Numerical Simulation of Dispersion and Ventilation of Hydrogen Clouds in Case of Leakage Inside a Large-Scale Industrial Building

Abstract

As the attention to using hydrogen as a potential energy storage medium for power generation and mobility increases, hydrogen production, storage, and transportation safety should be considered. For instance, hydrogen’s extreme physical and chemical properties and the wide range of flammable concentrations raise many concerns about the current safety measures in processing other flammable gases. Hydrogen cloud accumulation in the case of leakage in confined spaces can lead to reaching the hydrogen lower flammability limit (LFL) within seconds if the hydrogen is not properly evacuated from the space. At Jülich Research Centre, hydrogen mixed with natural gas is foreseen to be used as a fuel for the central heating system of the campus. In this work, the release, dispersion, formation, and spread of the hydrogen cloud in the case of hydrogen leakage inside the central utility building of the campus are numerically simulated using the OpenFOAM-based containmentFOAM CFD codes. Additionally, different ventilation scenarios are simulated to investigate the behavior of the hydrogen cloud. The results show that locating exhaust openings close to the ceiling and the potential leakage source can be the most effective way to safely evacuate hydrogen from the building. Additionally, locating the exhaust outlets near the ceiling can decrease the combustible cloud volume by more than 25% compared to side openings far below the ceiling. Also, hydrogen concentrations can reach the LFL in case of improper forced ventilation after only 8 s, while it does not exceed 0.15% in the case of natural ventilation under certain conditions. The results of this work show the significant effect of locating exhaust outlets near the ceiling and the importance of natural ventilation to mitigate the effects of hydrogen leakage. The approach illustrated in this study can be used to study hydrogen dispersion in closed buildings in case of leakage and the proper design of the ventilation outlets for closed spaces with hydrogen systems.

1. Introduction

The international commitment to limit the global temperature rise to 1.5 °C and the Net Zero Emissions (NZE) by 2050 Scenario raised global interest in renewable energies and energy storage technologies. Such a scenario could only be achieved by replacing hydrocarbon fuels with another energy storage source with minimal carbon emission during its life cycle. Hydrogen has different physical properties; its higher gravimetric energy density compared to natural gas, for instance, makes it a possible energy storage medium and an alternative fuel. Therefore, hydrogen research and infrastructure investments surged during the last decade, hoping to achieve the NZE by 2050. The global hydrogen demand is expected to reach 210 Mt by 2030 to cover the increasing transition towards the zero-emission target. This demand is expected to reach 390 Mt by 2040 and 530 Mt by 2050 [1].

Accordingly, this surge in hydrogen demand increases global hydrogen production, transportation, and usage in industrial and residential buildings. The extreme physical and chemical properties of hydrogen compared to other gaseous fuels raise different safety concerns about the prevention and handling of accidental leakage of hydrogen in both confined and open spaces. Therefore, hydrogen safety research gained much attention from industry, academia, and governments to ensure the safe usage of hydrogen.

To validate the usage of CFD simulation in analyzing hydrogen dispersion, Gianissi et al. [2] compared the capabilities of Reynolds-averaged Navier–Stokes (RANS), large Eddy simulations (LES), and laminar flow CFD approaches to simulate helium release with a low Reynolds number in a closed facility. By comparing the CFD and experimental results, they concluded that LES and RANS predicted the gas distribution inside the facility well. At the same time, the laminar flow approach managed to simulate the stratification. Also, Malakhov et al. [3] simulated hydrogen release in a semi-closed facility representing an underground tunnel. However, their study focused on simulating hydrogen release from different nozzle sizes and ventilation rates. After comparing experimental results, they concluded that such simulations can agree with experiments and be used to determine the different safety scenarios of hydrogen release.

