peak value of the CO level detected during the BEB fire is much higher than that measured for the burning ICEB (i.e., about 2790 ppm against approximately 2390 ppm).

6.2.3. CO₂ Concentration

Figure 15 shows that the combustion of the BEB—compared to the burning ICEB—leads to an increase in the CO₂ concentrations both downstream and upstream of the bus in flames, accounting for both cases of the naturally and mechanically ventilated tunnel. However, this increase in CO₂ concentrations is never such as to raise the CO₂ levels above the acceptability limit of 40,000 ppm [44] neither downstream nor upstream of the burning bus.

6.2.4. Temperature

With reference to the natural ventilation, Figure 16 shows that the BEB fire—compared to the burning ICEB—causes a modest increase in temperatures along the tunnel length, which are above the threshold value of 60 °C [45] at a distance from the fire center in the range of 75–200 m downstream and 70–190 m upstream of the bus in flames by considering both types of buses. In the case of longitudinal mechanical ventilation, it can be seen that the temperatures upstream of the fire are always of the order of 20 °C (i.e., the ambient temperature), while, along the tunnel portion downstream of the bus in flames, the combustion of the BEB—compared to the ICEB fire—causes a more significant increase in temperatures, which are >60 °C up to a distance of approximately 330 m and 295 m from the center of the BEB and ICEB in flames, respectively.

6.2.5. Radiant Heat Flux

From Figure 17, it is possible to observe that the combustion of the BEB, compared to that of the ICEB, does not result in a substantial increase in the radiant heat fluxes either downstream or upstream of the burning bus, in both cases of ventilation investigated. The radiant heat fluxes remain always below the tenability limit of 2 kW/m² [45], except near the fire center.

6.2.6. Visibility Distance

Figure 18 shows that, in both cases of ventilation investigated and for the entire tunnel length, the firefighters might be exposed to slightly worse visibility conditions during the burning of the BEB rather than the ICEB. Moreover, when the tunnel is naturally ventilated, the visibility distances along its entire length are always much less than the acceptability limit of 10 m [45], except near the two tunnel portals. Meanwhile, with reference to the mechanically ventilated tunnel, and for both types of bus fires, the visibility distances are always above 10 m along only the tunnel portion upstream of the bus in flames and of the order of about 1.5 m along the entire length of the tunnel portion downstream of the fire.

6.2.7. FED_{toxic gases}

From Figure 19 in both cases of ventilation investigated, the burning of the BEB—compared to the ICEB fire—causes a significant increase in the values of the FED_{toxic gases} parameter along the entire tunnel length.

Figure 19 also shows that in the case of natural ventilation, the values of the FED_{toxic gases} parameter along its entire length are, except near the two tunnel portals, always above the tenability limit of 0.1 [42], reaching a much higher peak during the burning of the BEB rather than the ICEB both downstream (approximatively 0.72 versus about 0.6) and upstream (approximatively 0.75 against about 0.63) of the fire.

In the case of mechanical ventilation, Figure 19 shows that the values of the FED_{toxic gases} parameter upstream of the bus in flames are almost always equal to zero, while they never drop below 0.1 along the entire length of the tunnel portion downstream of the fire, achieving a much higher peak during the combustion of the BEB rather than the ICEB (approximatively 1.75 against about 1.2).

In summary, the findings related to the naturally ventilated tunnel indicate that the firefighters entering the structure according to the direction of the air flow (i.e., from Portal A) to extinguish the burning bus and/or assist people still inside the tunnel might find more unsustainable environmental conditions in terms of HF and CO concentrations, temperatures, visibility distances, and FED_{toxic gases} in the event of the BEB fire compared to the ICEB fire. In this circumstance, the firefighters should wear even more appropriate smoke and firefighting equipment because the BEV fire, in addition to worsening the environmental conditions in the tunnel, also requires much more time to be extinguished than the ICEV fire [10].

7. Considerations and Discussions

The results previously considered and discussed have shown that the burning of a battery electric bus (BEB) in a bi-directional road tunnel might lead to a greater reduction in the safety level of both users and firefighters than the combustion of a similar internal combustion engine bus (ICEB). In contrast, certain studies in the current literature indicate that the fire of a Battery Electric Car (BEC) does not pose any additional risks to users compared to the burning of an analogous internal combustion engine car (ICEC) [16,46]. We also simulated the fire of a BEC in a road tunnel, finding results similar to those of Sturm et al. [16,46], which are not reported in this paper to save space. However, it is to be stressed that in this study, we investigated the consequences of the fire of a BEB powered by a Lithium-Ion Battery (LIB) with an energy capacity much higher than that of a BEC (i.e., 300,000 Wh against 80,000 Wh), thus potentially able to release a greater amount of Hydrogen Fluoride (HF) in the event of a fire than that of a BEC (i.e., 3 g/s against 1 g/s). Given this, when comparing the burning of a BEB with that of a similar ICEB, we might expect a higher risk in the former case than in the latter, as in our case.

