Characteristics of Hydrogen Leakage and Dissipation from Storage Tanks in an Integrated Hydrogen Production and Refueling Station

Abstract

Hydrogen, as a renewable and clean energy carrier, has the potential to play an important role in carbon reduction. Crucial to achieving this is the ability to produce clean sources of hydrogen and to store hydrogen safely. With the rapid development of the hydrogen industry, the number of hydrogen refueling stations (HRS) is increasing. However, hydrogen safety at HRS is of great concern due to the high risk of hydrogen leakage during storage. This study focused on an integrated hydrogen production and refueling station (IHPRS) in Weifang, China, and numerically simulated a hydrogen leakage accident in its storage area. The effects of the leakage aperture, the leakage direction and the ambient wind direction and speed on the leakage and dissipation characteristics of hydrogen were investigated. The results showed that the volume, mass and dissipation time of the flammable hydrogen cloud (FHC) increased with an increase in the leakage aperture. The installation of a canopy or densely packed equipment near the hydrogen storage area will seriously hinder the dissipation of the FHC. Ambient winds in the opposite direction of the leakage may cause high-concentration hydrogen to accumulate near the hydrogen storage tanks and be difficult to dissipate, seriously threatening the safety of the integrated station.

1. Introduction

In recent years, people have been facing increasingly serious environmental pollution problems and energy crises, and the call for using clean and renewable energy to replace traditional fossil energy sources is becoming louder [1,2]. Hydrogen, as a new type of energy carrier, possesses the merits of high efficiency, low carbon emissions and wide distribution in nature, which paves a feasible path for energy transition [3,4,5]. As a result, hydrogen energy is gradually receiving close attention from countries around the world, and its areas of application are being developed more and more widely [6,7]. Benefiting from continuous breakthroughs in hydrogen production, storage, and transportation technologies, the market scale of the hydrogen fuel cell vehicle industry has expanded [8,9]. Against this background, the number of hydrogen production stations and HRS has been increasing rapidly, and the primary type of hydrogen storage in the stations is currently high-pressure gaseous hydrogen storage [10,11,12]. However, due to compact storage space and high storage pressures, ensuring the safety of the storage and transportation of hydrogen is very challenging [13,14]. Hydrogen is characterized by easy leakage, the hydrogen embrittlement reaction of metals, low ignition energy and a wide combustion range [15,16,17]. Once a leakage accident occurs, the FHC formed by the leakage may be ignited, and it may lead to fires and explosions, which seriously threaten the personnel in the station and the inner building of the station [18]. Therefore, to ensure the safe and stable operation of a hydrogen production and refueling station, it is necessary to study the process of accidental hydrogen leakage and to analyze the characteristics of the leakage and diffusion of hydrogen in the station.

Some scholars have carried out experimental studies on the leakage and diffusion of hydrogen. Kobayashi et al. [19] conducted cryogenic compressed hydrogen leakage diffusion experiments, which showed that both the leakage flow rate and concentration of hydrogen increased with a decreasing supply temperature. An experimental study on hydrogen leakage at different initial pressures, nozzle diameters and ignition positions was carried out [20]. The hysteresis parameters and flame propagation characteristics of hydrogen leakage were quantitatively analyzed, and a prediction model for the hysteresis parameters of hydrogen leakage was developed on the basis of the van der Waals equation. Xin et al. [21] focused on the leakage behavior of an underground parking facility and built a scale-down model, and they obtained various data on different environmental conditions through experiments. Xu et al. [22] experimentally studied the hydrogen leakage and diffusion characteristics in a space with a large aspect ratio, and the results showed that the higher the initial leakage rate, the stronger the initial intensity of turbulence, which is more favorable to the mixing of hydrogen and air. Shu et al. [23] created a high-accuracy model of hydrogen leakage so that they obtained experimental data with a relatively low bias. Tanaka et al. [24] performed a hydrogen leaking test in a storage room. The results showed that the leakage diameter, the amount of hydrogen released and the indoor ventilation characteristics had a significant effect on the hydrogen concentration.

A number of scholars have focused on computational methods for analyzing the behavior and characteristics of hydrogen leakage. Vanlaere et al. [25] mainly investigated the distribution characteristics of hydrogen after it leaked in a confined space by means of a nilpotent analysis method. They proposed a risk reduction strategy with practical applications. Chang et al. [26] analyzed potential leakage accidents in a hydrogen production facility based on the Dynamic Bayesian Network method. They successfully predicted the probability of the system collapse and the logic of the accidents. He et al. [27] creatively utilized the ConvLSTM-based surrogate model to predict leakage in HRS accurately, which spent computing power more evenly on calculation. Rostamzadeh et al. [28] created a new method named MACB to apply to the convergence problem to calculations of hydrogen leakage so that they optimized the traditional scheme and sped up calculation in specific situations.

