You are on page 1of 8

Fuel 261 (2020) 116433

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Dynamic hazard evaluation of explosion severity for premixed hydrogen–air T


mixtures in a spherical pressure vessel
⁎ ⁎
Yun Zhanga, Weiguo Caoa, , Chi-Min Shub, , Mengke Zhaoa, Cunjuan Yua, Zhaobian Xiea,
Jinhu Lianga, Zhiqiang Songa, Xiong Caoa
a
School of Environmental and Safety Engineering, North University of China, Taiyuan 030051, PR China
b
Department of Safety, Health, and Environmental Engineering, National Yunlin University of Science and Technology, Yunlin 64002, Taiwan, ROC

G R A P H I C A L A B S T R A C T

Experiment Volume fraction: 30%

Explosion overpressure Fireba ll development


Simulation
Volume fraction: 30%

Explosive reaction
Technological mechanism of
premixed hydrogen-
approaches air mixtures has
been discussed

A R T I C LE I N FO A B S T R A C T

Keywords: To evaluate dynamic explosion severity levels of premixed hydrogen–air mixtures, pressure sensors were used to
Dynamic explosion severity levels test explosion pressure in a spherical pressure vessel (inner diameter: 0.34 m). ANSYS Fluent 19.0 three-di-
Premixed hydrogen–air mixtures mensional software was used to simulate the explosion process. The results revealed that when the hydrogen
Initial pressure volume fraction was 30 vol%, the explosion pressure and pressure rise rate reached maximum values at 1.0 atm.
Multi-dimensional transient explosion
As the initial pressure increased, the explosion pressure, and pressure rise rate increased gradually. At initial
parameters
Explosion reaction mechanism
pressures of 1.0, 1.2, 1.5, and 2.0 atm, the peak pressure levels were 0.85, 0.87, 0.92, and 0.99 MPa and the
pressure rise rates were approximately 198, 241, 302, and 558 MPa/s, respectively. The initial pressure exerted
greater effects on the pressure rise rate than did the explosion pressure. The simulation results were consistent
with the experimental findings. Moreover, the simulation yielded multi-dimensional transient explosion para-
meters—such as turbulent kinetic energy distribution—that are difficult to obtain in experiments. The findings
pertaining to the physical experiments, numerous simulations, and influence of initial pressure on the explosion
reaction mechanism of premixed hydrogen–air mixtures were discussed.

1. Introduction alleviate the negative effects of climate change, various green energy
solutions, including hydrogen gas, have been proposed as alternatives
Extensive burning of hydrocarbon fuels engenders considerable to fossil fuels for internal combustion engines. However, hydrogen oc-
concentrations of greenhouse gases, leading to climate change. To casionally engenders explosion accidents because of its large


Corresponding authors.
E-mail addresses: caoweiguoiem@nuc.edu.cn (W. Cao), shucm@yuntech.edu.tw (C.-M. Shu).

https://doi.org/10.1016/j.fuel.2019.116433
Received 12 August 2019; Received in revised form 9 October 2019; Accepted 14 October 2019
0016-2361/ © 2019 Elsevier Ltd. All rights reserved.
Y. Zhang, et al. Fuel 261 (2020) 116433

