i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Release characteristics of highly pressurized

hydrogen through a small hole

Sang Heon Han a,*, Daejun Chang a,*, Jong Soo Kim b
a
Division of Ocean System Engineering, Korea Advanced Institute of Science and Technology, 373-1 Guseong-dong, Yuseong-gu,
Daejeon 305-701, South Korea
b
Korea Institute of Science and Technology, 39-1 Hawolgok-dong, Seongbuk-gu, Seoul, South Korea

article info abstract

Article history: A hydrogen supplying system for hydrogen fuel cell cars is anticipated to utilize highly
Received 18 September 2012 pressurized hydrogen gas at pressures up to 700 bar. In this highly pressurized environ-
Received in revised form ment, large amount of hydrogen can be leaked from a relatively small hole caused by
13 November 2012 material and mechanical defects. A leaked hydrogen jet can reach very far in distance and
Accepted 16 November 2012 the size of the leak plays an important role in determining the safety of hydrogen recharge
Available online 23 January 2013 facilities. This study numerically investigated the concentration distribution and the mass
flux of hydrogen leaked from a highly pressurized source through a hole whose size is less
Keywords: than 1.0 mm. Numerical analysis was performed in axisymmetric coordinates on the
High pressure hydrogen leakage assumption that the hydrogen jet has a huge Froude number and that buoyancy forces can
Jet dispersion be negligible. The predicted hydrogen concentration along the centerline of a hydrogen jet
Small hole was compared with experimental data for verification of the numerical analysis and it
Mass flux satisfied the hyperbolic decay characteristics and matched the experiment well. The mass
fluxes for the various hole lengths of this study were found to be 5%e20% less than those
predicted using an isentropic flow assumption.
Copyright 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
reserved.

1. Introduction carbon energy strategy is significantly more costly than the

CCS (CO2 Capture and Storage) strategy applicable for the
Transportation, which accounts for approximately 20% of CO2 power and industry sectors, in that the marginal CO2 abate-
emission from by fossil fuel combustion in typical industri- ment cost for the transportation sector range from 200e500$/
alized countries, is the third major CO2 emission sector, tCO2, whereas those for the power and industry sectors range
following the power and industry sectors. However, formu- from 50e100$/tCO2 and 100e200$/tCO2, respectively. Never-
lating a CO2 reduction strategy for the transportation sector is theless, a non-carbon energy strategy is still necessary to
much more difficult because motor vehicles are nonstationary achieve a deep CO2 reduction from the transportation sector
and extremely small CO2 emission sources, emitting about to stabilize the atmospheric CO2 level by 2050. The IEA-BLUE
one millionth of the amount of large scale coal power plants Map scenario, developed with a specific target to CO2 stabili-
and blast furnaces. Consequently, in the Energy Technology zation by 2050, calls for a 50% CO2 emission reduction globally,
Perspectives 2008 [1], the International Energy Agency (IEA) which can be translated into an 80% CO2 emission reduction
recommended a non-carbon energy strategy, i.e., employing for typical industrialized countries and a 90% CO2 emission
hydrogen or electricity to power motor vehicles. This non- reduction for the USA.

* Corresponding authors. Tel.: 82 42 350 1574; fax: 82 42 350 1510.

E-mail addresses: freezia@kaist.ac.kr (S.H. Han), djchang@kaist.edu (D. Chang).
0360-3199/$ e see front matter Copyright 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijhydene.2012.11.071
3504 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2

Nomenclature T temperature, K
uz, ur velocity components, m/s
A coefficient in Eq. (21)
XH2 hydrogen concentration
a, b coefficients of line in Eq. (23)
Yk mass fraction of k-th species
CD discharge coefficient
z axial coordinate, m
Cp,k specific heat of k-th species, J/(kg K)
d diameter of hole, m Greek symbols
dps pseudo diameter of abrupt expansion, m 3 turbulent dissipation
e z; !
! er unit vectors, m m, mt, meff molecular viscosity, turbulent viscosity, and
k turbulent kinetic energy or decay coefficient effective viscosity, kg/s m
h specific enthalpy, J/kg r density, kg/m3
L length of hole, m g ratio of specific heats
m_ mass flux, kg/m2 s F dissipation function
Pr Prandtl number
Subscripts
p pressure, N/m2
k index for element or species
r radial coordinate, m
s isentropic
R gas constant, J/(kg K)
LFL lean flammability limit
Sc Schmit number

