Simulation of Possible Fire and Explosion Hazards of Clean Fuel Vehicles in Garages

: Clean fuel is advocated to be used for sustainability. The number of liqueﬁed petroleum gas (LPG) and hydrogen vehicles is increasing globally. Explosion hazard is a threat. On the other hand, the use of hydrogen is under consideration in Hong Kong. Explosion hazards of these clean fuel (LPG and hydrogen) vehicles were studied and are compared in this paper. The computational ﬂuid dynamics (CFD) software Flame Acceleration Simulator (FLACS) was used. A car garage with a rolling shutter as its entrance was selected for study. Dispersion of LPG from the leakage source with ignition at a higher position was studied. The same garage was used with a typical hydrogen vehicle leaking 3.4 pounds (1.5 kg) of hydrogen in 100 s, the mass ﬂow rate being equal to 0.015 kgs − 1 . The hydrogen vehicle used in the simulation has two hydrogen tanks with a combined capacity of 5 kg. The entire tank would be completely vented out in about 333 s. Two scenarios of CFD simulation were carried out. In the ﬁrst scenario, the rolling shutter was completely closed and the leaked LPG or hydrogen was ignited at 300 s after leakage. The second scenario was conducted with a gap height of 0.3 m under the rolling shutter. Predicted results of explosion pressure and temperature show that appropriate active ﬁre engineering systems are required when servicing these clean fuel vehicles in garages. An appropriate vent in an enclosed space such as the garage is important in reducing explosion hazards.


Introduction
Clean fuel [1][2][3] is expected to be widely used for vehicles in advocating for sustainability. Liquefied petroleum gas (LPG) has been used as fuel for vehicles on a limited scale as early as 1912 [1,4]. The desire to reduce the dependence on crude oil boosted the popularity of LPG fueled vehicles in 1970s and 1980s. With growing environmental concerns, the demand for alternative fuels is also increasing.
Although LPG is suitable for different types of vehicles, e.g., vans, truck and buses, this fuel is mainly used in passenger cars including taxis in many dense urban Asia-Oceania areas, such as Hong Kong. In order to reduce emissions from vehicles, the LPG Vehicle Scheme was introduced in 1999 [4,5].
• An incentive scheme in 2000 for replacing diesel taxis with LPG taxis. • An incentive scheme in 2002 for the replacement of diesel light buses with LPG or electric light buses.
The LPG taxi incentive scheme and the diesel light buses replacement scheme were completed at the end of 2003 and 2005, respectively. About 99.9% taxis and over 3100 light buses were switched to LPG through these schemes.
In April 2017, there were 18,152 LPG taxis and 4031 LPG mini buses registered in Hong Kong [6]. With the government's strong incentive for phasing out diesel commercial vehicles, the number of LPG taxis and mini buses is expected to surge continuously.
On 30 April 2017, a compressed natural gas (CNG) fuelled taxi caught fire and exploded in Buona Vista, Singapore [7]. Four people, including an SCDF officer, were injured. The incident once again alerted the public to the fire and explosion hazards of clean fuel vehicles.
The LPG filling station in Hong Kong is classified as a "notifiable gas installation" under the Gas Safety Ordinance Cap. 51 [8]. As part of the approval requirement, a quantitative risk assessment is required to demonstrate that the risk levels associated with stations are acceptable. Moreover, all operators of the stations shall receive proper training on filling operation and emergency response. Specially designed safety features are also required for LPG stations.
The safety of filling facilities has drawn a wide range of public concerns and detailed construction requirements have been formulated [9]. Yet, few people were aware that LPG vehicle maintenance may possibly be ticking time bombs until a deadly explosion occurred at a garage in Hong Kong in April 2015 [10] (Figure 1). In April 2017, there were 18,152 LPG taxis and 4031 LPG mini buses registered in Hong Kong [6]. With the government's strong incentive for phasing out diesel commercial vehicles, the number of LPG taxis and mini buses is expected to surge continuously.
On 30 April 2017, a compressed natural gas (CNG) fuelled taxi caught fire and exploded in Buona Vista, Singapore [7]. Four people, including an SCDF officer, were injured. The incident once again alerted the public to the fire and explosion hazards of clean fuel vehicles.
The LPG filling station in Hong Kong is classified as a "notifiable gas installation" under the Gas Safety Ordinance Cap. 51 [8]. As part of the approval requirement, a quantitative risk assessment is required to demonstrate that the risk levels associated with stations are acceptable. Moreover, all operators of the stations shall receive proper training on filling operation and emergency response. Specially designed safety features are also required for LPG stations.
The safety of filling facilities has drawn a wide range of public concerns and detailed construction requirements have been formulated [9]. Yet, few people were aware that LPG vehicle maintenance may possibly be ticking time bombs until a deadly explosion occurred at a garage in Hong Kong in April 2015 [10] (Figure 1). Although the use of hydrogen vehicles is not yet popular in Hong Kong, the number of hydrogen vehicles is surging all over the world; still, explosion hazards are not properly discussed. There have been many reported fires and explosions caused by fuel.