Since hydrogen fuel cell vehicles (FCVs) is one of the main applications, where hydrogen is expected to play an important role in the energy transition, hydrogen leakage from FCVs in closed spaces like tunnels is the focus of different research works. Ismail et al. [4] compared the results of simulating hydrogen release in a scaled tunnel with the experimental results and used the validated boundary conditions to simulate the same tunnel under different wind speeds. Their study showed a methodology to calculate the threshold hydrogen wind speed in the tunnel, after which a significant change in the behavior of the hydrogen cloud occurs. Regarding forced ventilation of tunnels, Suzuki et al. [5] studied the effect of the different ventilation flow rates and different configurations of tunnels. They have concluded that the configuration of tunnels, like their slopes and fan rooms, can significantly influence the hydrogen distribution and concentrations in the tunnel. A deeper analysis of the behavior of hydrogen jet dispersion below and around a car when hydrogen leaks from its tank inside a tunnel was studied by Koutsourakis et al. [6]. Their study numerically simulated hydrogen leakage from a typical 700 bar tank inside a 7.1 m high tunnel. The simulations showed that the hydrogen-filled space below the car started moving upwards to the tunnel’s ceiling within 2 s, and different flow phenomena were observed. Middah and Hansen [7] also studied the effect of the layout of tunnels in the case of hydrogen leakage from cars and buses to show how tunnels can be designed to reduce the probability of the worst case scenario of hydrogen accumulation in tunnels.

Another important application of hydrogen safety for FCVs is the study of hydrogen dispersion in underground garages. Several design parameters of underground carparks can significantly affect their safety in case of hydrogen leakage. One design parameter is the leakage and the ventilation outlets studied by Huang et al. [8]. Their study studied hydrogen dispersion after leakage from different locations inside the car park for different leakage times and ventilation outlets’ layouts. The simulations showed that leakage locations near the corners of the building caused larger concentrations near these corners. However, this did not affect the far-field hydrogen concentration. They also found that locating ventilation openings near the corners can lead to a faster evacuation of hydrogen from the space. Another parameter is the height of the ceiling and its structural elements, such as beams, for instance. Li et al. [9] studied the effect of the height of cross beams in the ceiling of underground car parks on hydrogen dispersion. They showed that the height of the beams can play an important role in the safe design of underground garages since it can block hydrogen distribution along the ceiling and, hence, form locations with high hydrogen concentrations. For residential garages, Hajji et al. [10] studied the effect of leakage location, duration, and mass flow rate on hydrogen concentration accumulated near the ceiling of a residential garage with a prismatic ceiling configuration. They concluded that when the leakage was located at the center of the garage, the formed hydrogen layer showed more stability in stratification. Also, longer leakage times can result in higher hydrogen concentrations accumulating near the ceiling. Hajji et al. [11] studied three different geometries of ventilation openings near the ceiling of the same garage in the former study. They have also studied four different aspect ratios for rectangular ventilation openings. They concluded that the transverse rectangular ventilation opening showed the highest hydrogen extraction rates.

Besides FCVs, another promising application of hydrogen in transportation is in the maritime sector. Since cargo ships are the backbone of world trade and their warehouses are responsible for about 3% of the global carbon emissions [12], using hydrogen in such applications is a great opportunity to cut these carbon emissions. Li et al. [13] studied the hydrogen distribution inside a fuel cell ship in the case of hydrogen leakage. They also studied different distributions of natural and mechanical ventilation openings and their effects on hydrogen distribution. They concluded that mechanical ventilation showed better evacuation performance for hydrogen, depending on the vents’ distribution and locations. Also, they found that hydrogen accumulates near walls and corners, which were the best locations to locate sensors. Xie et al. [14] also studied the effects of ventilation rates, hydrogen leakage diameters, and room temperatures on the hydrogen diffusion in a ship’s engine room. Their study found that higher ventilation rates can help decrease the hydrogen concentration. Also, they concluded that higher room temperature enhances the vertical stratification of hydrogen. Guan et al. [15] studied the effects of hydrogen pipe diameters and hydrogen detectors’ locations to develop some concepts for safe ship design. In their work, they simulated hydrogen leakage in a fuel cell room of a ship using CFD. Their study concluded that smaller pipe diameters can lower the risk of reaching flammable hydrogen concentrations. Additionally, they have noted that the location of hydrogen sensors is affected by the location of hydrogen leakage and the obstacles between the hydrogen source and the sensor.

From the state-of-the-art research summarized, it is clear that hydrogen dispersion due to leakage in complex, large-scale facilities, like factories, was not well studied. Because of the high potential of using hydrogen in industrial applications, such a study should be important to ensure the safe application of hydrogen in such buildings. Therefore, this study aims to simulate hydrogen cloud dispersion inside the central utility building at Jülich Research Centre with details like overhead piping and mechanical equipment. It should be noted here that the building is not originally designed to contain hydrogen pipes and equipment. Therefore, different leakage and ventilation scenarios are shown in this work to understand the behavior of hydrogen clouds in each case and to analyze the effectiveness of each ventilation strategy.