At this moment, it is unclear which will emerge as the Houf and Schefer [6,7] carried out experimental studies on the
dominant non-carbon energy source for the transportation dispersion of hydrogen jets arising from leakage with low
sector among hydrogen and electricity. To win the competition pressures.
to become the dominant non-carbon transportation energy Numerical studies on the release of highly pressurized
system, it is necessary to develop not only the hydrogen or hydrogen have been performed in two ways. The first and
electric vehicle technologies into a system capable of supplying major concern of this study [8e14] is performed by analyzing
power in as large a quantity as that of gasoline engine vehicle, the dispersion of the hydrogen jet into ambient air in view of
but more importantly the corresponding energy infrastructure safety. The domain sizes of these works are very large, such as
from the energy producers to the terminal users. In this regard, factory fields or urban sections. In this case, the precise
the hydrogen energy system is at a significant disadvantage characteristics of the abrupt hydrogen expansion just after
because it has yet to be built and the safety of the hydrogen exit through the cracked hole are neglected because of
energy system needs to be fully addressed. In particular, numerical inappropriateness. The mass flow rate for the inlet
hydrogen filling stations are under strict public scrutiny because boundary is calculated using an isentropic flow assumption.
it is where the energy system comes into contact with the public. The second way [15,16] focuses on the abrupt expansion
Hydrogen is one of the most reactive compounds. On the characteristics of the jet. This way explores the expansion
other hand, hydrogen has a very low volumetric energy length, pseudo diameter, and normal shock e essential data
density and a great tendency to buoyantly disperse away from for the Birch approach.
a leaking source. Consequently, the safety of hydrogen filling This study directly calculates both the mass flux and the
stations can be greatly improved by promoting the dispersion dispersion characteristics of a hydrogen jet leaked from
of leaking hydrogen before exposure to an ignition source. a highly pressurized source using numerical analysis. The
When developing the hydrogen station safety code that will mass flux can be obtained by calculating the flow field inside
guarantee the safety for such station operators as well as for a source reservoir, which is almost stagnant, and a cracked
the public, better quantified hydrogen jet characteristics will hole, which is being choked. The typical velocity and length
enable us to maximize the safety potential by specifying scales for high pressure hydrogen leakage are assumed to be
guidelines to promote the dispersion and dilution of the highly u w 1500 m/s, corresponding to the hydrogen sound speed,
reactive hydrogen. Therefore, it is the objective of this present and d w 1 mm, respectively. The resulting Froude number,
study to quantify the characteristics of hydrogen leaking from defined to be u2/gd, is on the order of 108, so that the buoyancy
a high pressure system through a small rupture hole. effect, which is capable of bending the hydrogen jet upward, is
Birch [2] demonstrated that the mean centerline concen- negligible and two dimensional axisymmetricity can be
tration profiles for various natural gas jets can be collapsed assumed to significantly reduce numerical efforts.
into a single curve if the longitudinal distance from a virtual
point source is non-dimensionalized by a virtual diameter
derived from the mass balance. Birchs approach for natural 2. Mathematical formulation
gas jets was later extended to hydrogen jets by Ruffin et al. [3].
The concentration profile of hydrogen jets, established by 2.1. Isentropic approach
leakage with constant pressures at up to 25 bar, was measured
by Shirvill et al. [4] with an oxygen sensor. Takeno et al. [5] Fig. 1 depicts the hydrogen jet leaked from a source, which can
measured the transient hydrogen concentration of a hypo- be a storage vessel or a flow line. When hydrogen is released
thetical scenario for a large scale leakage from a pipeline. through a hole created in a high pressure vessel or flow line,
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2 3505

energy and eddy dissipation rate equations governing the

thermo-fluidic characteristics can be written as:
INFINIT
RESERVOIR v 1v
ruz rrur 0 (5)
vz r vr

v 1v vp vszz 1 v
ruz uz rrur uz  rsrz (6)
vz r vr vz vz r vr

v 1v vp vszr 1 v sqq
ruz ur rrur ur  rsrr  (7)
vz r vr vr vz r vr r

  
 
v 1v v m vk 1v m vk
ruz k rrur k m t r m t Gr3
vz r vr vz sk vz r vr sk vr
(8)


  
 