An Example Garage
An example garage of size 8.1 m (x-direction) × 3.6 m (y-direction) × 3.9 m (z-direction) as shown in Figure 2a is used in the simulation. A roller shutter is provided at x = 8.1 m with a height of 3.6 m from ground level. The rolling shutter is assumed to be able to withstand a pressure of 0.2 bar.
The horizontal direction of the computational domain, i.e., x-direction, is extended to 1.5 m outside the roller shutter.
A typical amount of 95.5 L (48.705 kg) LPG, which is a full tank for a typical fiveseater taxi, would be released [24]. According to Van den Schoor et al. [25], the vapour mass rates for the most probable release scenarios of LPG are shown in Table 1. LPG is assumed to be composed of 40% propane and 60% butane. Although the use of hydrogen vehicles is not yet popular in Hong Kong, the number of hydrogen vehicles is surging all over the world; still, explosion hazards are not properly discussed. There have been many reported fires and explosions caused by fuel.

An Example Garage
An example garage of size 8.1 m (x-direction) × 3.6 m (y-direction) × 3.9 m (z-direction) as shown in Figure 2a is used in the simulation. A roller shutter is provided at x = 8.1 m with a height of 3.6 m from ground level. The rolling shutter is assumed to be able to withstand a pressure of 0.2 bar.
The horizontal direction of the computational domain, i.e., x-direction, is extended to 1.5 m outside the roller shutter.
A typical amount of 95.5 L (48.705 kg) LPG, which is a full tank for a typical five-seater taxi, would be released [24]. According to Van den Schoor et al. [25], the vapour mass rates for the most probable release scenarios of LPG are shown in Table 1. LPG is assumed to be composed of 40% propane and 60% butane.

Numerical Simulation
The grid system with mesh size used in software FLACS is 0.3 m × 0.3 m × 0.3 m as shown in Figure 2b.
According to a previous study by Hansen et al. [26], the FLACs simulation results were reasonably close to experimental data for different experimental configurations with a maximum grid size ∆x, ∆y and ∆z given in terms of the cloud volume V as: Max (∆x, ∆y, ∆z) = 0.1 × V 1/3 . For a cloud volume V of about 58.32 m 3 (8.1 m × 3.6 m × 2 m), the maximum allowable grid size would be 0.39 m.
Through assigning the boundary conditions of the computation domain, possible effects of surface compression at the roller shutter to the convective flow field are considered.
In the preliminary study of the explosion behaviour of an LPG taxi in a garage, three scenarios at different operation times judging from the fire investigation report are considered:  Simulation C1: The ignition process starts at 50 s after the start of the leakage. The ignition point is assumed to be x = 4.05 m, y = 1.45 m and z = 0.15 m with x, y, z directions shown in the figure.  Simulation C2:

Numerical Simulation
The grid system with mesh size used in software FLACS is 0.3 m × 0.3 m × 0.3 m as shown in Figure 2b.
According to a previous study by Hansen et al. [26], the FLACs simulation results were reasonably close to experimental data for different experimental configurations with a maximum grid size ∆x, ∆y and ∆z given in terms of the cloud volume V as: Through assigning the boundary conditions of the computation domain, possible effects of surface compression at the roller shutter to the convective flow field are considered.
In the preliminary study of the explosion behaviour of an LPG taxi in a garage, three scenarios at different operation times judging from the fire investigation report are considered: The ignition process starts at 50 s after the start of the leakage. The ignition point is assumed to be x = 4.05 m, y = 1.45 m and z = 0.15 m with x, y, z directions shown in the figure. The ignition process starts at 100 s after the start of the leakage.
• Simulation C3: The ignition process starts at 150 s after the start of the leakage. The monitoring points M1 and M3 are inside the garage, M2 is outside as in Table 2 to present CFD simulation results.