Section 2 shows the theoretical background of the CFD simulations applied in this work and describes the CFD setup and the grid used in the analysis. Then, in Section 3, the results of the CFD simulation are shown and analyzed. Finally, in Section 4, conclusions are explained.

2. Numerical Models and Setup

The literature introduces many turbulence models to simulate and capture the different turbulent flow phenomena. However, the turbulence models should be modified to enable the simulation of buoyancy-driven, low-density gases like hydrogen and helium. Hydrogen flows turbulently during its leakage from high-pressure vessels or pipes. Accordingly, the turbulent CFD models should correctly capture the hydrogen flow’s different turbulent flow features, like turbulent mixing and buoyancy. The applied turbulence models and their corrections to simulate the buoyancy of hydrogen are introduced. Also, the numerical setup of the CFD solution case is shown.

2.1. The containmentFOAM Package

The containmentFOAM simulation libraries are implemented, validated, and maintained by the Reactor Safety research group at Jülich Research Centre. The primary usage of these libraries is to analyze gas transport, such as hydrogen, during nuclear power plant accidents. Based on OpenFOAM ^® v9 (Greenshields and Weller [16]), containmentFOAM benefits from all features of the open-source CFD software of OpenFOAM in addition to other libraries and solvers designed for nuclear and hydrogen safety applications. The containmentFOAM libraries were extensively validated and verified against experimental results for modeling and simulations of nuclear applications, like the work of Kelm et al. [17], and for the non-nuclear hydrogen applications carried out by Yassin et al. [18].

2.2. Solution for Unsteady Multi-Species Gas Mixture

CFD simulations are based on solving the mass, momentum, and energy conservation equations. Fluid species should also be conserved in case of simulating a multi-gas fluid. This means that the mass of each gas, in the case of a non-reacting fluid, should be conserved [19]. Accordingly, solving the species conservation equation for each species enables the calculation of the mass fractions of each gas over the whole computational domain.

The mass conservation equation, in this case, reads

$${\displaystyle \frac{\partial \rho }{\partial t}}+\nabla ·\left(\rho \overrightarrow{U}\right)=0,$$

(1)

and the momentum conservation equation reads

$${\displaystyle \frac{\partial \rho \overrightarrow{U}}{\partial t}}+\nabla ·\left(\rho \overrightarrow{U}\otimes \overrightarrow{U}\right)=-\nabla p+\nabla ·\tau +\rho \overrightarrow{g},$$

(2)

where the shear stress tensor ($\tau$) takes the form

$$\tau =\rho (\nu +{\nu }_t)\left[\nabla \overrightarrow{U}+{\left(\nabla \overrightarrow{U}\right)}^T-{\displaystyle \frac{2}{3}}\delta \nabla ·\overrightarrow{U}\right],$$

(3)

where ${\nu }_t$ is the turbulent eddy viscosity, $\nu$ is the kinematic viscosity, and $\delta$ is the Kronecker delta. The species transport equation reads

$${\displaystyle \frac{\partial \rho Y_i}{\partial t}}+\nabla ·\left(\rho \overrightarrow{U}{Y}_i\right)=-\nabla ·{\overrightarrow{J}}_i,$$

(4)

where $Y_i$ is the mass fraction of species i, and ${\overrightarrow{J}}_k$ is the diffusion mass flux calculated from Fick’s law as

$${\overrightarrow{J}}_i=\rho \left(D_{i,m}+{\displaystyle \frac{{\nu }_t}{S{c}_t}}\right)\nabla Y_i.$$

(5)

Here $S{c}_t$ is the turbulent Schmidt number, and $D_{i,m}$ is the molecular diffusivity of species i. The energy conservation equations, in this case, read

$${\displaystyle \frac{\partial \rho h_{tot}}{\partial t}}+\nabla ·\left(\rho \overrightarrow{U}{h}_{tot}\right)={\displaystyle \frac{\partial p}{\partial t}}-\nabla ·{\overrightarrow{\dot{q}}}^{\prime \prime }+\nabla ·(\overrightarrow{U}·\tau )+\overrightarrow{U}·\left(\rho \overrightarrow{g}\right)-\nabla ·{\overrightarrow{\dot{q}}}_{rad}^{\prime \prime },$$

(6)

where $h_{tot}$ is the total enthalpy $h_{tot}=h+{\displaystyle \frac{1}{2}}{\left|\overrightarrow{U}\right|}^2$ and the potential energy represented by the term $\overrightarrow{U}·\left(\rho \overrightarrow{g}\right)$.