combustion limit range (4–75 vol%), tendency to have a negative detailed chemical reaction mechanism. However, the mechanism or
Joule–Thomson coefficient, and low ignition energy (0.02 mJ) [1]. factors affecting of spontaneous combustion were not thoroughly ana-
Therefore, investigating the hydrogen explosion process are helpful to lysed. Subsequently, Lee et al. and Wen et al. [31–33] have numerically
not only raise awareness of this hazard but also provide valuable data simulated and mainly solved the Navier–Stokes (N–S) equation by using
for reducing or preventing harmful accidents. unsteady two-dimensional (2D) axial symmetry and a highly detailed
Because of its low energy density, hydrogen is generally stored in multicomponent chemical reaction mechanism to simulate combustion
metal hydride storage and high-pressure storage systems. However, kinetics. In our previous study [34], a hydrogen–methane–air explosion
metal hydride storage systems have latent risks due to the instability of and its dynamic process of release under closed conditions were ana-
the metal constituting such systems. For example, Cheng et al. [2] and lysed. The experimental and simulation results revealed that the sec-
Chen et al. [3] have studied explosions caused by hydrogen storage ondary explosion under oxygen deficiency conditions was larger, and
materials releasing hydrogen at high temperature; they have found that the reaction mechanisms were preliminarily discussed. However, due to
hydrogen release increased the potential explosion damage incurred by limitations in experimental conditions and tools, the gas explosion
a combustion system. Accordingly, high-pressure storage technology is process under conditions of relatively high pressure was not studied,
an optimal alternative. Nevertheless, high-pressure hydrogen storage and a simulation process was used to resolve the N–S equation under 2D
systems involve a risk of leakage, which causes spontaneous combus- conditions. The potential effects of the gas flow field during the ex-
tion. Mironov et al. [4] investigated dynamic hydrogen diffusion plosion process were not discussed in detail.
combustion under spontaneous and forced ignition conditions; they This study first investigated the change law of explosion pressure of
observed that under conditions of large-scale hydrogen emissions, the spherical pressure vessels for premixed hydrogen–air mixtures at the
flame front velocity of hydrogen could reach 220 m/s, resulting in se- initial pressures of 1.0, 1.5, and 2.0 atm. Considering numerical re-
verely damaging shock waves. searches that can provide parameters which cannot be obtained in ex-
Hydrogen leakages that result in spontaneous ignition are serious. periments, ANSYS Fluent three-dimensional (3D) software was used for
However, hydrogen explosions caused by strong external ignition have a numerical simulation of the multi-dimensional transient reaction
been even more serious [5–8]. Thus, to better understand the hazards of process. The reaction mechanism of hydrogen at the different initial
hydrogen explosions, researchers had conducted considerable experi- explosion pressures was also investigated. The findings of this study can
mental studies to investigate explosion behaviors. For instance, Im- improve knowledge about dynamic hydrogen explosion processes and
amura et al. [9] and Schefer et al. [10] have studied hazards in hy- may serve as a reference in reducing hydrogen hazards.
drogen explosions and established empirical formula for predicting
flame configuration, temperature increase, and heat flow trajectory 2. Experimental
dynamics. Groethe et al. [11] have investigated on hydrogen–air de-
flagration enhancement by placing an obstacle in a hemispherical ex- Fig. 1 displays the gas explosion test device and a spherical pressure
perimental device (volume: 300 m3); and they found that the obstacle vessel with an inner diameter of 0.34 m. The device comprises several
did not increase the deflagration combustion rate when the blocking parts: A sphere (including the ignition system) with a maximum pres-
percentage is 11%. Moreover, Zheng et al. [12,13] and Zhang et al. [14] sure of 2 MPa, gas premix system, pressure recording device, and syn-
have probed the flame propagation velocity of multicomponent gas chronous control system. Before the test was started, the explosive
mixtures at a relatively high blocking rate. They have discovered that vessel was pumped to create a vacuum. First, hydrogen and air were
the flame velocity increased with the blocking rate; however, this as- premixed through a gas mass flow meter (range: 0–500 mL/min, and
sociation did not cause a deflagration-to-detonation transition. Flame uncertainty: ± 0.5%) and stored in a premixed cylinder. Then, the
propagation of premixed hydrogen–air mixtures was discovered to be premixed gas was injected into the explosive ball until the required
more intense under pure–oxygen conditions. Zhang et al. [15–17] have pressure required for testing was attained. After 20 min (to ensure the
analysed the detonation propagation behaviour of multicomponent gas gas was static or flowed slowly), an electrostatic ignition of 100 mJ was
mixtures in spiral obstacles. They have collectively examined the be- administered using an electrostatic generator. A synchronous100 mj
haviors of those mixtures and the dual effects of spiral obstacles–of control system was adopted to govern the pressure measurement of the
varying roughness–on the promotion and inhibition of detonation flame device. The data pertaining to the hydrogen explosions at various
characteristics. Cao et al. [18,19] have experimentally investigated the fractions are summarized in Table 1. All the experiments were con-
severity of hydrogen explosions for different hydrogen volume fractions ducted at room temperature, that is, at approximately 20–30 °C, with a
and reduced the destructibility of explosions through explosive release. humidity value of ca. 20–30%.
However, the research results were not consistent with the calculations
of the European standard EN 14994 [20], indicating that current safety 3. Numerical simulation
standards and specifications on hydrogen explosions require further
improvement. Additionally, formulating a universal law governing such 3.1. Geometric model
explosions is difficult because of inconsistencies in results and differ-
ences in experimental devices between studies as well as the limited A geometric model was established in accordance with the actual
availability of testing data [21]. experimental sphere, as shown in Fig. 2. The experimental sphere had
Computerized numerical simulation has rapidly developed as a ignition electrodes, and the geometric model was an irregular sphere
supplemental analysis approach, thus saving practical experimental with ignition electrodes instead of a regular sphere. To reflect the dy-
costs. Besides, the simulations can provide insight into hydrogen ex- namic changes that occurred during the simulated explosion more ef-
plosion mechanisms by analysing parameters that are unmeasurable in ficiently, dynamic tracking and data collection were conducted for
experiments. For example, such simulations enable modelling multi- pressures at the centre point of the ignition position and the positions R
dimensional transient reaction processes (using parameters including (sphere radius, 0.17 m), 2R/3, and R/3.
explosion flame propagation velocity, gas flow propagation velocity,
and explosion temperature field distribution) and capturing pressure 3.2. Mesh model
field distributions within an entire explosion range [22–26]. Molkov
et al. [27,28] and Li et al. [29] have modelled hydrogen combustion The ANSYS Fluent 3D software was used to replicate the explosion
using the large eddy simulation method, yielding a relatively consistent process of premixed hydrogen, as shown in Fig. 3. A tetrahedral mesh
computational fluid dynamics model for gaseous explosions. Liu et al. was used to divide the geometric model area, and all wall areas were
[30] simulated hydrogen ignition using the Euler equation and a successfully generated. To ensure the grid independence and the