Fig. 1 e Schematics of abrupt expansion (Ref. [2]). v 1v v m v3 1v m v3
ruz 3 rrur 3 m t r m t
vz r vr vz s3 vz r vr s3 vr
2
3 3
C3 1 G  C3 2 r 9
k k
the flow is choked near the exit of the hole. Neglecting viscous where
dissipation and introducing an adiabatic wall assumption
inside the hole, the flow is isentropic. At the choking point, all u uz !
e z ur !
er (10)
the flow variables can be obtained by combining the chocking
condition and the isentropic flow assumption: k2
meff m mt ; mt rCm (11)
3

g=g1

g1
1
2 2 P0 2
P2;s P0 ; T2;s T0 ; r2;s (1) 
2
2
2  2 
g1 g1 RT0 g 1 vuz vur vuz vur ur
G mt 2 2 2 (12)
s vz vr vz vr r


p 2g
u2;s c2;s gRT2;s RT0 (2)  
g1 vuz 2
szz meff 2  V$u (13)
vz 3
The flow velocity is sonic velocity at the choking point. The
mass flux can be determined with the flow variables calcu-  
vur 2
lated at the choking point: srr meff 2  V$u (14)
vr 3

g1
1
s
P0 2 2g  
_ s CD r2;s u2;s
m CD p (3) vuz vur
RT0 g 1 g1 szr srz meff (15)
vr vz
Leaving the hole, the flow experiences an abrupt expansion
 
as shown in Fig. 1. The abrupt expansion is completed in ur 2
sqq meff 2  V$u (16)
a very short distance (z10d), and it ends with a normal r 3
shock. The flow becomes incompressible after the normal The values of the model parameters are C13 1.44,
shock. If the entrainment of air is neglected during the abrupt C23 1.92, Cm 0.09, sk 1.0 and s3 1.3. The conservation
expansion, the hydrogen concentration follows the mean equations of energy and species equations are as follows:
axial concentration hyperbolic decay rule after the normal

  
 
shock: v 1v v m mt vh 1v m mt vh
ruz h rrur h r
s vz r vr vz Pr sh vz r vr Pr sh vr

1
1
g1=2g1 
 
kdps rair 2 kd rair 2 2 P0 1 2
XH2 CD (4) F  V$ r juj k u 17
z z0 rH2 z z0 rH2 g1 Pair 2

The derivation of Eq. (4) is described in Ref. [2]. In Eqs. 
  
 

v 1v v m mt vYi 1v m mt vYi
(1)e(4), z, z0, k, dps, g, CD are coordinates along the jet center- ruz Yi rrur Yi r
vz r vr vz Sc ss vz r vr Sc ss vr
line, virtual origin, decay coefficient, pseudo-diameter, ratio of (18)
specific heats and the discharge coefficient, respectively.
where:

2.2. Governing equations for flow, energy, and species X X ZT

h Yk hk Yk Cp;k TdT (19)
k k
The numerical problem is formulated by simultaneously Tref

solving for a high-pressure hydrogen reservoir, a choked

cracked hole and a hydrogen jet into the atmosphere. The v 1v
F uz szz ur szr ruz srz ur srr  (20)
axisymmetric continuity, momentum, turbulent kinetic vz r vr
3506 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2