Applying the simple gradient diffusion hypothesis (SGDH), the turbulent Schmidt and Prandtl numbers in Equation (6) take the values $S{c}_t=P{r}_t=0.9$. Kelm et al. [20] illustrate the complete details of the applied numerical approach.

2.3. k-$\omega$ SST Turbulence Model and Its Modification for Buoyancy

Since the case studied in this work involves hydrogen flow around complex structures, treating the flow near the walls is vital in simulating such a flow. The k-$\omega$ SST turbulence model [21] is used in this work to properly model turbulence near the wall using wall functions. Also, many validation simulations, like those in the work of Abe et al. [22] and Kampili et al. [23], have shown the importance of the buoyancy-induced turbulence effects to simulate the stratified gas layers. Because of the significant difference in density between the air and hydrogen, the buoyancy effect is crucial during the simulation. Accordingly, the k-$\omega$ SST turbulence model can be modified with turbulence production terms added to become [21]

$${\displaystyle \frac{\partial \left(\rho k\right)}{\partial t}}+\nabla ·\left(\rho \overrightarrow{U}k\right)=\nabla ·\left(\left(\mu +{\mu }_t{\sigma }_k\right)\nabla k\right)+\widetilde{P_k}-\rho \beta *\omega k+P_{k,b},$$

(7)

$$\begin{array}{c}{\displaystyle \frac{\partial \left(\rho \omega \right)}{\partial t}}+\nabla ·\left(\rho \overrightarrow{U}\omega \right)=\nabla ·\left(\left(\mu +{\mu }_t{\sigma }_{\omega }\right)\nabla \omega \right)+2\left(1-F_1\right){\displaystyle \frac{\rho {\sigma }_{\omega 2}}{\omega }}\nabla k·\nabla \omega \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hfill \\ \hfill +P_{\omega }+P_{\omega ,b}-Y_{\omega }.\end{array}$$

(8)

The variables in Equations (7) and (8) are shown in detail in [21]. By using the simple gradient diffusion hypothesis (SGDH), the production terms $P_{k,b}$ and $P_{\omega ,b}$ take the form

$$P_{k,b}=-{\displaystyle \frac{{\nu }_t}{{\sigma }_{\rho }}}{g}_i{\displaystyle \frac{\partial \rho }{\partial x_i}},$$

(9)

$$P_{\omega ,b}={\nu }_t((\gamma +1)C_3·max(P_{k,b},0)-P_{k,b}),$$

(10)

where ${\sigma }_{\rho }=1$ is the turbulent Schmidt number and $C_3$ is the dissipation coefficient.

2.4. Computational Setup

The central utility building with all its components is modeled using the FreeCAD software to generate the computational grid representing the void inside the structure shown in Figure 1, where air and the leaked hydrogen can flow. The grid is generated such that it takes into account all relevant equipment and pipes inside the building to enable the correct simulation of the flow. The following section discusses the generated grid, the hydrogen leakage parameters, and the simulated cases in this work.

2.4.1. Numerical Grid

Because of the complexity of the computational domain, a body-fitted Cartesian grid with proper boundary layer cells is generated to fit the CAD model of the building. A grid section showing the base cell size and the refined cells is shown in Figure 2. The base grid consists of cubic cells fitted to the geometry of the building and its internal structures. The base cubic cells of the grid have 0.5 m sides, as concluded from a detailed grid independence study explained in Appendix A. Additional refinement levels are applied directly above the leakage point and below the ceiling to enhance the accuracy of calculations of hydrogen concentrations and turbulence parameters in these areas. Such refinements were added according to the best practice guidelines (BPG) recommendations for CFD simulations for hydrogen safety [24]. Additionally, fine boundary-layer cells are added near the external and internal walls of the domain to ensure correct simulation of near-wall values of all fields.

In the case of natural ventilation scenarios explained later, the flow domain is extended about 15 m above the ventilation windows in the ceiling. This extension is applied to avoid any numerical effects of the boundaries representing the open atmosphere on the flow inside the building. Such a technique was applied in many other published studies, like the work performed by Giannissi et al. [25] and Matsuura et al. [26], and recommended by the BPG we referred to previously. From the explanation above, two different computational grids were used in this work: a grid with the extension above the building for natural ventilation and a grid with no extension used for other cases. The total volume of he computational domain of the two grids is around 15,050 m³ and 10,150 m³, respectively. Accordingly, the resulting extended and not-extended grids have about $15.7\times {10}^6$ and $11.3\times {10}^6$ cells, respectively.