2
Y. Zhang, et al. Fuel 261 (2020) 116433

Fig. 1. Schematic diagram of spherical pressure vessel.

Table 1
Hydrogen, oxygen, nitrogen, and air volume fraction data.
Hydrogen Oxygen Nitrogen Air (oxygen + nitrogen)
(vol.%) (vol.%) (vol.%) (vol.%)

10 18.9 71.1 90
20 16.8 63.2 80
28.6 15.0 56.4 71.4
30 14.7 55.3 70
40 12.6 47.4 60
50 10.5 39.5 50
60 8.4 31.6 40
70 6.3 23.7 30

Fig. 3. Structural mesh of computational domain.

convergence of the calculation results, five different grids were calcu-


lated. The optimal number of elements was 366,988 because any fur-
ther increase in the number of elements had a negligible effect on the
calculation results. Each time step in the simulation involved 20
iterations, and the calculation residual of each time step was less than
0.001, which ensured the convergence of the calculation results.
An ignition patch (volume: 524 mm3; temperature: 1200 K) was
gradually applied to ignite the premixed hydrogen–air mixtures. In the
simulation, a particularly large ignition patch resulted in a short flame
propagation time, engendering a time scale error. Furthermore, a lower
temperature resulted in a longer flame propagation process or even
prevented this process. Thus, to ensure calculation result convergence
Fig. 2. Geometric model of computational domain. and accuracy, simultaneously verifying the mesh independence, time