The simulation results also showed that the longitudinal mechanical ventilation system, by spreading hot gases and toxic substances toward Portal B, may worsen the environmental situations to which the tunnel occupants downstream of the fire are exposed. In this context, to realize a transverse or semi-transverse ventilation system rather than a longitudinal one might serve as a potential solution.

8. Summary and Conclusions

The main reason justifying this research lies in the need to quantitatively evaluate the additional risks that a battery electric bus (BEB) might pose—compared to a similar internal combustion engine bus (ICEB)—for the safety of the evacuees and firefighters in the event of a tunnel fire. The investigated bi-directional tunnel is 850 m long and is assumed to be under congested traffic conditions, as well as naturally or mechanically ventilated. The two aforementioned types of buses (BEB or ICEB), which were considered to be located in the middle of the tunnel length when they caught fire, were assumed to have the same geometric characteristics, heat of combustion, CO, and soot yields and differed from each other for both the HRR_max and HF release rate: (i) BEB with an HRR_max of 34 MW and an HF release rate of 3 g/s; (ii) ICEB with an HRR_max of 30 MW and an HF release rate of 1.5 g/s.

To achieve the objectives of this study, a 3D computational fluid dynamics (CFD) model was performed, and the FDS tool was used to solve this model. The Evac tool was applied to simulate the consequences on the escape process of the users toward a safe place. The FDS code was preliminary calibrated against a fire involving a battery electric vehicle (BEV) in a confined space, and then the optimal mesh resolution to be used in the numerical modeling was defined by performing a grid sensitivity analysis.

In both cases of ventilation investigated, the CFD outcomes revealed that the burning of the BEB, in comparison to that of its ICEB counterpart, significantly worsened the environmental situations for the tunnel occupants while evacuating downstream of the fire, especially in terms of toxic gases concentrations (i.e., HF, CO, and CO₂), which also led to a significant increase in the values of the FED_{toxic gases} parameter.

Regarding the safety of the firefighters entering the tunnel to extinguish the burning bus and/or help users still in the structure, the simulation results related to the naturally ventilated tunnel showed that they might find highly unsustainable environmental conditions upstream of the bus in flames since the HF and CO levels, temperatures, visibility distances, and FED_{toxic gases} were found to exceed the corresponding acceptability limit, especially during the BEB fire.

Therefore, the results provided by this study showed that the fire of a BEB in a road tunnel, compared to that of an ICEB, may represent an additional threat to the safety of the evacuees and firefighters. In this respect, although tunnel fires happen less frequently than tunnel traffic accidents, their impact may be more negatively relevant to user safety since a large portion of the tunnel can quickly become engulfed in hot gases, toxic substances, and smoke. Considering this, the evaluation of user safety in the event of a BEB fire in road tunnels is a crucial issue in real life, and it requires particular attention from the authorities at different levels.

Although this paper adds knowledge to the field of research concerning the safety of road tunnels in the event of a fire involving a battery electric bus, there are still some points that deserve further investigation. Particularly, experimental tests should be conducted for a more consolidated verification of the results obtained. Moreover, the influence of factors such as stress and panic on users’ behavior during their evacuation process toward a safe place is another aspect to be addressed with future studies. Additional investigations should also be made to define a database of traffic accidents involving different categories of EVs (e.g., cars, vans, buses, heavy goods vehicles, dangerous goods vehicles, etc.) with the aim of building a reliable event tree to be used in quantitative risk analysis. The results might be expressed in terms of F/N curves (i.e., the cumulative frequency F of incidents involving electric vehicles as a function of the potential number N of fatalities). In this respect, certain studies on the F/N curves related to internal combustion engine vehicles can be found, for example, in Caliendo and De Guglielmo [47]. Further studies should also be made to assess the influence of tunnel geometry (e.g., length, longitudinal slope, number of lanes, etc.) on the safety of the occupants and rescue teams in the event of a BEB fire. Finally, the evaluation of the effects on the safety of the evacuees and firefighters due to the use of certain fire suppression technologies (e.g., water mist systems, advanced foams, and fire blankets) to mitigate BEB fire hazards could be a further extension of the present paper. Consequently, additional studies are needed to address these relevant research issues.