2.4.2. Hydrogen Leakage Parameters

The hydrogen leakage from the pipe is assumed to occur at the indicated locations, namely, P1 and P2, in Figure 1. The leakage point P1 was chosen to be in the middle of the pipe with no obstacle between the leakage point and the ceiling. On the other hand, the leakage point P2 was chosen to be located under an intermediate ceiling in the structure to represent trapped hydrogen clouds. This pipe hole is assumed to be 40 mm in diameter and oriented 21.8° above the horizontal line to represent a pipe rupture due to collision with moving equipment. From the technical data of the building presented by the building operator, a hydrogen pipe with an 80 mm nominal diameter passes through the building and transports hydrogen to the boilers at 0.6 bar (gauge). The total length of the pipe running through the building is 89.7 m. Assuming that there is only one emergency shutdown valve (ESDV) at the pipe inlet to the building, the total volume of hydrogen inside the pipe running through the building is 0.436 m³.

From the parameters given and assumed above, the hydrogen pipe in case of leakage after the closure of the ESDV can be treated as a leaking tank with the same volume and pressure as the pipe. The leakage parameters can be calculated using the methodology explained by Molkov and Saffers [27] and Schefer et al. [28]. The leakage model used here shows that the complete charge of hydrogen, in this case, is wholly leaked after 0.245 s after the ESDV closure with an initial flow rate of 0.12423 m³/s. From the leakage model mentioned earlier, the leakage density is 0.9591 kg/m³, and the temperature is 255.87 K at the leakage location.

2.4.3. Boundary Conditions

All solid boundaries, i.e., pipes and mechanical equipment inside and the external walls, are treated as walls. This means these boundaries have non-slip conditions for velocity, fixed value temperatures, and zero-gradient for pressure. The external sides of the extended region explained in Section 2.4.1 are considered outlet–inlet boundaries. This outlet–inlet condition treats the boundary as zero-gradient in the case of outlet flow and with a fixed zero value in the case of inlet flow. The hydrogen flow is introduced to the domain using the volumetric source term approach explained by Kotchourko et al. [29]. Applying such a model to simulate leakage is valid due to the relatively low leakage volume flow rate, 0.124 m³/s, and accordingly low leakage velocity of 98.85 m/s. Such a velocity is well below the speed of sound, 346.05 m/s at 298.15 K ambient temperature. Accordingly, the jet effects can be neglected to benefit from the advantages of modeling the leakage as a mass source. This approach increases the required time step of the simulations at a Courant-Friedrichs–Lewy number (CFL) = 1 and, hence, significantly reduces the computational time of the simulation.

For all solid walls in the domain, the temperature boundary condition is assumed to be fixed at ambient temperature = 298.15 K. The external equipment temperatures in the building are kept as close as possible to the ambient temperature for the safety of the personnel working there. However, the exact temperature of the equipment during operation is not considered in this work.

In this study, different assumptions are considered to simplify the simulation inputs without reducing the applicability of the results in realistic scenarios. These assumptions are as follows:

• There is no temperature difference between the air inside the building and the air outside the building. This means no airflow occurs between the inside and outside of the building due to temperature difference and, hence, density difference.
• There is stagnant wind around the building. Considering wind speed and direction would result in induced airflow inside the building. This requires further detailed study. However, it is not considered in the current work.
• Leakage stops after 10 s from the start of the leakage. It is assumed that the emergency shutdown system in the building takes a few seconds after being triggered by hydrogen sensors until it is fully functional, i.e., completely shuts down the hydrogen flow in the pipe. The analysis of the effects of employing different sensor technologies and optimizing their locations is not in the scope of the current study.

2.5. Studied Cases and Scenarios

Different leakage scenarios are simulated in this work to study the behavior of the hydrogen cloud. From the parameters mentioned in the former section, the time needed for the pipe to stop leaking and the change in the flow rate during the discharge are calculated. Knowing these parameters, four different scenarios are studied:

• Scenario 1 (Extreme scenario): It is assumed that the hydrogen leakage from the leakage point P1 is undetected, and the hydrogen leaks inside the building without ventilation. This case is studied as a worst-case scenario (WCS) where no automatic action is taken to mitigate hydrogen leakage or accumulation.
• Scenario 2 (Natural ventilation): The hydrogen leaks from P1 and is detected and the shutdown happens 10 s after the leakage begins, as aforementioned. This results in an emergency shutdown of the hydrogen supply to the building. The roof ventilation windows are also opened to evacuate the hydrogen–air mixture from the building. The windows are assumed to be fully opened.
• Scenario 3 (Mechanical ventilation): The leakage from P1 is stopped also after 10 s, and the hydrogen evacuation outside the building is started. However, in this scenario, the existing ventilation fan and air outlets in the building are assumed to be used to evacuate the hydrogen–air mixture. Before the detection of hydrogen, i.e., after 10 s from the start of the leakage, the total volume flow rate of exhaust air is assumed to be 0.03 m³/m²/min, as recommended by the National Fire Protection Association (NFPA): Hydrogen Technology Code [30]. The total ventilation flow is calculated to be 4.371 m³/s.

After the detection, two sub-scenarios are studied: Scenario 3-a, where the ventilation rate is increased to 13.888 m³/s ≈ 5 air-changes per hour (ACH). This flow rate is the maximum capacity of the installed ventilation fans in the building. In Scenario 3-b, five times the ventilation flow rate, i.e., 25 ACH, recommended by FM Global Property Loss Prevention Data Sheet [31], is evacuated from the building. This flow is calculated to be 69.44 m³/s.

• Scenario 4 (Ceiling ventilation): This scenario assumes the same parameters as in Scenario 3-a. The only difference is that the exhaust outlets are located on the ceiling of the machine room. The primary purpose of this scenario is to study the effect of the location of the exhaust outlets on the size of the hydrogen cloud in the building.
• Scenario 5 (Leakage with obstacles): This scenario assumes the same parameters as in Scenario 3-a, with the leakage point P2 indicated in Figure 1.

In each scenario, hydrogen concentrations were calculated at seven different measurement points to quantitatively study the behavior of hydrogen inside the building. Figure 3 shows the locations of the hydrogen calculation points.

3. Simulations and Results

Comparing the scenarios explained in the last section leads to a deeper understanding of the hydrogen cloud behavior and a better assessment of risks resulting from hydrogen leakage. This section compares the hydrogen concentration values at specific locations in the building and the combustible hydrogen cloud volume to give an idea about the proper ventilation strategy.

3.1. Computational Time and Effort

The simulations of the five scenarios mentioned above required different computational efforts depending on the number of cells and the simulation time step, resulting from limiting the Courant–Friedrichs–Lewy (CFL) number to 0.99. The grid extended above the building used for the natural ventilation case was simulated for 24 s from the start of the leakage and required 6340 core hours (core-h) per simulation second (sim. s). For the grid with no extension used in the forced ventilation scenarios, the simulation cases were simulated for 60 s and required 1130 core-h/sim. s.

This significant difference in computational effort happens for different reasons. The first reason is the difference in the number of cells. The extended grid has almost 40% more cells than the other grid. The second reason is the relatively faster flow of hydrogen near the refined ceiling cells, which decreases the time step required to keep the CFL value close to 0.99. Other computer reasons for communication between cores could also have different effects, but these are not studied in this work.

3.2. Scenario 1: No-Ventilation, No-Detection

In this extreme scenario, the hydrogen continues to flow into the space without any outlet outside the building, which leads to its continuous accumulation. Figure 4a shows the contour surfaces of the concentrations 1%, 2%, 3%, and 4% of hydrogen within the building after 21 s of continuous hydrogen leakage. A thick hydrogen cloud accumulates below the ceiling, and the thickness of this cloud increases near the corners due to buoyancy and stratification.

It should be noted that this scenario is extreme and should not occur in practical cases since different safety measures have already been applied in the existing hydrogen pipeline in the building. One of the existing safety systems is the ESDV, whose operation is discussed in the following scenarios.

3.3. Scenario 2: Natural Ventilation

In the case of natural ventilation of the building shown in Figure 4b, the hydrogen cloud spreads horizontally below the ceiling until it meets the ventilation windows in the ceiling shown in Figure 1. As soon as the hydrogen reaches the openings, it starts to flow outside the building, driven by the buoyancy forces as shown in Figure 4b. A rapid dilution of the hydrogen concentration near the ceiling occurs after shutting down the hydrogen flow in the pipeline.

A significant drop in hydrogen concentration and cloud thickness near the ceiling can be seen, especially in the areas covered with the ceiling windows. This is clear from the comparison of Figure 4a,b, where the contour surfaces show less than 4% concentration areas. Nevertheless, the ventilation openings influence parts of the hydrogen cloud flow outside the area.