Pressure Explosion
Pressure explosion comes from the building up of pressure during combustion. For monitoring point M1, i.e., near the roller shutter in the garage, the variations of pressure in the above three scenarios are as shown in Figure 3a-c. The ignition process starts at 100 s after the start of the leakage.  Simulation C3: The ignition process starts at 150 s after the start of the leakage. The monitoring points M1 and M3 are inside the garage, M2 is outside as in Table 2 to present CFD simulation results.

Pressure Explosion
Pressure explosion comes from the building up of pressure during combustion. For monitoring point M1, i.e., near the roller shutter in the garage, the variations of pressure in the above three scenarios are as shown in Figure 3a-c. In simulation C1, before the ignition at t = 50 s, pressure is built up inside of the garage due to the leakage of LPG. However, the pressure is below 0.2 barg and the roller shutter is kept in the closed position. At t = 50 s, the ignition process starts and a rapid surge in pressure is observed. At about t = 53 s, the pressure reaches 0.2 barg and a drop in pressure is noted due to the opening of the roller shutter for controlling pressure rise. In simulation C1, before the ignition at t = 50 s, pressure is built up inside of the garage due to the leakage of LPG. However, the pressure is below 0.2 barg and the roller shutter is kept in the closed position. At t = 50 s, the ignition process starts and a rapid surge in pressure is observed. At about t = 53 s, the pressure reaches 0.2 barg and a drop in pressure is noted due to the opening of the roller shutter for controlling pressure rise.
In simulation C2, alike to Simulation C1, pressure is built up constantly inside of the garage due to the leakage of LPG before the ignition. A jump of pressure is observed during the ignition at t = 100 s, followed by a drop in pressure due to the opening of the roller shutter at 0.2 barg.
In simulation C3, pressure is built up constantly inside of the garage due to the leakage of LPG. At t = 102 s, the pressure is over 0.2 barg and the roller shutter is opened due to the over pressure. A drop in pressure due to the opening of the roller shutter at 0.2 barg and a surge in pressure is observed during the ignition at t = 150 s.
For monitoring point M2, i.e., near the roller shutter out of the garage, as shown in Figure 4a, surges are noted at the time of ignition in all three scenarios. The maximum surge in pressure is noted in Simulation C2. In simulation C2, alike to Simulation C1, pressure is built up constantly inside of the garage due to the leakage of LPG before the ignition. A jump of pressure is observed during the ignition at t = 100 s, followed by a drop in pressure due to the opening of the roller shutter at 0.2 barg.
In simulation C3, pressure is built up constantly inside of the garage due to the leakage of LPG. At t = 102 s, the pressure is over 0.2 barg and the roller shutter is opened due to the over pressure. A drop in pressure due to the opening of the roller shutter at 0.2 barg and a surge in pressure is observed during the ignition at t = 150 s.
For monitoring point M2, i.e., near the roller shutter out of the garage, as shown in Figure 4a, surges are noted at the time of ignition in all three scenarios. The maximum surge in pressure is noted in Simulation C2.  For monitoring point M3, i.e., near the ceiling above the point of ignition in Figure  4b, the maximum pressures recorded in all three scenarios are 0.2 barg. For simulations C1 and C2, the pressures are built after the ignition of leaked LPG. However, the pressure in simulation C3 is built up to 0.2 barg due to the accumulation of LPG in the garage.

Temperature
The transient temperature distributions for the three scenarios are compared in  For monitoring point M3, i.e., near the ceiling above the point of ignition in Figure 4b, the maximum pressures recorded in all three scenarios are 0.2 barg. For simulations C1 and C2, the pressures are built after the ignition of leaked LPG. However, the pressure in simulation C3 is built up to 0.2 barg due to the accumulation of LPG in the garage.