Here, F denotes source terms of the energy equation due to consistency at the 4th probe, but they show some deviation at
dissipation work. In this study, both the turbulent Prandtl the 5th probe at a 9 m distance. The experimental data from
number sh and the turbulent Schmit number ss are taken to be d 1.0 mm show some deviation from the 4th probe, but not
0.9. The ideal gas law is used for the state equation. as severe as the d 0.5 mm case. The figure shows that the
downstream characteristics of the experiment are much more
susceptible to disturbances by wind, especially for smaller
3. Results and discussion hole diameters and lower source pressures.
Among a number of causes contributing to the deviation
3.1. Comparison between experiment and prediction far downstream, the following three causes appear to be
outstanding. First, the virtual wall introduced for numerical
KIST (Korea Institute of Science and Technology) measured stability may have contributed to the over-prediction of the
the concentration of a released hydrogen jet from a highly hydrogen concentrations in the numerical analysis. As the jet
pressurized chamber, representing a high pressure vessel. is being developed, the jet diameter becomes wide enough to
The hydrogen concentration was measured along the jet be affected by the virtual wall, which prevents the jet from
centerline for three cases of small leak hole diameters further expanding. Consequently, the numerical analysis
(d 0.5 mm, 0.7 mm and 1.0 mm). The hole diameters are would over-predict the hydrogen concentration profiles,
chosen in such a way that the hole area increases by particularly in the downstream region. Second, there is
approximately two for each diameter increase. For each measurement error associated with the alignment of the jet
diameter, the measurements are carried out for four cases of centerline. The sampling probes might not be placed exactly
chamber pressures (P0 100 bar, 200 bar, 300 bar and 400 bar) along the centerline of the jet exit. Third, because the exper-
using a gas sampling method. However, KIST failed in iment was performed outside, the effects of wind could not be
measuring the hydrogen concentration for d 1.0 mm and totally eliminated. Consequently, such a small discrepancy in
P0 400 bar because of the jet duration was too short to the centerline alignment and wind effect could have resulted
complete the measurement. The concentration of hydrogen in under-detecting the centerline hydrogen concentrations.
was measured at five locations along the hydrogen jet
centerline e 1.0 m, 3.0 m, 5.0 m, 7.0 m and 9.0 m. 3.2. Dispersion of hydrogen jet
The geometrical model for this study is depicted in Fig. 2. It
has a pressure-inlet boundary at a distance enough far from Fig. 4 shows (a) pressure distribution, (b) velocity profile, and
the leak hole. A fictitious slip wall is adopted for computa- (c) temperature profile for the hydrogen jet for the case with
tional stability. The fictitious wall should be carefully located d 1.0 mm and P0 200 bar. Fig. 4(a) shows that a significant
in two reasons. The first reason is that the fictitious wall pressure drop occurs only around the hole. The region around
should not disturb the main hydrogen jet flow by putting it too the hole can be divided into three zones e the hole inlet, the
close to the axis. The second reason is that computational inner region of the hole, and the exit of the hole. The hydrogen
stability cannot be achieved by locating it too far from the axis. flow experiences a large pressure drop in the inlet zone of the
Its location is different for different computational conditions. hole; the strongest pressure drop of the three. The pressure
In combination with the fictitious wall, an air intake inlet drops approximately 85 bar in this zone. In the second zone,
boundary is added to the computation because it is necessary the pressure drop continues to meet the chocked conditions
to supply fresh air into the main flow. Consequently, the just before the flow exits the hole. At the choking point, the
actual boundary surrounding the jet is divided into the two pressure is 85.0 bar, 20.3 bar less than the isentropic flow
boundaries. experiences because of the viscosity. The large pressure drop
Fig. 3 shows the measurement and prediction. The is completed in the exit zone by an abrupt expansion process,
prediction was achieved by using FLUENT [17]. Both the and its value is almost the same as the pressure drop inside
experimental and numerical results exhibit a gradual mono- the hole.
tonic decrease of the centerline hydrogen concentration with The flow inside the source reservoir remains almost static
excellent agreement until the 3rd probe at a 5 m distance from until it starts to accelerate near the inlet of the hole as
the exit for all diameters. Such an agreement indicates that depicted in Fig. 4(b). The average velocity magnitude of the
the experiment and the numerical simulation were carried pressure inlet boundary of the reservoir is only approximately
out in a physically reasonable manner. The experimental data 0.05 m/s but the flow is accelerated up to 200 m/s at the inlet of
for d 0.5 mm begin to show significant deviations from their the hole. The hydrogen flow continues to be accelerated after
numerical counterparts from the 4th probe at a 7 m distance. passing through the inlet of the hole. Then, the velocity
In the case of d 0.7 mm, both results still maintain magnitude reaches sonic velocity just before the flow leaves
the hole and the flow is chocked around the location of 0.95L.
When the high pressure flow is exposed to the ambient air, it
Air intake
Fictitious Wall (Slip Boundary) experiences abrupt expansion. This expansion accompanies
Inlet
Outlet a large decrease in the flow density so that flow accelerates
Reservior once more. The flow has a maximum velocity magnitude at
Inlet
x
a distance of 10d from the hole, and it accelerates up to a Mach
Axis
r number of M 5.7 at the end of the expansion.
The quick expansion of the flow ends with the normal
Fig. 2 e Geometric model used for the computation. shock. In Fig. 4(c), the normal shock is equal to the highly
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2 3507