Many scholars have utilized the computational fluid dynamics (CFD) method to study hydrogen leakage numerically. Choi et al. [29] simulated the hydrogen leakage diffusion process of fuel cell vehicles in an underground car park. They found that the volume of the flammable region did not increase linearly in the initial stage but increased rapidly after a latent period. Shentsov et al. [30] modeled the release and dispersion of a high-pressure hydrogen storage tank in a parking lot by CFD, investigated the effects of the release angle, canopy height and ventilation rate on the hydrogen diffusion, and proposed a safety strategy based on the results. Wang et al. [31] used FLACS software to simulate hydrogen leakage in a confined room and considered various restrictions, including the side walls and corners. Patel et al. [32] simulated the stratification characteristics of hydrogen after leakage in a semi-enclosed space with ANSYS FLUENT. They investigated the effect of the arrangement of the number and location of the vents on the reduction in the hydrogen concentration to derive an optimal ventilation scheme. Malakhov et al. [33] used STAR CCM+ to simulate the distribution of hydrogen leakage in a semi-enclosed space with different initial leakage pressures and different leak opening sizes. They investigated ways to improve the efficiency of forced ventilation to guide hydrogen safety issues. Tian et al. [34] simulated the pattern of leakage from a high-pressure hydrogen storage tank in a variety of scenarios using CFD methods. They analyzed the safety distances as well as the variation of the hydrogen concentration in the overall space. Thomas et al. [35] studied hydrogen leakage accidents using numerical simulations and found that larger leakage rates were more likely to form an FHC in open space. Kikukawa et al. [36] simulated the leakage of a hydrogen dispenser in an HRS using FLUENT software (version 6.2). They verified the possibility of constructing an HRS near a gas station by the distance of the diffusion of hydrogen. Qian et al. [37] analyzed diverse scenarios, including various locations of leakage in hydrogen storage tanks and wind effects in an HRS. They found that the presence of obstacles greatly affected the shape and diffusion distance of the FHC, and the closer the leakage location was to the obstacle, the larger the contour of the FHC and the more irregular the shape. Patel et al. [38] conducted a parametric study of the diffusion and explosion of hydrogen in an HRS by means of FLACS software, analyzed the range of acceptable safety distances for various scenarios and assessed the risk of accidental leakage and explosion in an HRS. Wang et al. [39] modeled a hydrogen leakage accident from a heavy truck in an HRS with a large canopy structure by CFD methods. They performed a deterministic assessment of the accident risk. Gao et al. [40] conducted a leakage model based on hydrogen storage tanks in a nuclear station. They obtained diffusion contours on account of a variety of factors. Han et al. [41] investigated the diffusion law of hydrogen leakage for leakage holes with different diameters and different leakage pressures by numerical simulation and explored the evolution of the FHC. Xiao et al. [42] studied the leakage and diffusion behavior of hydrogen using CFD methods and predicted the diffusion distance of the FHC via artificial neural networks. Cui et al. [43] concentrated on the HRS in Ningbo Harbor. The impact of wind, the roof shape and the air humidity on the diffusion of hydrogen leakage was investigated. They found that headwinds significantly increased the volume of the FHC, sloping roofs could promote the diffusion of hydrogen, and the air humidity had a negligible effect.

The above studies focused on the diffusion characteristics of high-pressure hydrogen during the leakage process. However, there are fewer studies on the dissipation characteristics of hydrogen during the dissipation process when the leakage is over. In addition, most of the studies on hydrogen leakage are based on scenarios with small spaces or simplified building layouts. However, there are fewer studies on actual hydrogen-related scenarios with large spaces and complex building layouts. In this study, the leakage characteristics of high-pressure hydrogen in an integrated station were investigated, as well as the dissipation characteristics of hydrogen after the leakage stopped. Furthermore, the purpose of this study was to guide the construction of an IHPRS in Weifang, China, which is of great practical significance. The integrated station is characterized by a large space and a complex building layout, and the factors influencing the hydrogen leakage and dissipation in this scenario were investigated, including the leakage aperture, the leakage direction, and the ambient wind direction and speed.

2. Numerical Modeling of Hydrogen Leakage and Dissipation

2.1. Mathematical Model

2.1.1. Governing Equations

The process of hydrogen leakage and dissipation in an IHPRS needs to satisfy the continuity equation, the momentum conservation equation and the energy conservation equation [3].

The continuity equation is expressed as

$${\displaystyle \frac{\partial \rho }{\partial t}}+{\displaystyle \frac{\partial \rho u}{\partial x}}+{\displaystyle \frac{\partial \rho v}{\partial y}}+{\displaystyle \frac{\partial \rho w}{\partial z}}=0$$

(1)

where $\rho$ is the density of the gas (kg/m³); $t$ is the time (s); and $u,v,w$ are the components of the velocity in the $x,y,z$ directions (m/s), respectively.

The momentum conservation equation is expressed as follows

$$\left\{\begin{array}{l}{\displaystyle \frac{\partial \rho u}{\partial t}}+\nabla \cdot (uu)=-{\displaystyle \frac{\partial p}{\partial x}}+{\displaystyle \frac{\partial {\tau }_{xx}}{\partial x}}+{\displaystyle \frac{\partial {\tau }_{yx}}{\partial y}}+{\displaystyle \frac{\partial {\tau }_{zx}}{\partial z}}+F_x\\ {\displaystyle \frac{\partial \rho v}{\partial t}}+\nabla \cdot (vu)=-{\displaystyle \frac{\partial p}{\partial y}}+{\displaystyle \frac{\partial {\tau }_{xy}}{\partial x}}+{\displaystyle \frac{\partial {\tau }_{yy}}{\partial y}}+{\displaystyle \frac{\partial {\tau }_{zy}}{\partial z}}+F_y\\ {\displaystyle \frac{\partial \rho w}{\partial t}}+\nabla \cdot (wu)=-{\displaystyle \frac{\partial p}{\partial z}}+{\displaystyle \frac{\partial {\tau }_{xz}}{\partial x}}+{\displaystyle \frac{\partial {\tau }_{yz}}{\partial y}}+{\displaystyle \frac{\partial {\tau }_{zz}}{\partial z}}+F_z\end{array}\right.$$

(2)

where $u$ is the sum of the velocity components in the $x,y,z$ directions (m/s); $p$ is the pressure (Pa); $\tau$ is the viscous stress (Pa); and $F_x,F_y,F_z$ are the components of the volume force F in the $x,y,z$ directions (N), respectively.

The energy conservation equation is expressed as

$${\displaystyle \frac{\partial \left(\rho c_{p}T\right)}{\partial t}}+\nabla \cdot \left(\rho c_{p}uT\right)={\displaystyle \frac{\partial }{\partial x}}\left(k{\displaystyle \frac{\partial T}{\partial x}}\right)+{\displaystyle \frac{\partial }{\partial y}}\left(k{\displaystyle \frac{\partial T}{\partial y}}\right)+{\displaystyle \frac{\partial }{\partial z}}\left(k{\displaystyle \frac{\partial T}{\partial z}}\right)+\varphi +S_h$$

(3)

where $c_p$ is the constant-pressure specific heat capacity (J/(kg·K)), $T$ is the temperature (K), $k$ is the thermal conductivity (W/(m·K)), $\varphi$ is the dissipation function and $S_h$ is the heat source within the fluid (W/m³).