3
Y. Zhang, et al. Fuel 261 (2020) 116433

step, ignition patch, and temperature was emphasised and executed in combustion, and fewer hydrogen molecules participated in the com-
this study. bustion; additional increases in the hydrogen fraction thus resulted in
decreases in the explosion pressure. At 30 vol%, Pmax and (dP/dt)max
3.3. Numerical methods were 0.85 and 199 MPa/s, respectively. Consequently, according to Eq.
(1), the maximum explosion index was 53.83 MPa·m/s.
ANSYS Fluent 3D software was used to simulate the explosions of The numerical simulation of the dynamic explosion process was
premixed hydrogen–air mixtures. Mass, energy, momentum, and spe- conducted using Fluent 3D software to observe the three-dimensional
cies transport conservation equations constituted the governing equa- transient explosion process. This simulation revealed the field dis-
tions, details of which can be found in a previous publication [35]. tribution of the explosion pressure, airflow propagation velocity, and
Scale-adaptive simulation turbulence models, other than the Reynolds- explosion temperature. It also enabled further analysis of the influence
averaged N–S model [36,37], were adopted to simulate the explosions of various other parameters on the severity of dynamic explosion pro-
of the premixed hydrogen–air mixtures. The equations used in the si- cess. Fig. 5 presents the simulation results regarding the influence of
mulation were expressed previously by [38]. Moreover, a flame front four parameters on the dynamic explosion process: explosion tem-
propagation model was established by solving the transport equation of perature spatial distribution, gas flow velocity, explosion pressure, and
density-weighted average reaction process variables [34]. These equa- turbulent kinetic energy. Compared with existing test methods, the
tions were detailed in previously [39]. present study’s methods yielded more detailed data about the explosion
process. For example, this study determined that the timing of the
4. Results and discussion maximum values of various parameters was not consistent. Specifically,
gas flow velocity reached its maximum level 5 ms after ignition,
4.1. Dynamic explosion process of premixed hydrogen–air mixtures whereas pressure, turbulent kinetic energy, and flame temperature
reached their maximum levels 7, 12, and 12 ms after ignition, respec-
The maximum pressure (Pmax) and explosion index (Kg) are con- tively (Fig. 7).
stitute characteristic parameters representing ignition and explosion At the initiation of combustion, the entire reaction system tem-
severity. They are essential for accurate hazard evaluation [40]. The perature and flame propagation speed were low (Fig. 5). The flame
explosion index Kg can be calculated with respect to the maximum front extended outside position R/2 from the centre of the sphere in
pressure rise rate (dP/dt)max: 3 ms (Fig. 5(a)). Apart from the local turbulence zone affected by the
ignition electrode, the combustion gas exhibited a laminar flow in all
dP 4πR 3 directions inside the sphere (Fig. 5(b)). As the reaction progressed, the
Kg = ⎛ ⎞ 3 ⎛ ⎞
⎝ dt ⎠max ⎝ 3 ⎠ (1) gas continued to spread, gradually filling the entire sphere. After 5 ms,
the maximum gas flow speed was 200 m/s. the gas flow commenced to
where t is the time; P is the explosion pressure; and R is sphere radius.
be restrained by the spherical wall surface and then forms reflection.
In this study, experiments were conducted on premixed hydro-
Turbulence gradually dominated the combustion process, thus resulting
gen–air mixtures at the following hydrogen volume fractions: 10, 20,
in an increase in turbulent kinetic energy. In theory, turbulent kinetic
28.6, 30, 40, 50, 60, and 70 vol%, as shown in Fig. 4.
energy is related to turbulence velocity and fluid quality, which can be
P and (dP/dt) initially increased with the hydrogen volume fraction
used to estimate turbulence intensity [41]. Within 5 ms of ignition, the
increased, and then decreased as the volume fraction continued to in-
turbulent kinetic energy level was low (Fig. 5(c)), and the evolution of
crease. At very low hydrogen volume fractions, the oxygen concentra-
turbulent kinetic energy lagged behind the change in airflow velocity.
tion was insufficient to sustain a reaction. Accordingly, P and (dP/dt)
12 ms after ignition, the maximum turbulent kinetic energy value was
were primarily affected by the hydrogen volume fraction. Furthermore,
reached 150 m2/s2.
a heat transfer analysis revealed that the number of molecules available
The temperature and pressure inside the sphere also increased. The
for combustion increased with the hydrogen volume fraction; hence,
inner wall pressure reached a maximum of 0.90 MPa at 7 ms (Fig. 5(d)).
the total heat released and the relative strength of the entire combus-
Furthermore, the flame front reaches the boundary at 5 ms after igni-
tion process increased. P and (dP/dt) thus enhanced the combustion
tion (Fig. 5(a)), the flame temperature rises further during a later time
process and attained a maximum value. The optimal hydrogen volume
and reaches the maximum value at 12 ms. Moreover, the explosion
fraction (30 vol%) was slightly larger than the equivalent ratio fraction
pressure reached the maximum value of 1.12 MPa at 16 ms (Fig. 5(d)),
(28.6 vol%). At higher hydrogen volume fractions, the quantity and
indicating that the chemical reaction in the sphere were continuing and
density of oxygen in the confined space were insufficient for

(a) Explosion pressure (P) (b) Pressure rise rate (dP/dt)

Fig. 4. Experimental explosion severities (pressure and pressure rise rate) of hydrogen–air mixture at different volume fractions. (a) explosion pressure (P), (b)
pressure rate rise (dP/dt).

4
Y. Zhang, et al. Fuel 261 (2020) 116433

(a) Spatial distribution of explosion temperature

(b) Spatial distribution of gas flow velocity

(c) Spatial distribution of turbulent kinetic energy

(d) Spatial distribution of explosion pressure

Fig. 5. Simulation graphs during explosion process of hydrogen–air mixtures. Volume fraction: 30%.

the hydrogen was not completely exhausted. establishing heat transfer by radiation required time. Under the con-
The temperature reached a maximum value above 2000 K at 12 ms straint of the wall surface, the gas flow and pressure wave were re-
after the pressure reached its maximum value. This was because flected to the centre of the sphere. When the airflow velocity and

5
Y. Zhang, et al. Fuel 261 (2020) 116433

(a) Different volume fractions (b) Different distances from central


position. Volume fraction: 30%

Fig. 6. Simulation explosion pressure history of hydrogen–air mixture. (a) different volume fractions, (b) different distances from central position. Volume fraction:
30%.