3.4. Scenario 3: Forced Ventilation

Figure 4c shows that the hydrogen cloud spreads rapidly near the ceiling despite active ventilation. By analyzing the formed hydrogen cloud in this case, it is found that the hydrogen cloud accumulates above the exhaust outlets. The outlets are about 2.5 m below the ceiling of the machine hall, which violates the NFPA 2 [30] Art. 6.17, the Occupational Safety and Health Association (OSHA) Art. 1910.103 [32], and many other safety standards. These standards mandate the placement of hydrogen evacuation outlets at a maximum of 30 cm below the highest point of the building. This value is recommended for the hydrogen cloud accumulation below the ceiling, as shown in Figure 5. Even in the simulations of Scenario 3-b, i.e., the exhaust flow rate is 25 ACH, the exhaust system is still not as effective as natural ventilation because the hydrogen cloud escapes the surrounding area of the evacuation outlet. This shows the importance of correctly locating the exhaust openings to prevent hydrogen dispersion in case of leakage.

3.5. Scenario 4: Forced Ventilation from Ceiling

The proposed exhaust openings in this scenario have the same number, diameters, and flow rates as in scenario 3-a. The only difference between this scenario and Scenario 3-a is the location of the openings, which is, in this case, on the ceiling. Figure 4d shows a significant improvement in the evacuation of the hydrogen cloud from the building only by changing the location of the outlets and their direction, especially around the openings, as seen in this figure.

Another result from the simulation of this scenario is the limited area of effect of these outlets. The outlets are only effective in the surrounding area because they are concentrated in a small area compared to the cloud. Once the cloud disperses out of this area, the hydrogen–air mixture can no longer be evacuated from the building with the same effectiveness. Similar to the previously discussed scenario, the location of the openings plays a crucial role in their effectiveness and, hence, in the hydrogen evacuation rates.

3.6. Scenario 5: Leakage with Obstacles

The simulation of the spread of hydrogen cloud when leaking from point P2 shown in Figure 1 shows different behavior of hydrogen. The hydrogen gas rises towards the ceiling, driven by the buoyancy force. After impinging with the intermediate floor between the ground and the top roof of the building, the cloud is split into two volumes: One continues rising towards the top ceiling, and the other remains trapped below the intermediate floor. Consequently, the trapped volume of hydrogen accumulates rapidly under the intermediate floor. Figure 6 shows the contour surfaces for hydrogen concentration below the top ceiling and the intermediate floor. After the hydrogen leakage from the pipe stops, i.e., after 10 s, the trapped hydrogen cloud starts to leak slowly towards the top roof. This happens due to the buoyancy force and the diffusion from high to low hydrogen concentrations that continuously forces hydrogen to flow around the edge of the intermediate floor in the upward direction. This eventually leads to the depletion of the combustible hydrogen cloud from below the floor at a much slower rate than in other scenarios.

3.7. Hydrogen Concentrations

To quantitatively assess the hydrogen concentration change in different locations in the building, its change over time is monitored at the measuring locations shown in Figure 3. Figure 7a shows hydrogen concentrations in the case of applying mechanical ventilation, i.e., Scenario 3-a, in solid lines, compared to the no-ventilation scenario, i.e., Scenario 1, in dashed lines. The same colors in this figure represent the same measuring point. The hydrogen cloud behaves almost the same for about 15 s by comparing the mechanical ventilation and no-ventilation scenarios. After this period, hydrogen concentrations tend to drop in the case of mechanical ventilation, while the concentrations increase in the case of no ventilation. Such behavior is expected since the hydrogen cloud takes a few seconds to reach the exhaust outlets near the ceiling and flow outside the building. The continuous hydrogen leakage from the pipe and the lack of an exhaust opening from the building in Scenario 1 lead to a constant build-up of hydrogen inside the building.

On the other hand, Figure 7b compares Scenario 3-a in solid lines with the natural ventilation scenario, i.e., Scenario 2 in dashed lines. The figure shows a significant advantage of natural ventilation over forced ventilation in the rapid discharge of hydrogen outside the building. This occurs due to the buoyancy forces driving hydrogen toward the ceiling and through the openings. Also, the large area of the ceiling openings plays an essential role in facilitating hydrogen evacuation through them.