100 100
d=0.5mm d=0.5mm
d=0.7mm d=0.7mm
d=1.0mm d=1.0mm
d=0.5mm d=0.5mm

Hydrogen Concentration

Hydrogen Concentration
d=0.7mm d=0.7mm
-1 d=1.0mm -1 d=1.0mm
10 10

-2 -2
10 10

-3 -3
10 10
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Distance(m) Distance(m)

(a) P0 =100 bar (b) P0 =200bar

0 0
10 10
d=0.5mm d=0.5mm
d=0.7mm d=0.7mm
d=1.0mm d=1.0mm
d=0.5mm d=0.5mm
Hydrogen Concentration

Hydrogen Concentration

d=0.7mm d=0.7mm
-1 d=1.0mm -1
10 10

-2 -2
10 10

-3 -3
10 10
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Distance(m) Distance(m)

(c) P0 =300 bar (d) P0 =400bar

Fig. 3 e Comparison between experiment and prediction. (a) P0 [ 100 bar, (b) P0 [ 200 bar, (c) P0 [ 300 bar and (d) P0 [ 400 bar.

clustered contour lines of temperature. The normal shock is is decreased to 208 K at the exit of the hole. After hydrogen is
located approximately 10d from the hole. The temperature, emitted from the hole, it experiences abrupt expansion, as
density, and pressure of the flow are drastically lowered by the discussed already. The temperature is decreased down to 60 K
expansion process, whereas the flow has hypersonic velocity. by this expansion process. This is the lowest temperature
The flow with a lowered density and hypersonic velocity observed in the flow and soon recovers to the ambient air
recognizes the ambient air as a strong non-moving body and temperature through the normal shock.
a normal shock is formed. After experiencing the normal Here, the dilution length of a specific hydrogen mole frac-
shock, the flow recovers and the temperature, density, and tion is introduced and is defined as the distance from the hole
pressure rise to that of the ambient air. to the point of the hydrogen mole fraction along the jet
The temperature of the hydrogen is kept at 293 K in almost centerline. In this study, the variables L1LFL, L1/2LFL, and L1/4LFL
the entire reservoir. The temperature of hydrogen begins to will be used for three typical dilution lengths. The variable
change after the flow enters the hole. Hydrogen is heated by 1LFL denotes the lean flammability limit which is 0.04,
viscous work up to 370 K inside the thermal boundary layer, expressed in mole fraction. The variables 1/2LFL and 1/4LFL
which is formed near the wall with the hole. However, the are 0.02 and 0.01, expressed in mole fraction. As seen in Fig. 5,
temperature of the flow outside the thermal boundary layer is the dilution length of a specific hydrogen mole fraction is the
continuously decreased by the acceleration of the fluid flow. longest distance from the hole to the point of a specific
The average temperature of flow at the chocking point is hydrogen mole fraction. Fig. 5 is the 2-D contour plot of the
approximately 245 K, which is slightly larger than the hydrogen mole fraction for d 0.5 mm. The incremental step
temperature of the isentropic flow (243 K). The temperature for the contour plot is 0.01 and the outmost contour line has
3508 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2

The dilution lengths for 1LFL, 1/2LFL, and 1/4LFL are plotted
in Fig. 6. In the case of d 1.0 mm, the data for 300 bar and
400 bar are not available because the jet lengths for 1/4LFL
2.31E5
exceed the domain of the experiment and the numerical
1.97E7
1.14E7
simulation. Both the experiment and the simulation give well
9.46E6 matched results for L1LFL and L1/2LFL, which is correlated with
the reasoning explained in Fig. 3. L1LFL>1=2LFL , dilution length
(a) Pressure
from 1LFL to 1/2LFL, is almost equal toL1LFL, which well agreed
with the result of hyperbolic decay. Because a point of 1/4LFL
was far beyond the concurrency region where experiment
1300 m/s
and simulation were in good agreement, the measured and
predicted L1/4LFL showed large deviation from each other.