Temperature
The transient temperature distributions for the three scenarios are compared in Figure 5.  In simulation C1, the surge in temperature is localized near the point of leakage and ignition. However, the extent of temperature increase in simulations C2 and C3 is much greater than that of C1. Similar post-explosion transient temperature profiles are noted in both simulations C2 and C3. In simulation C1, the surge in temperature is localized near the point of leakage and ignition. However, the extent of temperature increase in simulations C2 and C3 is much greater than that of C1. Similar post-explosion transient temperature profiles are noted in both simulations C2 and C3. The area of high temperature then propagates in the upward direction which is a natural phenomenon due to the hot gas movement.
It is also noted that the transient temperature profiles in simulation C2 are extended towards positive x-direction when compared with that of C3. In simulation C2, higher explosion pressure is generated by the explosion which would push the hot gas and cause the elongation of transient temperature profile.
The temperature profiles measured at monitoring point 1 are shown in Figure 6a. The area of high temperature then propagates in the upward direction which is a natural phenomenon due to the hot gas movement.
It is also noted that the transient temperature profiles in simulation C2 are extended towards positive x-direction when compared with that of C3. In simulation C2, higher explosion pressure is generated by the explosion which would push the hot gas and cause the elongation of transient temperature profile.
The temperature profiles measured at monitoring point 1 are shown in Figure 6a.  For C3, the maximum temperature at the monitoring point 1 is observed at t = 152.8 with T = 462.04 K.
The temperature profiles measured at monitoring point 2 are shown in Figure 6b. The temperature profiles measured at monitoring point 3 are shown in Figure 6c. Among the three monitoring points, the maximum surge in temperature is noted in Simulation C2.

Horizontal Component of Velocity
The horizontal component of fluid velocity U at the three monitoring points M1 to M3 are shown in Figure 7. Positive U refers to out-flowing of fluid from the garage while negative U means the flowing of fluid into the garage. For C3, the maximum temperature at the monitoring point 1 is observed at t = 152.8 with T = 462.04 K.
The temperature profiles measured at monitoring point 2 are shown in Figure 6b. The temperature profiles measured at monitoring point 3 are shown in Figure 6c. Among the three monitoring points, the maximum surge in temperature is noted in Simulation C2.

Horizontal Component of Velocity
The horizontal component of fluid velocity U at the three monitoring points M1 to M3 are shown in Figure 7. Positive U refers to out-flowing of fluid from the garage while negative U means the flowing of fluid into the garage.  Among the three monitoring points, the maximum change in velocity is also noted in Simulation C2.

Discussion
Among the three monitoring points, the maximum surge in velocity is also noted in Simulation C2.
In the above simulation scenarios, it is noted that scenario C2 would result in a greater increase in pressure, temperature and velocity than that of C3.
In C3, the pressure is built up constantly inside of the garage due to the leakage of LPG until the pressure is over 0.2 barg. Then the roller shutter is opened due to the over pressure and LPG is allowed to diffuse out of the garage. The tragic consequences of the explosion are lessened in Simulation C3.
To further study the effect of ventilation on explosion [27][28][29][30][31], the following simulations are conducted: • Simulation C2-1: Simulation configuration is exactly the same as for C2, with the ignition process started at 100 s after the start of leakage. Among the three monitoring points, the maximum change in velocity is also noted in Simulation C2.