The hydrogen is mixed with air after leakage, and the expression for the component transportation equation is [43]

$${\displaystyle \frac{\partial \left(\rho c_s\right)}{\partial t}}+\nabla \cdot \left(\rho c_{s}u\right)={\displaystyle \frac{\partial }{\partial x}}\left(D_s{\displaystyle \frac{\partial \left(\rho c_s\right)}{\partial x}}\right)+{\displaystyle \frac{\partial }{\partial y}}\left(D_s{\displaystyle \frac{\partial \left(\rho c_s\right)}{\partial y}}\right)+{\displaystyle \frac{\partial }{\partial z}}\left(D_s{\displaystyle \frac{\partial \left(\rho c_s\right)}{\partial z}}\right)+R_s$$

(4)

where $c_s$ is the volume fraction of component $s$, $D_s$ is the diffusion coefficient of component $s$ in air (m²/s) and $R_s$ is the production rate of component $s$.

2.1.2. Turbulence Model

In the process of the leakage and dissipation of high-pressure hydrogen, the flow situation is very complicated. The realized k–ε turbulence model is widely used in CFD to simulate gas diffusion problems due to its high accuracy in the simulation of complex turbulence problems [41,44]. Therefore, to better simulate the leakage and dissipation process, the realized k–ε turbulence model was used in this study. The transport equations of turbulent kinetic energy $k$ and dissipation rate $\epsilon$ are expressed as follows [45]

$${\displaystyle \frac{\partial }{\partial t}}(\rho k)+\nabla \left(\rho ku\right)=\nabla \left[\left(\mu +{\displaystyle \frac{{\mu }_t}{{\sigma }_k}}\right)\nabla k\right]+G_k+G_b-\rho \epsilon -Y_M+S_k$$

(5)

$${\displaystyle \frac{\partial }{\partial t}}(\rho \epsilon )+\nabla \left(\rho \epsilon u\right)=\nabla \left[\left(\mu +{\displaystyle \frac{{\mu }_t}{{\sigma }_{\epsilon }}}\right)\nabla \epsilon \right]+\rho C_1{S}_{\epsilon }-\rho C_2{\displaystyle \frac{{\epsilon }^2}{k+\sqrt{v\epsilon }}}+C_{1\epsilon }{\displaystyle \frac{\epsilon }{k}}{C}_{3\epsilon }{G}_b+S_{\epsilon }$$

(6)

where $G_k$ and $G_b$ represent the turbulent kinetic energy due to the velocity gradient and buoyancy (J), respectively; ${\mu }_t$ is the turbulent viscosity (kg/(m·s)); ${\sigma }_k$ and ${\sigma }_{\epsilon }$ are the turbulent Prandtl numbers for the $k$ and $\epsilon$ equations, respectively; $Y_M$ denotes the effect of turbulent fluctuation on $\epsilon$ in a compressible fluid; $C_1$ is the empirical coefficient, $C_1=\max [0.43,\eta /\eta +5]$, $\eta =Sk/\epsilon$, $S=\sqrt{2S_{ij}{S}_{ij}}$; $C_2$ is a constant; $v$ is the correction factor; $C_{1\epsilon }$ and $C_{3\epsilon }$ are constants describing the effect of buoyancy on $\epsilon$; and $S_k$ and $S_E$ are user-defined source terms. ${\sigma }_k$, ${\sigma }_{\epsilon }$, $C_2$, $C_{1\epsilon }$ and $C_{3\epsilon }$ are dimensionless constants with values of 1.0, 1.2, 1.9, 1.44 and 0.8, respectively, which are the default values for turbulence models in FLUENT.

2.1.3. Virtual Nozzle Model

The high-pressure, under-expanded supersonic gas ejected from the leakage port forms a more complex shock wave structure when it contacts the outside air, and there is a throttling phenomenon at the leakage port. Therefore, a large number of meshes were required to ensure the accuracy of the simulation results, and the computational cost was very high. In this study, the Molkov virtual nozzle model [46] was used to rationally simplify the boundary conditions at the leakage port to save calculation costs, which provided an equivalent inlet condition. The schematic of the Molkov virtual nozzle model is shown in Figure 1. The isentropically expanding gas is ejected from a high-pressure reservoir (Level 1), first through the nozzle orifice (Level 2) and later released at ambient pressure (Level 3) through the virtual nozzle. The calculation procedure of the Molkov virtual nozzle model is as follows.

According to the Abel–Nobel equation of state [47], the density of the hydrogen in the tank ${\rho }_1$ can be obtained as [37]

$${\rho }_1={\displaystyle \frac{p_1}{b{p}_1+R_{H_2}{T}_1}}$$

(7)

where b is the residual capacity factor, 7.691 × 10⁻³ (m³/kg), and $R_{H_2}$ is the gas constant for hydrogen, 4124.24 (J/(kg·K)).

The density at the orifice ${\rho }_2$ can be obtained by solving the transcendental equation of isentropic expansion [37]

$${\left({\displaystyle \frac{{\rho }_1}{\left(1-b{\rho }_1\right)}}\right)}^{\gamma -1}={\left({\displaystyle \frac{{\rho }_2}{\left(1-b{\rho }_2\right)}}\right)}^{\gamma -1}\left[1+{\displaystyle \frac{(\gamma -1)}{2{\left(1-b{\rho }_2\right)}^2}}\right]$$

(8)

where $\gamma$ is the specific adiabatic index of hydrogen gas, with a value of 1.4. The temperature at the orifice $T_2$ is solved by the following equation [46]:

$$T_1/T_2=1+(\gamma -1)/2{(1-b{\rho }_2)}^2$$

(9)

The pressure at the orifice $P_2$ can then be obtained from the Abel–Nobel equation of state.