pressure wave reached 12 and 16 ms, respectively, the maximal values Fig. 6(b) presents the dynamic change in pressure at the positions R
of 180 m/s and 1.12 MPa were observed again at the centre of the (the wall), 2R/3, R/3, and 0 from the centre of the sphere ignition
sphere. The pressure at the centre of the sphere was considerably higher during the simulation process. Before the explosion pressure at different
than the maximum wall pressure obtained previously, indicating the locations reached the first maximum peak, the maximum value of ex-
limitations in existing methods of testing maximum wall pressure plosion pressure gradually increased with an increase at the distance
(using a wall pressure sensor). Therefore, it is necessary to evaluate the from the ignition centre. The increasing trends of explosion pressure at
change in explosion pressure under different experimental conditions. different locations were consistent. The reflection of the vessel interior
in terms of pressure oscillation. The pressure curves at different posi-
tions were observed after the explosion pressure reached its first peak.
4.2. Explosion pressure results under ambient pressure The second explosion pressure peaks at R, 2R/3, R/3, and 0 were 0.77,
0.79, 1.01, and 1.12 MPa, respectively. The pressures at R/3 and 0 were
Fig. 6 presents simulated explosion pressure levels for the hydro- higher than the maximum wall pressure in the experiment. In Fig. 6(b),
gen–air mixtures. Comparing the explosion pressure–time curves for the box with the black dotted lines demarcating the region between
different hydrogen volume fractions in the simulation (Fig. 6(a)) and 13.3 and 18.8 ms indicates the explosion pressures at R/3 and 0, which
experimental (Fig. 4) measurements conducted at the vessel’s wall were greater than the maximum wall pressure. During this period, the
surface revealed that experimental and simulation results were con- system was in its most violent phase. As the reaction time progressed,
sistent. When explosion pressure increased, the error between the si- the pressure oscillation gradually synchronized, and the peak pressure
mulation and experiment was within 10%, and the pressure change in in the vessel stabilized at 0.7 MPa.
the pressure testing point on the wall of the vessel was basically con-
sistent. At the same hydrogen volume fraction, the simulated maximum
explosion pressure was slightly higher than its experimental counter- 4.3. Effect of initial pressure on explosion severity
part; this was primarily due to the complete and uniform mixing of
hydrogen and air in the simulation. As the hydrogen volume fraction To further evaluate the influence of initial pressure on explosion
increased, the error between the simulation and experiment decreased severity, pressure values of 1.2, 1.5 and 2.0 atm were deliberately se-
gradually, further indicating the reliability of the simulation in mod- lected to probe the severity of explosion on the basis of the above study
elling the hydrogen explosion process. with initial pressure of 1.0 atm. At 30 vol%, the explosion pressure

(a) Experiment (b) Simulation

Fig. 7. Explosion pressure history of hydrogen–air mixture at different initial pressures. Volume fraction: 30%. (a) experiment, (b) simulation.

6
Y. Zhang, et al. Fuel 261 (2020) 116433

hydrodynamics and diffusional-thermal instability would result in the


self-accelerative propagation of a spherically expanding flame, conse-
quently enhancing an explosion. Similar phenomena have been ob-
served by Hu et al. [44], Xie et al. [45], and Cai et al. [46]. Therefore, in
industrial applications where hydrogen is stored under high pressure,
the risk of human errors or operating accidents is extraordinarily high.
Thus, the explosion process of hydrogen must be thoroughly under-
stood.