3.3. Mass flux of released jet

(b) Velocity
When the reciprocal of hydrogen concentration is taken from
Eq. (4), the following line equation is obtained:
300K
12
rH2
1 rair z z  z z 
0 0
300K 370K s

g1=2g1 A (21)
60K XH2 2 p d d d d
CD Po =Pair
337K
g1
(c) Temperature Fig. 7 is the reciprocal plot of the measured data with
respect to z/d for P0 300 bar. As discussed above, the data of
Fig. 4 e Computational results for d [ 1.0 mm and the 1st, 2nd, and 3rd probes are well matched with the
P0 [ 200 bar. computational results. It can be said that the data of 1st, 2nd,
and 3rd probes were cross verified by the experiment and the
prediction. Then, only the data of 1st, 2nd, and 3rd probes
the value of 0.01. The criteria of setting the fictitious wall were used in plotting the figures. The line in each graph is the
boundary are determined based on these contour results. The best fitted line for the measurement data. Each graph again
diameter of the fictitious wall has at least two times larger shows the well matched behavior between the measurement
than maximum diameter of 0.01 contour line. and the prediction. The hyperbolic decay characteristics are

0.04 0.02 0.01

(a) P0=100bar

0.01
0.04 0.02

(b) P0=200bar

0.02 0.01
0.04

(c) P0=300bar

0.01
0.02
0.04

m
0 1 2 3 4 5 6 7 8 9 10

(d) P0=400bar

Fig. 5 e Hydrogen concentration contours for d [ 0.5 mm.

 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2 3509

16.0 16.0 16.0

14.0 14.0 14.0

12.0 12.0 12.0

Dilution length (m)

Dilution length (m)

Dilution length (m)

10.0 10.0 10.0

8.0 8.0 8.0

6.0 6.0 6.0

4.0 4.0 4.0

2.0 2.0 2.0

0.0 0.0 0.0

100 200 300 400 100 200 300 400 100 200 300 400
Pressure(bar) Pressure(bar) Pressure(bar)

(a) d=0.5 mm (b) d=0.7 mm (c) (c) d=1.0 mm

Fig. 6 e Dilution length for various hole diameters.

120 120
Experiment Experiment
Prediction(0.5mm) Prediction(0.7mm)
100 100

80 80
1/X H2

1/X H2

60 60

40 40

20 20

0 0
0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000
z/d z/d

(a) d=0.5 mm (b) d=0.7 mm

120
Experiment
Prediction(1.0mm)
100

80
1/X H2

60

40

20

0
0 2000 4000 6000 8000 10000
z/d

(c) d=1.0 mm
Fig. 7 e Reciprocal of hydrogen concentra for three holes with P0 [ 300 bar.
3510 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2

valid up to relatively large axial distances. The z/d of the 3rd prediction shows some deviation from that of the isentropic
probe for the three holes are 5,000, 7,000, and 10,000 in order of flow model. Because the coefficient A is inversely proportional
hole diameter. to the mass flux (the strength of the jet), this says that the
Eq. (21) tells that the coefficient A of the isentropic flow is mass flux of the actual flow is less than that of the isentropic
solely dependent on the source pressure. However, the gradi- flow. This is because of dissipation work, which the flow
ents of the three lines, A, are slightly different from another in experiences during passing through a hole. Fig. 4(c) clearly
Fig. 7. The same plots for various source pressures show rela- shows the temperature build-up near the wall of a hole by this
tively large differences in the gradients, as shown in Fig. 8, in dissipation work.
which each dashed line is obtained with the isentropic flow Two more cases of L, 10 mm and 20 mm, were tested to
assumption. These two figures reveal that the coefficient A of investigate the effect of dissipation work on the jet strength
the actual flow is also dependent on L, d, and P0. However, the and mass fluxes. When highly pressurized hydrogen flows
dependence on L and d is relatively small compared to that of through a pipe or a hose between a storage tank and a fueling
source pressure. All the coefficients e both measured and tip, its thickness is most likely less than 30 mm because of its
predicted e are listed in Table 1. small inner diameter. Then, a thickness less than 30 mm is
The last column in Table 1 is for the coefficient A of the more meaningful to take into account. Fig. 9 shows the mass
isentropic flow. The last column of each case is approximately fluxes for d 0.7 mm. In case of L 30 mm, the mass fluxes of
25% less than the others. This is due to viscous work existing P0 100 bar, 200 bar, 300 bar, 400 bar are reduced to 84.1%,
in the actual flow. Although the concentration of the numer- 84.8%, 85.3%, 85.4% of those of isentropic flow, respectively.
ical prediction fulfills the hyperbolic decay rule as well as that It is natural that the longer the hole length is, the larger the
of the measurement, the coefficient A of the numerical reduction is.