The velocity of hydrogen at the orifice $V_2$ can be calculated from the sound velocity equation [46]:

$$V_2^2=\gamma R_{H_2}{T}_2/{(1-b{\rho }_2)}^2$$

(10)

Assuming that the velocity of hydrogen at the virtual nozzle $V_3$ is equal to the local speed of sound, $V_3$ can be expressed as [46]:

$$V_3^2=\gamma R_{H_2}{T}_3$$

(11)

Then, the temperature at the virtual nozzle $T_3$ can be obtained by solving the equation of conservation of energy per unit of mass, which is solved as follows [46]:

$$c_p{T}_2+V_2^2/2=c_p{T}_3+V_3^2/2$$

(12)

$$T_3={\displaystyle \frac{2T_2}{(\gamma +1)}}+{\displaystyle \frac{(\gamma -1)}{(\gamma +1)}}·{\displaystyle \frac{P_2}{{\rho }_2\left(1-b{\rho }_2\right)R_{H2}}}$$

(13)

Finally, the virtual nozzle diameter $d_3$ can be obtained by solving the continuity equation [46]

$$d_3=d_2{\left({\rho }_2{V}_2/{\rho }_3{V}_3\right)}^{1/2}$$

(14)

where the hydrogen density at the notional nozzle ${\rho }_3$ can be calculated as $P_3$ being equal to the ambient pressure as well as the velocity of the gas $V_3$.

2.2. Numerical Modeling and Boundary Conditions

In the study, ANSYS ICEMCFD [48] was used to model the IHPRS geometrically and to discretize the computational region. The computational region was a rectangular area containing all the buildings and equipment in the station, with a length of 219 m, a width of 181 m and a height of 50 m. Furthermore, buildings and equipment within the computational region were considered solid regions and were not included in the computational region, i.e., the computational region contained only fluids. The overall layout of the IHPRS and the details of specific areas are shown in Figure 2.

There were two groups of hydrogen storage tanks in the hydrogen storage area, each with 12 tanks arranged in a 3 × 4 array. The detailed parameters of the hydrogen storage tanks are shown in

Table 1. Detailed parameters of the hydrogen storage tanks.

Property	Value	Unit
Hydrogen storage tank type	III	/
Operating pressure	20	MPa
Outer diameter	0.559	m
Thickness of the tank wall	0.0184	m
Length	11.58	m
Volume	2.36	m3

. The geometry of the hydrogen storage tank was simplified to a hexahedral structure to facilitate modeling. The compressor area was located next to the hydrogen storage area, which contained five hydrogen compressors and a canopy. The buffer tank area was to the west of the hydrogen storage area, which contained five buffer tanks with an external diameter of 4.4 m, a height of 3.5 m and an operating pressure of 1.6 MPa. The dimensions in the figure correspond to the height of the building.

The energy model, turbulence model and component transport model in ANSYS FLUENT (version 2020R2) were selected to establish the numerical model of hydrogen leakage. The mixture of hydrogen and air was set up as an incompressible ideal gas. No-slip boundary conditions were used for both the building walls and the ground. The boundary condition at the leakage port was set as the mass flow rate inlet. The boundary of the fluid region except the ground was set as the pressure outlet, and the windy side boundary was set as the velocity inlet. Moreover, the initial ambient temperature and ambient pressure were set to 293 K and 101,325 Pa, respectively; the effect of gravity was considered, and the gravitational acceleration was set to 9.81 m/s².

In ANSYS FLUENT software, the pressure-based transient solver was selected to solve the numerical model, and the semi-implicit separation algorithm SIMPLE [49] was selected as the solution method. The discrete format of the convective terms was set as a second-order upwind format to ensure the stability and accuracy of the solution. The second-order implicit format was selected as the discrete format for the time term.

Due to the complexity of the leakage process of high-pressure hydrogen, simplifying assumptions needed to be made for the numerical simulation model. In this study, it was assumed that the leakage aperture and leakage pressure remained constant during the leakage process; the external environmental factors, such as wind speed, temperature, humidity, etc., remained constant, and the mixture of hydrogen and air was regarded as an ideal gas.

2.3. Model Validation and Mesh Independence Analysis

2.3.1. CFD Model Validation

To ensure the accuracy of the model, this study compared the experimental data of hydrogen leakage from the National Institute of Standards and Technology (NIST) [50] with the simulation results of the CFD numerical model established in this study. The experimental scenario was a garage with a length and width of 6.10 m and a height of 3.05 m, modeled as shown in Figure 3a. The hydrogen was released by a dispenser located in the center of the garage with a constant leakage flow rate and a volumetric flow rate of 994 L/min. There were four sensors in the garage for measuring the hydrogen concentration, and these four monitoring points had the same horizontal location coordinates, except for the height, which were 1.14 m, 1.52 m, 1.90 m and 2.59 m. The four monitoring points were labeled M1, M2, M3 and M4 from low to high, as shown in Figure 3a. In this validation model, the basic settings of the simulation remained the same as those described in the previous section. A comparison of the simulation curves of the hydrogen volume fraction with time at each monitoring point with the experimental data is shown in Figure 3b. As can be seen from the figure, the simulated values were slightly lower than the experimental values in the early stages but converged with the leakage time.

Table 2. Simulated and experimental results and their relative errors at the end of the leakage.

Monitoring Point	Experimental Value	Simulated Value	Relative Error (%)
M1	0.07311	0.07396	1.163
M2	0.08203	0.08392	2.304
M3	0.08658	0.08845	2.160
M4	0.09174	0.09205	0.3379

shows the simulated and experimental results and their relative errors at the end of the leakage. It can be seen that the simulated results were slightly higher than the experimental results, but the relative errors of the simulated results ranged from 0.3379% to 2.304% for the different monitoring points. Overall, the simulation results were in good agreement with the experimental data.