5. Conclusions

This study was evaluated the explosion severity of various premixed


hydrogen–air mixtures. The results reveal the following:
When the initial reaction pressure was 1 atm, P and (dP/dt) in-
creased with the hydrogen volume fraction increased, attained a max-
imum value, and then decreased at higher volume fraction amounts.
Moreover, the dynamic explosion process of premixed hydrogen–air
Fig. 8. Experimental pressure rise rate history of hydrogen–air mixtures at
mixtures was revealed. The pressure, turbulent kinetic energy, and
different initial pressures.
flame temperature attained maximum levels later than the gas flow
velocity did. Under the wall constraint, the airflow and pressure wave
increased gradually with the initial reaction pressure (Fig. 7); the peak were reflected to the centre of the sphere. The air velocity and pressure
explosion pressure obtained from the simulation was slightly higher wave, in turn, attained a maximum value at the centre of the sphere
than that obtained from the experiment, consistent with the hydrogen again. The peak pressure at the centre of the sphere was notably higher
explosion pressure curves at different hydrogen volume fractions under than the experimental peak wall pressure, indicating the limitations of
standard atmospheric pressure. Relative to the history of the (dP/dt) of existing methods of evaluating the dynamic explosion severity.
the hydrogen–air mixtures at different initial pressures (Fig. 8), the rate On the stage of increasing explosion pressure, the simulation results
of change of explosion pressure due to the change in initial pressure reproduced the experimental process well. The pressure changes at the
increased slightly. However, the rate of (dP/dt) increase was larger. In pressure test points of the vessel wall were consistent, which indicated a
the experiment, when the initial pressure reached 2 atm, Pmax and (dP/ reliable result. The hydrogen explosion process can be reproduced and
dt)max increased to 1.0 atm and 558 MPa/s, respectively. analysed through simulation with confidence. Combined with the
Based on the Arrhenius equation and the Lewis theory of molecular pressure of the simulation process, the prominent pressure oscillation
collision [42,43], the average relative velocity of molecules A and B was observed in the pressure curves at different positions due to the
that effectively collide during the reaction of hydrogen and oxygen is reflection inside the vessel after the explosion pressure attained its first
given as presented in Eq. (2): peak. Between 13.3 and 18.8 ms, the explosion pressure of R/3 and 0
1/2 1/2 was always greater than the maximum value of the wall pressure. As the
8k T 8k T 8k T
v¯r = (v¯A2 + v¯B2 )1/2 = ⎛ B r + B r ⎞
⎜ ⎟ = ⎜⎛ B r ⎟⎞ reaction time progressed, the pressure oscillation period gradually
⎝ πm A πmB ⎠ ⎝ πμm ⎠ (2) synchronized, and the maximum value of the pressure in the vessel
stabilized at 0.7 MPa.
where kB is the Boltzmann constant; Tr is the reaction temperature; μm
An increase in the initial reaction pressure led to a decrease in the
is the equivalence to mass μ = mAmB/(mA + mB); and v̄A and v̄ B are the
distance between molecules. The number of molecules that collides
motion velocities of molecules A and B, respectively.
increased, which expedited the chemical reaction rate and promoted an
The number (ZAB) of collisions of molecules A and B at a unit time
increase in the explosion pressure. Subsequently, increasing the ex-
and unit volume can be obtained as presented in Eq. (3):
plosion pressure expedited the chemical reaction rate, which increased
1/2
the total chemical reaction rate. The rate of explosion pressure in-
2 ⎛ 8kB Tr ⎞
ZAB = πrAB ⎜ ⎟ NA NB creased due to the coupling effect of the initial pressure and explosion
⎝ πμm ⎠ (3)
pressure. Therefore, in industrial applications where hydrogen is stored
where rAB is the sum of radius of molecules A and B; and NA and NB are under high pressure, the risk of human errors or operating accidents is
the number of molecules A and B in unit volume, respectively. extremely high. Thus, the explosion process of hydrogen must be
At a certain reaction temperature, the distance between molecules thoroughly understood.
decreases as the initial pressure of the reaction system increases. Eqs.
(2) and (3) indicate that the number of molecules colliding increases Declaration of Competing Interest
gradually and the chemical reaction rate is thus accelerated; conse-
quently, Pmax increases. Fig. 9 confirms that the peak explosion pressure The authors declare that they have no known competing financial
increases linearly with the initial pressure. interests or personal relationships that could have appeared to influ-
In the closed reaction system, the increase in the initial reaction ence the work reported in this paper
pressure accelerated the chemical reaction rate and promoted the in-
crease in explosion pressure. While the increase in the explosion pres- Acknowledgements
sure further expedited the chemical reaction. Accordingly, the total
chemical reaction rate increased and (dP/dt) exhibited the law of ex- This work was greatly supported by the Natural Science Foundation
ponential growth due to the coupling effect of initial pressure and ex- of China (11802272), the China Postdoctoral Science Foundation
plosion pressure. When the initial pressure increased from 1.0 to (2019M651085), the Natural Science Foundation of Shanxi Province
2.0 atm, the maximum value of the explosion pressure increased by (201801D121285), the Cultivation Programs for Young Scientific
1.16 folds and the maximum value of (dP/dt) increased by 2.82 folds Research Personnel of Higher Education Institutions in Shanxi Province
(Fig. 9). From the perspectives of flame propagation, as the initial (1912200059MZ), and the Scientific and Technological Innovation
pressure increased, the flame propagation speed was higher and com- Programs of Higher Education Institutions in Shanxi (201802079 and
bustion duration was shorter. A possible explanation for this is that 2017152).

7
Y. Zhang, et al. Fuel 261 (2020) 116433

(a) Explosion pressure and pressure rise rate (b) Rate of explosion severity vs. initial
peak vs. initial pressure pressure

Fig. 9. Experimental explosion behaviors of hydrogen–air mixtures at different initial pressures. (a) explosion pressure and pressure rise rate peak vs. initial pressure,
(b) rate of explosion severity vs. initial pressure.