120 120
E x p e rim e n t E x p e rim e n t
C F D c a lc u la tio n C F D c a lc u la tio n
100 I s e n tro p ic 100 I s e n tro p ic

80 80

60 60
1 /X

1 /X

40 40

20 20

0 0
0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000
z /d z /d

(a) P =100 bar (b) P =200 bar

120 120
E x p e rim e n t E x p e rim e n t
C F D c a lc u la tio n C F D c a lc u la tio n
100 I s e n tro p ic 100 I s e n tro p ic

80 80

60 60
1 /X

1 /X

40 40

20 20

0 0
0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000
z /d z /d

(c) P =300 bar (d) P =400 bar

Fig. 8 e Reciprocal of hydrogen concentration for various source pressures with d [ 0.7 mm.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2 3511

Table 1 e Coefficient A (measurement, prediction), 310L3.

0.4

P0 d 0.5 mm d 0.7 mm d 1.0 mm Isentropic

(bar) 0.35 P0=100bar
P0=200bar
100 (11.34, 10.354) (10.069, 10.320) (10.778, 11.266) (e, 7.781) P0=300bar
200 (7.804, 7.379) (6.965, 7.380) (7.867, 8.257) (e, 5.502) 0.3 P0=400bar
300 (6.880, 6.113) (5.796, 6.100) (6.155, 6.378) (e, 4.492) Fitted line
400 (6.050, 5.235) (5.285, 5.266) (e, 5.617) (e, 3.891)

Reduction Ratio
0.25

At this point, the reduction ratio is introduced to investi- 0.2

gate the strength of the mass flux reduction of the actual flow
compared to that of the isentropic flow. The reduction ratio
(RR) is defined as: 0.15

_
mmass flux of actual flow
RR 1  : (22) 0.1
_ s mass flux of isentropic flow
m

As is expected, the larger the L/d ratio is, the larger the
reduction ratio is. Interestingly, the reduction ratio is not as 0.05
sensitive to the source pressure. The sensitivity to the source
pressure is almost zero in case of L/d 10. When the reduction 0
0 10 20 30 40 50 60 70
ratio is plotted with respect to L/d as shown in Fig. 10, it is
almost linearly dependent on L/d: L/d


L L Fig. 10 e Reduction ratio RR.
RR ya b : (23)
d d

When Eq. (22) is combined with Eq. (3) and Eq. (23), the mass
flux can be determined with the following equation: The variables a and b of the fitted line are 4.707  102 and


 2.361  103, respectively. The maximum error of the fitted
L line is 1.57% at an L/d 60, and the average error is 0.5%.
_
m 1  RR m _ s T0 ; P0
d

g1
1
s

P0 2 2g L
CD p 1ab (24)
RT0 g 1 g1 d 4. Conclusions

A 2D axisymmetric simulation was performed to investigate

25000 the characteristics of a hydrogen jet formed by high pressure
release through a small hole. The calculation was performed
I s e n tro p ic for the following experimental cases; a specific hole length
L = 10mm
(L 30 mm), three hole sizes (d 0.5 mm, 0.77 mm and
L = 20mm
20000 1.0 mm) and four chamber pressures (P0 100 bar, 200 bar,
L = 30mm
300 bar, and 400 bar). In addition to the experimental cases,
calculations were performed for two more hole lengths to
M a s s F lu x ( k g / m s )

analyze the mass flux characteristics through a small hole.

15000 This study introduced an air intake inlet and a fictitious
wall to ensure the accuracy and stability of the numerical
calculation. This approach gave a good agreement with the
experimental data until the 3rd probe. The measured and
predicted concentration fulfilled the hyperbolic decay char-
10000
acteristics in this concurrency region. However, the
measurement and prediction showed large deviation at two
farthest probe points from the release hole because of
centerline alignment inaccuracy of probes and wind effect.
5000
The dilution lengths of 1LFL (Lean Flammable Limit, 0.04)
and 1/2LFL were obtained for three hole diameters and four
release pressures The dilution length of 1/2LFL was almost
two times longer than that of 1LFL, which well agreed with the
0 result of hyperbolic decay curve. Because a point of 1/4LFL
100 200 300 400
was far beyond the concurrency region, the measured and
P re s s u re ( b a r)
predicted dilution lengths of 1/4LFL showed large deviation
Fig. 9 e Mass fluxes for d [ 0.7 mm. from each other.
3512 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 3 5 0 3 e3 5 1 2