2.3.2. Mesh Independence Analysis

In this study, the hydrogen leakage and dissipation model of the IHPRS was meshed by using ANSYS ICEMCFD, as shown in Figure 4a. The structured mesh was used for all areas, and the mesh was refined for the core area of the leakage near the hydrogen storage tank groups, as well as at the leakage ports, as shown in Figure 4b,c. To verify the independence of the meshes, a total of five sets of meshes were plotted, with quantities of 3.57, 4.25, 5.18, 6.19, and 6.94 million, respectively. Three monitoring points (M_1, M_2 and M_3) were set up with the coordinates (114.8, 1.5, 16), (112, 2, 4), and (118, 6, −6), respectively, as shown in Figure 4. The independence of the meshes was verified by the variation of the hydrogen volume fraction with time at the monitoring points and by the volume of the FHC at the end of the leakage. Figure 5a shows the variation of the volume fraction of hydrogen with time at the three monitoring points for different meshes, and it can be seen that the simulation results were very close when the number of meshes increased to 6.19 million. Figure 5b shows the volume of the FHC at the end of the leakage with different meshes. After the number of meshes reached 6.19 million, the results of the volume of the FHC were very close to each other, and there was only a difference of 0.78 m³ from the results of the mesh with 6.94 million meshes. Therefore, to ensure the accuracy of the simulation results and, at the same time, reduce the computational cost, a mesh quantity of 6.19 million was selected for the subsequent numerical simulation. The computation time for each case was about 50 to 60 h.

3. Results and Discussion

3.1. Effect of Leakage Aperture on Hydrogen Leakage and Dissipation Behavior

3.1.1. Leakage Processes at Different Leakage Apertures

The leakage aperture is an important factor affecting the process of high-pressure hydrogen leakage. To study the effect of different leakage apertures on the hydrogen leakage process, the hydrogen leakage process was simulated for leakage apertures of 8 mm, 12 mm, 16 mm, 20 mm and 24 mm. In this set of simulations, the ambient wind speed was 0 m/s, and the leakage direction was toward the office buildings and plant buildings. For different leakage apertures, the hydrogen leakage flow rate was different. Therefore, the time required for all the hydrogen in the tank to leak was different. The boundary conditions at the leakage aperture are shown in

Table 3. Boundary conditions at the leakage port for different leakage apertures.

Leakage Aperture	Simplified Leakage Port Boundary Conditions for Virtual Nozzles	Leakage Duration
8 mm	$u_2$ = 1187.6 m/s$; d_2$ = 78.6 mm; T = 244.3 K	59.80 s
12 mm	$u_2$ = 1187.6 m/s$; d_2$ = 117.9 mm; T = 244.3 K	26.56 s
16 mm	$u_2$ = 1187.6 m/s$; d_2$ = 157.2 mm; T = 244.3 K	14.92 s
20 mm	$u_2$ = 1187.6 m/s$; d_2$ = 196.5 mm; T = 244.3 K	9.56 s
24 mm	$u_2$ = 1187.6 m/s$; d_2$ = 235.8 mm; T = 244.3 K	6.64 s

. To observe the leakage behavior of hydrogen, the FHC was selected to be displayed as cloud diagrams. The FHC is a hydrogen-air mixture with a hydrogen volume fraction between 0.04–0.75.

The process of hydrogen leakage under different leakage apertures is shown in Figure 6. In the early stage of the leakage, the flow rate at the leakage aperture was very large, and the high-pressure hydrogen was ejected to the front and formed an FHC. At 1 s, there was already a gap among the FHC of different leakage apertures in terms of the volume and diffusion distance, and this gap increased as the leakage proceeded. At 4.5 s, the FHC with a large leakage aperture (16 mm, 20 mm, 24 mm) was significantly larger than that with a small leakage aperture (8 mm, 12 mm) in volume. It had already diffused to the far enclosure. At this time, due to the resistance of the air and the enclosure, the speed of the head of the FHC decreased, and the buoyancy effect gradually dominated. The tendency of the FHC to move in the vertical direction became obvious, and a part of the FHC continued to diffuse forward over the enclosure. At the same time, the FHC at the smaller leakage aperture was still some distance away from the enclosure wall. At the moment of their respective leakage termination, the FHC at different leakage apertures differed significantly in their shapes. When the leakage aperture was 8 mm, the FHC reached the enclosure at the end of the leakage, but it did not diffuse along the enclosure walls. The FHC was in the shape of a long strip. As the leakage aperture increased, the FHC could diffuse to the wall at the end of the leakage process and diffused along the enclosure to the sides and over the wall to the back and upwards. In addition, the larger the leakage aperture, the larger the volume of the FHC that gathered near the enclosure at the end of the leakage process.

The diffusion distances of the FHC in the x, y and z directions at the end of the leakage with different apertures are shown in

Table 4. Diffusion distances of the FHC at the end of leakage under different leakage apertures.

Leakage Aperture (mm)	Diffusion Distance in Each Direction (m)
	X	Y	Z
8	4.23	6.61	39.15
12	31.56	15.31	53.36
16	58.72	21.37	62.17
20	57.26	19.18	58.56
24	51.27	17.72	55.70

. The x and z directions are the horizontal directions, and y is the vertical direction. It can be seen that with an increase in the leakage aperture, the diffusion distance of the FHC at the end of the leakage showed a tendency to increase first and then decrease. When the leakage aperture was 8 mm, although the leakage time was the longest, the diffusion distance of the FHC at the end of the leakage was the smallest in all three directions, especially in the x and y directions. When the leakage aperture was 16 mm, the diffusion distance of the FHC was the largest in all three directions. Further increasing the leakage aperture made the leakage time shorter, and the FHC was not yet sufficiently diffused in all directions, which led to a decrease in the diffusion distance at the end of the leakage.

3.1.2. Dissipation Processes at Different Leakage Apertures

After the leakage stopped, although no hydrogen continues to be ejected from the hydrogen storage tank, the hydrogen that has been leaked out will continue to diffuse in the air until it was completely dissipated. The FHC still posed a large potential risk to the nearby buildings and people during this time. Therefore, it was necessary to study the dissipation process of the FHC after the leakage stopped. To compare and analyze the dissipation process of the FHC under different conditions, the end moment of the hydrogen leakage process under different conditions was defined as the beginning moment of the dissipation process.