Discussion
Among the three monitoring points, the maximum surge in velocity is also noted in Simulation C2.
In the above simulation scenarios, it is noted that scenario C2 would result in a greater increase in pressure, temperature and velocity than that of C3.
In C3, the pressure is built up constantly inside of the garage due to the leakage of LPG until the pressure is over 0.2 barg. Then the roller shutter is opened due to the over pressure and LPG is allowed to diffuse out of the garage. The tragic consequences of the explosion are lessened in Simulation C3.
To further study the effect of ventilation on explosion [27][28][29][30][31], the following simulations are conducted:  Simulation C2-1: Simulation configuration is exactly the same as for C2, with the ignition process started at 100 s after the start of leakage. With the presence of a gap of 0.3 m at the bottom of the roller shutter, a smaller increase in temperature, velocity and pressure induced by the ignition at t = 100 s is observed.
The results highlighted the importance of low-level ventilation in a garage, which might effectively lessen the tragic consequences of an explosion.
The same simulated garage was used. According to the fuel leak simulation by Swain [38], a hydrogen vehicle leaked 3.4 pounds (1.5 kg) of hydrogen in 100 s. The mass flow rate is then assumed to be 0.015 kgs −1 . Taking the Toyota Mirai as a reference hydrogen vehicle in the simulation, the Mirai has two hydrogen tanks with a three-layer structure made of carbon fibre-reinforced polymer and the tanks have a combined capacity of 5 kg. The whole tank would be completely vented out in about 333 s. Two more simulations were carried out on hydrogen vehicles:  Simulation H1 The ignition process of hydrogen starts at 333 s after the start of leakage. For M1 near the roller shutter inside the garage, the pressure is built up constantly inside of the garage due to the leakage of hydrogen before the ignition as shown in Figure  9. A jump in the pressure is observed at ignition at t = 333 s. However, the small increase in pressure is still far from being sufficient to open roller shutter, although the full tank of hydrogen has leaked out.  With the presence of a gap of 0.3 m at the bottom of the roller shutter, a smaller increase in temperature, velocity and pressure induced by the ignition at t = 100 s is observed.
The results highlighted the importance of low-level ventilation in a garage, which might effectively lessen the tragic consequences of an explosion.
The same simulated garage was used. According to the fuel leak simulation by Swain [38], a hydrogen vehicle leaked 3.4 pounds (1.5 kg) of hydrogen in 100 s. The mass flow rate is then assumed to be 0.015 kgs −1 . Taking the Toyota Mirai as a reference hydrogen vehicle in the simulation, the Mirai has two hydrogen tanks with a three-layer structure made of carbon fibre-reinforced polymer and the tanks have a combined capacity of 5 kg. The whole tank would be completely vented out in about 333 s. Two more simulations were carried out on hydrogen vehicles: The ignition process of hydrogen starts at 333 s after the start of leakage. For M1 near the roller shutter inside the garage, the pressure is built up constantly inside of the garage due to the leakage of hydrogen before the ignition as shown in Figure 9. A jump in the pressure is observed at ignition at t = 333 s. However, the small increase in pressure is still far from being sufficient to open roller shutter, although the full tank of hydrogen has leaked out. With the presence of a gap of 0.3 m at the bottom of the roller shutter, a smaller increase in temperature, velocity and pressure induced by the ignition at t = 100 s is observed.
The results highlighted the importance of low-level ventilation in a garage, which might effectively lessen the tragic consequences of an explosion.
The same simulated garage was used. According to the fuel leak simulation by Swain [38], a hydrogen vehicle leaked 3.4 pounds (1.5 kg) of hydrogen in 100 s. The mass flow rate is then assumed to be 0.015 kgs −1 . Taking the Toyota Mirai as a reference hydrogen vehicle in the simulation, the Mirai has two hydrogen tanks with a three-layer structure made of carbon fibre-reinforced polymer and the tanks have a combined capacity of 5 kg. The whole tank would be completely vented out in about 333 s. Two more simulations were carried out on hydrogen vehicles:  Simulation H1 The ignition process of hydrogen starts at 333 s after the start of leakage. For M1 near the roller shutter inside the garage, the pressure is built up constantly inside of the garage due to the leakage of hydrogen before the ignition as shown in Figure  9. A jump in the pressure is observed at ignition at t = 333 s. However, the small increase in pressure is still far from being sufficient to open roller shutter, although the full tank of hydrogen has leaked out.  The ignition process of hydrogen starts at 100 s after the start of leakage. The temperature measured at M3 near the ceiling above the point of ignition is provided in Figure 10. The ignition process of hydrogen starts at 100 s after the start of leakage. The temperature measured at M3 near the ceiling above the point of ignition is provided in Figure 10. The transient temperature distributions for scenario H2 are shown in Figure 11. The ignition process of hydrogen starts at 100 s after the start of leakage. The temperature measured at M3 near the ceiling above the point of ignition is provided in Figure 10. The temperature of hot gas is found much lower than the case of LPG vehicle explosion. Unlike the transient temperature profile in Figure 5, the hot gas region is located at the upper part of the garage and pushed toward the ends of the garage.
Although the ignition time is at t = 100 s for simulation C2 for LPG taxis and H2 for hydrogen taxis, the surge in temperature at M3 near the ceiling above the point of ignition is found to be much greater in case of LPG leak in C2 ( Figure 12). Two more simulations on lower-level ventilation for H2 are carried out:  Simulation H2a Simulation configuration is exactly the same as H2, with the ignition process started at 100 s after the start of leakage (d = 0 m).  Simulation H2b Figure 11. Transient temperature distributions of simulation H2.
The temperature of hot gas is found much lower than the case of LPG vehicle explosion. Unlike the transient temperature profile in Figure 5, the hot gas region is located at the upper part of the garage and pushed toward the ends of the garage.
Although the ignition time is at t = 100 s for simulation C2 for LPG taxis and H2 for hydrogen taxis, the surge in temperature at M3 near the ceiling above the point of ignition is found to be much greater in case of LPG leak in C2 ( Figure 12). The temperature of hot gas is found much lower than the case of LPG vehicle explosion. Unlike the transient temperature profile in Figure 5, the hot gas region is located at the upper part of the garage and pushed toward the ends of the garage.
Although the ignition time is at t = 100 s for simulation C2 for LPG taxis and H2 for hydrogen taxis, the surge in temperature at M3 near the ceiling above the point of ignition is found to be much greater in case of LPG leak in C2 ( Figure 12). Two more simulations on lower-level ventilation for H2 are carried out:  Simulation H2a Simulation configuration is exactly the same as H2, with the ignition process started at 100 s after the start of leakage (d = 0 m).  Simulation H2b Two more simulations on lower-level ventilation for H2 are carried out: Simulation configuration is exactly the same as H2, with the ignition process started at 100 s after the start of leakage (d = 0 m).