References [24] Li HT, Chen XK, Shu CM, et al. Experimental and numerical investigation of the
influence of laterally sprayed water mist on a methane-air jet flame. Chem Eng J
2019;356:554–69.
[1] Crowl DA, Jo YD. The hazards and risks of hydrogen. J Loss Prev Process Ind [25] Ugarte OJ, Akkerman V, Rangwala AS. A computational platform for gas explosion
2007;20:158–64. venting. Process Saf Environ Prot 2016;99:167–74.
[2] Cheng YF, Meng XR, Ma HH, et al. Flame propagation behaviors and influential [26] Sezer H, Kronz F, Akkerman V, et al. Methane-induced explosions in vented en-
factors of TiH2 dust explosions at a constant pressure. Int J Hydrogen Energy closures. J Loss Prevent Proc Ind 2017;48:199–206.
2018;43:16355–63. [27] Molkov V, Makarov D, Schneider H. LES modelling of an unconfined large-scale
[3] Chen SH, Tang Y, Yu HS, et al. The rapid H2 release from AlH3 dehydrogenation hydrogen–air deflagration. J Phys D: Appl Phys 2006;39:4366–76.
forming porous layer in AlH3/hydroxyl-terminated polybutadiene (HTPB) fuels [28] Molkov V, Verbecke F, Makarov D. LES of hydrogen-air deflagrations in a 78.5-m
during combustion. J Hazard Mater 2019;371:53–61. tunnel. Combust Sci Technol 2008;180:796–808.
[4] Mironov VN, Penyazkov OG, Ignatenko DG. Self-ignition and explosion of a 13-MPa [29] Li GQ, Du Y, Wang SM, et al. Large eddy simulation and experimental study on
pressurized unsteady hydrogen jet under atmospheric conditions. Int J Hydrogen vented gasoline-air mixture explosions in a semi-confined obstructed pipe. J Hazard
Energy 2015;40:5749–62. Mater 2017;339:131–42.
[5] Li YC, Bi MS, Li B, et al. Effects of hydrogen and initial pressure on flame char- [30] Liu YF, Tsuboi N, Sato H, et al. Direct numerical simulation on hydrogen fuel jetting
acteristics and explosion pressure of methane/hydrogen fuels. Fuel from high pressure tank, Proceedings of the 20th International Colloquium on the
2018;233:269–82. Dynamics of Explosions and Reactive Systems, Montreal, Canada; 2005.
[6] Li YC, Bi MS, Li B, et al. Explosion hazard evaluation of renewable hydrogen/am- [31] Lee HJ, Park JH, Kim SD, et al. Numerical study on the spontaneous-ignition fea-
monia/air fuels. Energy 2018;159:252–63. tures of high-pressure hydrogen released through a tube with burst conditions. Proc
[7] Zhang C, Wen J, Shen XB, et al. Experimental study of hydrogen/air premixed flame Combust Inst 2015;35:2173–80.
propagation in a closed channel with inhibitions for safety consideration. Int J [32] Lee BJ, Jeung IS. Numerical study of spontaneous ignition of pressurized hydrogen
Hydrogen Energy 2019;44:22654–60. released by the failure of a rupture disk into a tube. Int J Hydrogen Energy
[8] Shen XB, Zhang C, Xiu G, et al. Evolution of premixed stoichiometric hydrogen/air 2009;34:8763–9.
flame in a closed duct. Energy 2019;176:265–71. [33] Wen JX, Xu BP, Tam VHY. Numerical study on spontaneous ignition of pressurized
[9] Imamura T, Hamada S, Mogi T, et al. Experimental investigation on the thermal hydrogen release through a length of tube. Combust Flame 2009;156:2173–89.
properties of hydrogen jet flame and hot currents in the downstream region. Int J [34] Rao GN, Zhang Y, Cao WG, et al. Experimental and numerical studies of premixed
Hydrogen Energy 2008;33:3426–35. methane-hydrogen/air mixtures flame propagation in closed duct. Can J Chem Eng
[10] Schefer RW, Houf WG, Williams TC, et al. Characterization of high-pressure, un- 2018;96:2684–9.
derexpanded hydrogen-jet flames. Int J Hydrogen Energy 2007;32:2081–93. [35] Zhang Y, Jiao FY, Huang Q, et al. Experimental and numerical studies on the closed
[11] Groethe M, Merilo E, Sato Y. Large-scale hydrogen deflagrations and detonations. and vented explosion behaviors of premixed methane-hydrogen/air mixtures. Appl
Int J Hydrogen Energy 2007;32:2125–33. Therm Eng 2019;159:113907.