The dissipation process of the FHC under different leakage apertures is shown in Figure 7. It can be seen that the FHC kept its previous movement trend and continued to diffuse. With the hydrogen continuously mixing with the surrounding air, the FHC gradually dissipated. When the dissipation time was 3 s, the FHC was basically distributed in the vicinity of the enclosure. The FHC with a smaller leakage aperture dissipated faster, and the diffusion distance did not increase. However, the FHC with a larger leakage aperture dissipated more slowly, and the diffusion distance increased further. When the dissipation time was 6 s, the FHC under the 8 mm leakage aperture had completely dissipated, the FHC under the 12 mm leakage aperture only had a small amount attached to the enclosure, the FHC under the 16 mm leakage aperture still existed in the vicinity of the enclosure wall and the airspace above the plant in the form of a cluster. The distribution of the FHC at 20 mm and 24 mm leakage apertures was similar to that at 16 mm. However, the diffusion distance still increased further, and the residual FHC was bigger.

To further quantify the effect of the leakage aperture on the hydrogen dissipation process, the volume and mass of the FHC were calculated in this study using the user-defined function (UDF) in FLUENT.

The variation of the volume and mass of the FHC with dissipation time at different leakage apertures is shown in Figure 8. It can be seen that both the volume and mass of the FHC at the end of the leakage increased with the increase of the leakage aperture. From Figure 8a, it can be seen that the volume of the FHC increased and then decreased with dissipation time for 20 mm and 24 mm leakage apertures. In contrast, the volume of the FHC kept decreasing when the leakage aperture was less than or equal to 16 mm. This indicates that when the leakage aperture was large, the influence range of the FHC continued to increase even if the leakage had stopped. From Figure 8b, it can be seen that the mass of the FHC at each leakage aperture continued to decrease after the leakage stopped. The reduction in the mass of the FHC gradually slowed down due to the lower kinetic energy of the FHC in the later stages of the dissipation process and the decrease in the mixing rate of the hydrogen with the surrounding air.

Throughout the dissipation process, the dissipation time of the FHC increased with the increase of the leakage aperture, which implies a longer duration of the potential hazard. At the same time, the peak values of the volume and mass of the FHC also increased with the increase of the leakage aperture, which meant that the peak value of the potential hazard range and the peak value of the hazard degree were larger. The leakage was worst when the leakage aperture was 24 mm, and the peak values of the volume and mass of the FHC reached 4556.13 m³ and 28.05 kg, respectively. Furthermore, the plant buildings and office building areas were densely populated and contained electrical circuits and appliances, etc., which present a certain risk of ignition. If the FHC were to ignite in this area, it would cause serious injury to people.

3.2. Effect of Leakage Direction on Hydrogen Leakage and Dissipation Behavior

3.2.1. Leakage Processes at Different Leakage Directions

In the event of a leakage accident, the buildings around the leakage vent will have a direct impact on the hydrogen leakage behavior. For different leakage directions, the buildings contacted with the high-pressure hydrogen after it is ejected from the leakage vent are also different, and the hydrogen cloud will be subject to different obstruction effects. To study the characteristics of hydrogen leakage and dissipation in an integrated station more comprehensively, six representative leakage directions were considered, as shown in

Table 5. Cases for each direction of hydrogen leakage.

Cases	Leakage Directions
Case 1	Leakage toward the plant buildings and office building
Case 2	Leakage toward the HRS building
Case 3	Leakage toward the buffer tank and hydrogen production workshop
Case 4	Leakage toward the compressor area
Case 5	Leakage toward the adjacent hydrogen storage tank group
Case 6	Leakage toward the ground

. In this set of simulations, the ambient wind speed was 0 m/s, and the leakage aperture was 20 mm. The process of hydrogen leakage under different leakage directions is shown in Figure 9.

In Case 1, the leakage direction was toward the plant buildings and office building. Due to the openness of the leakage area, the hydrogen was not hindered at the beginning of the leakage, and then it moved to the far enclosure. It was hindered by the wall surface and diffused to both sides and beyond the enclosure wall. In Case 2, the leakage direction was toward the HRS building. The hydrogen reached the HRS building at the early stage of the leakage and diffused in all directions due to the obstruction of its walls. At the end of the leakage, the flammable hydrogen formed a fan-shaped cloud with thicker edges and diffused to the open areas on the left and upper sides of the station building as well as to the adjacent blast wall. In Case 3, the leakage direction was toward the buffer tank and hydrogen production workshop. The leaked hydrogen first moved to the buffer tank, bypassed it, converged behind it and then continued forward to the hydrogen production workshop, where it diffused in all directions after being obstructed by the workshop walls. In Case 4, the leakage direction was toward the compressor area. The hydrogen was impeded by the compressor at the beginning of the leakage, and a portion of the FHC moved through the compressor area and onward to the electrical substation room. In contrast, the majority of the FHC was impeded by the compressor and canopy, and collected in the compressor area. In Case 5, the leakage direction was toward the adjacent hydrogen storage tank group. The hydrogen was impeded by the adjacent hydrogen storage tank group immediately after leakage, and most of the hydrogen moved through the gap of the tank group and to the compressor area and the electrical substation room. In Case 6, the leakage direction was toward the ground. The hydrogen was ejected to the ground and moved along the ground in all directions, diffusing into the HRS building, the buffer tanks and the compressor area, eventually forming a large FHC at the ground level centered on the hydrogen storage tanks.

Table 6. Diffusion distances of the FHC at the end of leakage under different leakage directions.