Although actual hydrogen jet flow well fulfilled the tests for 40 MPa pressurized hydrogen. Int J Hydrogen Energy
hyperbolic decay characteristics as described in the Birchs 2007;32:2144e53.
work [2], its mass flux showed a large deviation from the [6] Schefer RW, Houf WG, Williams TC. Investigation of small-
scale unintended releases of hydrogen: buoyancy effects. Int
isentropic estimation of mass flux. This is due to dissipation
J Hydrogen Energy 2008;33:4702e12.
work which pure hydrogen flow experiences during passing [7] Schefer RW, Houf WG, Williams TC. Investigation of small-
through the hole. The mass reduction ratio, defined as the scale unintended releases of hydrogen: momentum-
ratio between actual mass flux and isentropic estimation, was dominated regime. Int J Hydrogen Energy 2008;33:6373e84.
found to be almost linearly dependent on L/d for the range of [8] Swain MR, Filoso P, Grilliot ES, Swain MN. Hydrogen leakage
source pressures and hole lengths of this study. Then, the into simple geometric enclosures. Int J Hydrogen Energy
2003;28:229e48.
mass flux of the actual hydrogen jet flow can be calculated
[9] Mohamed K, Paraschivoiu M. Real gas simulation of
with the linear relation of Eq. (24) within the maximum error
hydrogen release from a high-pressure chamber. Int J
of 1.57%. Hydrogen Energy 2005;30:903e12.
[10] Tchouveleva AV, Cheng Z, Agranat VM, Zhubrin SV.
Effectiveness of small barriers as means to reduce clearance
Acknowledgments distances. Int J Hydrogen Energy 2007;32:1409e15.
[11] Benard P, Tchouvelev A, Hourri A, Cheng Z, Angers B. High
pressure hydrogen jets in the presence of a surface. In:
This research was a part of the project titled Development of
Second international conference on hydrogen safety; 11e13
Technology for CO2 Marine Geological Storage funded by the September, 2007. San Sebastian, Spain.
Ministry of Land, Transport and Maritime Affairs, Korea. [12] Venetsanos AG, Baraldi D, Adams P, Heggem PS,
Wilkening H. CFD modelling of hydrogen release, dispersion
and combustion for automotive scenarios. J Loss Prev
references Process Ind 2008;21:162e84.
[13] Hourri A, Angers B, Benard P, Tchouvelev A, Agranat V.
Numerical investigation of the flammable extent of semi-
[1] International Energy Agency. Energy technology confined hydrogen and methane jets. Int J Hydrogen Energy
perspectives 2008: scenarios & strategies to 2050. Paris: 2011;36:2567e72.
OECD/IEA; 2008. [14] Salva JA, Tapia E, Iranzo A, Pino FJ, Cabrera J, Rosa F. Safety
[2] Birch AD, Brown DR, Dodson MG, Swaffield F. The structure study of a hydrogen leak in a fuel cell vehicle using
and concentration decay of high pressure jets of natural gas. computational fluid dynamics. Int J Hydrogen Energy 2012;
Combust Sci Technol 1984;36:249e61. 37:5299e306.
[3] Ruffin E, Mouilleau Y, Chaineaux J. Large scale characterization [15] Peneau F, Pedro G, Oshkai P, Benard P, Djilali N. Transient
of the concentration field of supercritical jets of hydrogen and supersonic release of hydrogen from a high pressure vessel:
methane. J Loss Prev Process Ind 1996;9:279e84. a computational analysis. Int J Hydrogen Energy 2009;34:
[4] Shirvill LC, Roberts P, Butler CJ, Roberts TA, Royle M. 5817e27.
Characterization of hazards from jet releases of hydrogen. [16] Khaksarfard R, Kameshki MR, Paraschivoiu M. Numerical
In: First international conference on hydrogen safety; 8e10 simulation of high pressure release and dispersion of
September, 2005. Pisa, Italy. hydrogen into air with real gas model. Shock Waves 2010;20:
[5] Takeno K, Okabayashia K, Kouchia A, Nonaka T, 205e16.
Hashiguchia K, Chitose K. Dispersion and explosion field [17] FLUENT. 6.3 users guide. Lebanon: Fluent Inc.; 2006.