• Simulation H2b
Simulation configuration is exactly the same as H2, with the ignition process started at 100 s after the start of leakage, except with a gap of 0.3 m located at the bottom of the roller shutter (d = 0.3 m). Simulation configuration is exactly the same as H2, with the ignition process started at 100 s after the start of leakage, except with a gap of 0.3 m located at the bottom of the roller shutter (d = 0.3 m).
The pressure variations measured at M3 in Simulation H2a and H2b are shown in Figure 13a. Due to the 0.3 m opening at the bottom of the roller shutter, the pressure fails to build up in Simulation H2b. The temperature variations measured at M3 in Simulation H2a and H2b are shown in Figure 13b. Similar temperature profiles are noted in both cases. The peak temperature measured in Simulation H2b is slightly lower than that in H2a. Since hydrogen is lighter than the air, the leaked hydrogen would accumulate in the upper part of the garage. Lowlevel ventilation could not improve the situation in this case. The slight drop is attributed to the heat exchange with the atmosphere through the 0.3 m opening.

Conclusions
The explosion hazards of LPG and hydrogen vehicles are outlined in this paper. The technology of LPG fuelled vehicles has been well established for a long time. The major hazard of LPG fuelled vehicle is the gross leakage under failure condition of the fuel tanks and associated piping [39].
Our simulation results highlighted the importance of low-level ventilation in lessening the hazardous consequences of an LPG vehicle explosion.
Comparing the simulation results of C2 and H2, it is noted that the resulted surge in temperature and pressure by the explosion at t = 100 s are smaller in the case of hydrogen. However, the explosion risk of hydrogen should not be overlooked. The temperature variations measured at M3 in Simulation H2a and H2b are shown in Figure 13b. Similar temperature profiles are noted in both cases. The peak temperature measured in Simulation H2b is slightly lower than that in H2a. Since hydrogen is lighter than the air, the leaked hydrogen would accumulate in the upper part of the garage. Lowlevel ventilation could not improve the situation in this case. The slight drop is attributed to the heat exchange with the atmosphere through the 0.3 m opening.

Conclusions
The explosion hazards of LPG and hydrogen vehicles are outlined in this paper. The technology of LPG fuelled vehicles has been well established for a long time. The major hazard of LPG fuelled vehicle is the gross leakage under failure condition of the fuel tanks and associated piping [39].
Our simulation results highlighted the importance of low-level ventilation in lessening the hazardous consequences of an LPG vehicle explosion.
Comparing the simulation results of C2 and H2, it is noted that the resulted surge in temperature and pressure by the explosion at t = 100 s are smaller in the case of hydrogen. However, the explosion risk of hydrogen should not be overlooked.
Hydrogen could leak out of a system which is considered as gas-tight for other gases. Hydrogen also has a wide range of flammable limit (4-75%) while the flammability limit of LPG only ranges from 2.1-9.5%. Besides, the gross calorific value of hydrogen is 158.9 MJkg −1 which is much higher than 50.49 MJkg −1 of LPG [24].
Although the surge in temperature at M3 near the ceiling above the point of ignition for simulation H2 is much smaller than that in the case of an LPG leak in C2, this is due to the large difference in the flow rates adopted in the simulations. Given a large enough flow rate of hydrogen, the consequences of an hydrogen explosion would also be devastating.

Data Availability Statement:
The data presented in this study is available on request from the authors.