Cases	Diffusion Distance in Each Direction (m)
	X	Y	Z
Case 1	57.26	19.18	58.56
Case 2	70.90	22.54	18.49
Case 3	40.66	21.56	44.12
Case 4	51.96	11.97	34.63
Case 5	63.82	13.24	23.03
Case 6	73.46	12.06	43.10

shows the diffusion distances of the FHC in the x, y and z directions at the end of the leakage under different leakage directions. In Cases 2, 5 and 6, the diffusion distance was larger in the x direction. In Cases 1, 2 and 3, the diffusion distance in the y direction was larger, which indicated that wide planar obstacles had a significant hindering effect on the movement of hydrogen. The FHC diffused along the wall in all directions, although it lost more kinetic energy after encountering the obstruction of this obstacle head-on. In Cases 4 and 5, the diffusion distance in the y direction was smaller because the canopy inhibited the vertical movement of the FHC. In Case 3, the FHC, after being obstructed by the buffer tanks, bypassed the buffer tanks and continued to move to the hydrogen purification plant, which indicated that the obstacles in the curved structure had a small obstructive effect on the hydrogen jet and did not change the overall trend of hydrogen leakage. In Case 5, the hydrogen jet, after being obstructed by the adjacent hydrogen storage tank group, continued to move forward through the gap until it crossed the compressor area and reached the electrical substation room. The FHC at the end of the leakage diffused 63.82 m along the leakage direction. This indicated that the groups of obstacles with gaps also had a less obstructive effect on the hydrogen jet and its movement.

3.2.2. Dissipation Process at Different Leakage Directions

The dissipation process of the FHC under different leakage directions is shown in Figure 10. By observing the changes in the FHC in each case at 4 s of dissipation, it could be seen that the diffusion distance of the FHC in each direction continued to increase after the leakage stopped. By comparing the morphology of the FHC at 8.5 s, it could be seen that a more open and unconfined area was more beneficial for the dissipation of the FHC. The FHC in Cases 1, 2 and 3 were located in a relatively open area. Their dissipation speeds were significantly faster than those in Cases 4, 5 and 6, among which the FHC in Cases 1 and 2 had basically completely dissipated. On the other hand, in Cases 4, 5 and 6, plenty of the FHC remained in the compressor area, where the free diffusion of the FHC was hindered by the densely packed equipment and the canopy above, and their dissipation rate was obviously slower.

The variation of the volume and mass of the FHC with dissipation time at different leakage directions is shown in Figure 11. In Case 2, the volume and mass of the FHC decreased from the time the leakage stopped and completely dissipated in the shortest possible time, and the peaks in volume and mass were minimized. Therefore, the hazard duration, the peak hazard range and the peak hazard degree were all minimized in Case 2. Except for Case 2, the FHC under all other cases increased in volume after the leakage stopped, i.e., the hazard range expanded further, among which, the peak volume of the FHC under Case 3 was the largest, up to 4538.2 m³, which had the largest hazard range. In Cases 1, 2 and 3, the volume and mass of the FHC decreased more rapidly with the dissipation time, which further indicated that the open area was favorable for the dissipation of the FHC. On the contrary, the slower dissipation of the FHC in Cases 4, 5, and 6 implied a longer hazard duration. Moreover, the hydrogen clouds in these three cases moved slowly to the electrical substation room, where there was a certain risk of ignition for the hazard duration due to the electronic equipment involved, and the FHC could cause a fire and explosion if ignited. Case 5, which leaked towards the adjacent hydrogen storage tank group, was the worst leakage, with the slowest dissipation rate, the longest hazard duration, and the largest mass of the FHC, which implied that it had the largest hazard degree. The common cause of slow dissipation in Cases 4, 5, and 6 was the accumulation of the FHC under the canopy. Therefore, it is important to avoid setting up a canopy near the hydrogen storage area in an integrated station so that the leaked hydrogen can dissipate quickly rather than gathering under the canopy, which poses a great risk to the safety of the integrated station.

3.3. Effect of Ambient Wind Directions on Hydrogen Leakage and Dissipation Behavior

3.3.1. Leakage Processes at Different Ambient Wind Directions

The ambient wind direction has a significant impact on the process of hydrogen leakage. In this section, the hydrogen leakage process under different ambient wind directions was simulated. Ambient winds from north, south, west and east directions were considered, with a wind speed of 8 m/s. The leakage aperture was 20 mm, and the leakage direction was toward the office building.

Figure 12 shows the process of hydrogen leakage under different ambient wind directions, and the hydrogen leakage behaviors under all four wind directions changed significantly compared with the windless case. Figure 13 shows the airflow velocity distribution and streamlines at the end of leakage under different ambient wind directions. When the wind direction was north, most of the airflow was blocked by the HRS building. The leaking hydrogen was subjected to a strong and complex vortex when it moved to the area close to the enclosure wall. Its movement along the leakage direction was blocked, and it moved to the vertical direction under the effect of the buoyancy force. The FHC barely diffused to the enclosure wall at the end of the leakage. When the wind direction was south, the leaking hydrogen lost much of its kinetic energy under the strong headwind. Its diffusion distance along the leakage direction was obviously reduced. The FHC gathered in the vicinity of the hydrogen storage tank groups at the end of the leakage. When the wind direction was east, the leaked hydrogen was affected by the lateral wind from east to west and shifted to the west, and the FHC at the end of the cloud had less kinetic energy. At the end of the leakage, the FHC diffused to the vicinity of the hydrogen production workshop. When the wind direction was west, the leaked hydrogen was deflected to the east by the lateral wind. Its movement was hindered by the pump room, and the FHC gathered near the pump room area at the end of the leakage.

3.3.2. Dissipation Processes at Different Ambient Wind Directions

Figure 14 shows the dissipation process of the FHC under different ambient wind directions. When the wind direction was north, the FHC lost its momentum in the horizontal direction after the leakage stopped. It continued to diffuse in the vertical direction under the effect of local eddy currents and buoyancy forces. When the wind direction was south, the FHC diffused in a northwesterly direction under the influence of the airflow from the southeast in the integrated station. When the wind direction was east, the FHC diffused to the west and dissipated rapidly. When the wind direction was west, the FHC diffused eastward to the transformer area under the effect of the airflow and dissipated more slowly due to the impeded movement.