3.8. Hydrogen Combustible Cloud

Another method to compare the effectiveness of ventilation systems is by comparing the combustible hydrogen cloud volumes in different scenarios, as shown in Figure 8. The volume of the combustible cloud is calculated by summing the volumes of all grid cells in which the volume fraction of hydrogen is between the lower and the LFL and HFL values, i.e., 4% and 75% respectively.

In Scenario 1, the size of the hydrogen cloud continuously increases since there is no outlet for hydrogen to flow outside the building. Furthermore, Scenario 3-a is the least effective among the scenarios where the leakage occurs at point P1, even after increasing the exhaust rate in Scenario 3-b to five times the maximum fan capacity, i.e., 25 ACH.

This occurs due to the exhaust outlets’ location, about 2.5 m below the ceiling, as explained in Section 3.4. Therefore, changing the location of the outlets to be on the ceiling, as in Scenario 4, significantly improves the effectiveness of the exhaust system despite it having the same flow rate as in Scenario 3-a. Also, locating exhaust outlets on the ceiling in Scenario 4 reduces the peak volume of the cloud by almost one third compared to Scenario 3-a. On the other hand, the effectiveness of the ceiling ventilation drops after about 20 s after the start of the leakage. This drop occurs because the ceiling outlets are only effective in the area close to these outlets. As the cloud disperses inside the building, the hydrogen cloud flows far from the outlets and evacuates almost pure air from the building instead of a hydrogen-rich mixture.

Scenario 5, where the leakage point is assumed to occur at P2 in Figure 1 under the intermediate floor, shows a different behavior than the other scenarios. First, the peak cloud volume is reached after about 10 s from the start of the leakage; however, it is lower than the previous scenarios illustrated. After reaching the peak volume, the volume of the combustible hydrogen cloud starts to decrease at a lower rate compared to other scenarios until it vanishes completely after about 82 s.

This scenario shows that the presence of an obstacle between the leakage point and the exhaust outlets hinders the free motion of hydrogen and traps the cloud in other pockets, depending on the geometry of the obstacle. In the studied scenario, the intermediate floor redistributes the combustible cloud volume over a longer time. This means a lower peak volume with a longer time of the flammable cloud inside the building, which enhances the hydrogen combustion risk.

4. Conclusions

This work shows a numerical simulation study of hydrogen cloud dispersion inside a large-scale building not originally designed to contain hydrogen pipes and equipment. Realistic hydrogen flow and leakage parameters are considered in this study to investigate the concentration change with time and the effect of different ventilation strategies. This study considered five different scenarios: no ventilation, natural ventilation, forced ventilation from side openings, and forced ventilation from ceiling openings.

Under the explained leakage parameters and the assumptions considered in this study, the main driving force for the hydrogen cloud is the buoyancy force. Therefore, natural ventilation is more effective than forced ventilation from side openings. However, placing the ventilation outlets on the ceiling of the space leads to the effective evacuation of hydrogen, which almost matches the effectiveness of natural ventilation. Additionally, once the hydrogen cloud moves away from the effective area of the outlet, the hydrogen keeps diffusing away to the rest of the building. In the case of hydrogen leakage located below other structures inside the building, the hydrogen cloud was partially trapped below these structures, which led to a redistribution of the combustible cloud in different locations and, accordingly, increased the risk of combustion.

From these simulation results, it is concluded that the ventilation openings should be located as close as possible to the ceiling of the area where hydrogen is expected to leak. These exhaust openings should also be placed above the potential hydrogen leakage locations and at other locations to ensure enough area in the ceiling to evacuate hydrogen effectively. Another conclusion from this work is to design buildings that contain hydrogen sources with no obstacles between the potential hydrogen leakage locations and the ventilation outlets. Such a design should lead to avoiding combustible hydrogen cloud redistribution and forming multiple combustible pockets below these obstacles. Finally, the simulations show the importance of studying the effectiveness of the exhaust system rather than focusing only on the exhaust flow rate.

Many future studies can be based on the current research to cover conditions and parameters not included in this study. As mentioned in Section 2.4.3, wind speeds and atmospheric temperature outside the building are not considered here. Therefore, future studies should consider wind speeds and the induced airflow due to different temperatures, wind speeds, and wind directions. Another potential future work is to analyze other hydrogen mitigation techniques, such as passive auto-catalytic recombiners (PARs), on hydrogen concentrations. Also, future work could optimize the locations of sensors that detect hydrogen after a delay is short enough to avoid any combustible hydrogen levels in the building.