[12] Zheng K, Yu MG, Liang YP, et al. Large eddy simulation of premixed hydrogen/ [36] Menter FR. Two-equation eddy-viscosity turbulence models for engineering appli-
methane/air flame propagation in a closed duct. Int J Hydrogen Energy cations. AIAA J 1994;32:1598–605.
2018;43:3871–84. [37] Egorov Y, Menter FR. Development and application of SST-SAS turbulence model in
[13] Zheng K, Yu MG, Zheng LG, et al. Effects of hydrogen addition on methane-air the DESIDER project. Adv Hybrid RANS-LES Model 2008:261–70.
deflagration in obstructed chamber. Exp Therm Fluid Sci 2017;80:270–80. [38] Menter FR, Egorov Y. The scale-adaptive simulation method for unsteady turbulent
[14] Zhang Q, Wang YX, Lian Z. Explosion hazards of LPG-air mixtures in vented en- flow predictions. Part 1: theory and model description. J Flow Turbul Combust
closure with obstacles. J Hazard Mater 2017;334:59–67. 2010;85:113–38.
[15] Zhang B, Liu H, Yan BH. Effect of acoustically absorbing wall tubes on the near-limit [39] Zimont V, Polifke W, Bettelini M, et al. An efficient computational model for pre-
detonation propagation behaviors in a methane-oxygen mixture. Fuel mixed turbulent combustion at high Reynolds numbers based on a turbulent flame
2019;236:975–83. speed closure. J Gas Turbines Power 1998;120:526–32.
[16] Zhang B, Liu H. The effects of large scale perturbation-generating obstacles on the [40] Cao WG, Qin QF, Cao W, et al. Experimental and numerical studies on the explosion
propagation of detonation filled with methane–oxygen mixture. Combust Flame severities of coal dust/air mixtures in a 20-L spherical vessel. Powder Technol
2017;182:279–87. 2017;310:17–23.
[17] Zhang B, Liu H, Yan BH. Investigation on the detonation propagation limit criterion [41] Li J, Li QQ, Wang YT, et al. Fundamental flame characteristics of premixed H2–air
for methane-oxygen mixtures in tubes with different scales. Fuel 2019;239:617–22. combustion in a planar porous micro–combust. Chem Eng J 2016;283:1187–96.
[18] Cao Y, Guo J, Hu KL, et al. Effect of ignition location on external explosion in [42] Lewis JC, Van Kranendonk J. Intercollisional interference effects in collision-in-
hydrogen–air explosion venting. Int J Hydrogen Energy 2017;42:10547–54. duced light scattering. Phys Rev Lett 1970;24:802–4.
[19] Cao Y, Guo J, Hu KL, et al. The effect of ignition location on explosion venting of [43] Lewis JC, Van Kranendonk J. Theory of intercollisional interference effects. I. in-
hydrogen–air mixtures. Shock Waves 2017;27:691–7. duced absorption. Can J Phys 1976;50:2881–901.
[20] NFPA 68. Standard on explosion protection by deflagration venting. Quincy, [44] Hu EJ, Huang ZH, He JJ, et al. Experimental and numerical study on laminar
Massachusetts, USA: National Fire Protection Association; 2013. burning velocities and flame instabilities of hydrogen–air mixtures at elevated
[21] Wang K, He YR, Liu ZY, et al. Experimental study on optimization models for pressures and temperatures. Int J Hydrogen Energy 2009;34:8741–55.
evaluation of fireball characteristics and thermal hazards induced by LNG vapor [45] Xie YL, Wang JH, Cai X, et al. Self-acceleration of cellular flames and laminar flame
cloud explosions based on colorimetric thermometry. J Hazard Mater speed of syngas/air mixtures at elevated pressures. Int J Hydrogen Energy
2019;366:282–92. 2016;41:18250–8.
[22] Bind VK, Roy S, Rajagopal C. A reaction engineering approach to modeling dust [46] Cai X, Wang JH, Zhao HR, et al. Flame morphology and self-acceleration of syngas
explosions. Chem Eng J 2012;207–208:625–34. spherically expanding flames. Int J Hydrogen Energy 2018;43:17531–41.
[23] Di Sarli V, Di Benedetto A, Russo G. Large eddy simulation of transient premixed
flame–vortex interactions in gas explosions. Chem Eng Sci 2012;71:539–51.

You might also like