Figure 15 shows the variation of the volume and mass of the FHC with dissipation time at different wind directions after the leakage stopped. The peak values of the volume and mass of the FHC decreased in all four wind directions compared with the windless case, which implied that the hazard range and the hazard degree were reduced. However, it is worth noting that when the wind direction was west, the dissipation rate of the FHC was lower than the natural dissipation in windless conditions. The lower rate of dissipation was due to the lateral airflow that caused the FHC to move into the transformer area where the equipment was densely packed. Because of the large number and complexity of electrical circuits in the transformer area, there would be a higher risk of ignition of the FHC, which could easily lead to an explosion of the FHC. However, this area was sparsely populated, and an explosion would mainly result in damage to the surrounding equipment and buildings. The leakage was worst when the wind direction was south. Since the airflow direction was opposite to the leakage direction, the movement of the FHC was blocked. Then, it accumulated in a high concentration near the hydrogen storage tank groups, so its dissipation rate was the slowest, and the hazard lasted the longest.

3.4. Effect of Ambient Wind Speed on Hydrogen Leakage and Dissipation Behavior

3.4.1. Leakage Processes at Different Ambient Wind Speeds

The ambient wind speed has a direct influence on the process of hydrogen leakage. In this section, the process of hydrogen leakage under different ambient wind speeds was simulated. Four ambient wind speeds of 4 m/s, 8 m/s, 12 m/s and 16 m/s were considered. The ambient wind direction was south. The leakage aperture was 20 mm, and the leakage direction was toward the office building.

Figure 16 shows the process of hydrogen leakage under different ambient wind speeds. Figure 17 shows the airflow velocity distribution and streamlines at the end of leakage under different ambient wind speeds. It can be seen that the leaking hydrogen was affected by the airflow from the southeast direction, and its movement along the leakage direction was hindered. When the wind speed was 4 m/s, the wind level was a breeze. Although the movement of leaking hydrogen along the leakage direction was affected by the wind resistance, due to the small wind speed, the obstruction effect was not obvious. The leaking hydrogen still moved along the leakage direction to the enclosure. The head of the cloud, with a small speed, was shifted to the west for a certain distance by the action of the airflow from the southeast direction. When the wind speed was 8 m/s, due to the increase in the wind speed, the wind resistance of the leaked hydrogen in the process of movement became larger, its movement along the leakage direction was obviously restricted, and at the end of the leakage, the FHC gathered in the vicinity of the hydrogen storage tank groups. When the wind speed was 12 m/s, the wind force was strong, the speed of the airflow that hindered the movement of hydrogen along the leakage direction further increased, and the FHC diffused to the northwest direction under the action of the airflow. When the wind speed increased to 16 m/s, the wind force was very strong, and the speed of the airflow acting on the FHC reached more than 10 m/s. The hydrogen dissipated rapidly under the high-speed airflow during the leakage process, and the volume of the FHC at the end of the leakage was significantly reduced when compared with that at a wind speed of 12 m/s.

3.4.2. Dissipation Processes at Different Ambient Wind Speeds

Figure 18 shows the dissipation process of the FHC under different ambient wind speeds. As the wind speed increased, the diffusion distance of the FHC along the leakage direction gradually became smaller, and the movement toward the northwest direction was gradually enhanced. Figure 19 shows the variations of the volume and mass of the FHC with dissipation time after the leakage stopped under different wind speeds. When the wind speed was small, the FHC expanded further after the leakage stopped, and its volume increased and then decreased, similar to that under conditions of no wind. When the wind speed was greater, the volume of the FHC kept decreasing after the leakage stopped. Furthermore, the peaks of both the volume and mass of the FHC decreased with an increase in the wind speed, which implied that the increase in the wind speed could effectively reduce the peaks of the hazard range and degree. The leakage was worst when the wind speed was 8 m/s, as the airflow greatly reduced the kinetic energy of the leaking hydrogen in the station, and an FHC with a high concentration gathered near the hydrogen storage tanks, which had the slowest dissipation speed and the longest hazard duration. The highly concentrated FHC near the storage tanks may lead to a serial explosion of the tanks if there was a source of ignition. When the wind speed increased to 12 m/s and above, the FHC diffused in the northwest direction under the effect of a strong wind, and its dissipation speed was obviously accelerated. The volume and mass of the FHC were at a low level during the whole dissipation process, which meant that the potential hazard range and degree were small.

4. Conclusions

In this study, the process of hydrogen leakage and dissipation from storage tanks in an IHPRS was numerically studied. The effect of the leakage aperture, the leakage direction, the ambient wind direction and the ambient wind speed on hydrogen leakage and dissipation characteristics were investigated. The main conclusions are as follows.

(1)

As the leakage aperture increased, the volume and mass of the FHC increased as the hydrogen storage tank emptied. During the subsequent dissipation, the volume, mass and dissipation time of the FHC increased with an increasing leakage aperture, implying that leakage with a large aperture resulted in a larger hazard range, hazard degree and a longer hazard duration.

(2)

When the leakage direction was toward the ground, the leaking hydrogen formed a huge FHC on the ground level, and the hazard range was extremely large. The areas with densely packed equipment, especially in the presence of a canopy, seriously slowed down the dissipation of the FHC. Therefore, a canopy or densely packed equipment near hydrogen storage areas should be avoided.

(3)

Ambient winds can significantly affect the hydrogen leakage behavior. When the ambient wind direction was opposite to the leakage direction, the movement of hydrogen was hindered by air flows, which may have resulted in a highly concentrated FHC that gathered near the storage tanks and was difficult to dissipate. If the FHC ignites, it may lead to serial explosions of the storage tanks, which may seriously threaten the safety of the integrated station.

Due to the short duration of leakage from hydrogen storage tanks in this study, the probability of multiple simultaneous leaks was relatively small and was not considered. Meanwhile, multiple simultaneous leaks should be considered for leakage with longer durations.