You are on page 1of 8

Process Safety and Environmental Protection 151 (2021) 20–27

Contents lists available at ScienceDirect

Process Safety and Environmental Protection


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

Experimental study of liquefied gas dynamic leakage behavior from a


pressurized vessel
Xiyan Guo a , Wei Tan a , Liyan Liu a , Cenfan Liu b,∗ , Guorui Zhu a,∗
a
School of Chemical Engineering and Technology, Tianjin University, Tianjin, 300072, China
b
Key Laboratory of Special Equipment Safety and Energy-saving for SMR, CSEI, Beijing, 100029, China

a r t i c l e i n f o a b s t r a c t

Article history: The accidental releases of pressurized liquefied gases from tanks involve violent phase transformation,
Received 24 December 2020 which would generate two-phase releases and flashing jets. To investigate the evolution of leakage
Received in revised form 6 April 2021 behaviors and make an analysis of near-field jet flow characteristics, a small-scale liquefied gas release
Accepted 4 May 2021
experiment has been established. Leakage holes with different length-diameter ratio (LDR) have been
Available online 12 May 2021
used to analyze interactions between the leakage holes and release behaviors. Morphological charac-
teristics of jet expansion angles have been obtained by high-speed camera, and jet velocities have been
Keywords:
measured by particle image velocimetry (PIV). Meanwhile, the depressurization process, variation of
Small scale release experiment
Flashing jet
temperature in the tank and mass outflow rates were obtained. Results show that, despite the LDR is
Two-phase flow varying, expansion angles and jet velocities behavior in the same tendency: decreases in the initial and
Model validation maintains in a stable value for a period. The stable velocity status was worked by the balanced pressure
R134a drop. Thereafter, an empirical two-phase mass outflow rate model is developed based on the experi-
mental data, which is related to the nozzle geometric parameter and upstream pressure, 90 % of the
experimental data are within ±12 % of the prediction. Therefore, the empirical two-phase model can be
supported in the mass flow rate evaluation, especially for the release cases that LDR are smaller than 5.00
but larger than 2.00. The experimental data is beneficial for providing further insight into the prediction
of accidental release and risk assessment for the liquefied gas transportation and storage.
© 2021 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction phase, and two-phase flashing jets will generate. A violent pressure
drop may release enormous evaporated gas and energy, which will
To address issues of oil over-consumption, greenhouse gas emis- cause an accidental two-phase release. It has a destructive effect
sions, and pollutant emissions, green and clean fossil fuels such as on the vessel, even leads to huge economic losses and catastrophic
liquefied natural gas (LNG) have been proposed widely (Horvat, consequences to surroundings (Stawczyk, 2003).
2018; Zhu et al., 2020a,b; Chen et al., 2004; Zhang et al., 2020; Dan To evaluate the degree of accident leakage, factors such as
and Lee, 2014). For convenient transportation and storage, these mass outflow rate, physico-chemical properties, flammability and
liquefied gases are stored as liquid phase in vessels under suffi- toxicity are often studied (Stawczyk, 2003). Since the mass out-
ciently high pressure. However, if such a vessel is damaged due to flow rate is a vital element in the prediction of accidental hazard
flawed material, fatigue, corrosion and so on, cracks or holes may analysis (Woodward, 2009), various theoretical models have been
develop (Deaves et al., 2001; Bellegoni et al., 2021). As results of a researched to gain more accurate prediction results. Based on
failure of vessel with saturated liquid, a sharp decrease of pressure the conservation of mechanical energy, the models presented by
and shift of the thermodynamic balance would cause instant boiling Daniel and Joseph (1990) are widely used (Johnson and Cornwell,
of some portion of the liquid phase (Galeev et al., 2013). Mean- 2007; Zhu et al., 2020a,b). Besides, these models provide theoret-
while, the saturated liquids will vaporize instantaneously when ical basises for the latter modified models (Zhu, 2014; Li et al.,
discharge to the ambient. The vapor can be released with the liquid 2016; Luo et al., 2006). Khajehnajafi and Shinde (1994) presumed
the transient leakage was a quasi-steady process, and developed
a theoretical model to calculate the mass outflow rate, which can
be applied in the tank-pipe system. Considering depressurization
∗ Corresponding authors.
and phase change was caused by flashing, Britter et al. (2011) pre-
E-mail addresses: liucenfan288@163.com (C. Liu), zhuguorui@tju.edu.cn
sented a model which can evaluate the mass outflow rate of gas,
(G. Zhu).

https://doi.org/10.1016/j.psep.2021.05.005
0957-5820/© 2021 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
X. Guo et al. Process Safety and Environmental Protection 151 (2021) 20–27

flashing spray. With high volatility and low saturation temperature,


Nomenclature the mechanism of the flashing spray with volatile cryogen can be
studied. Wang et al. (2017) measured the spray cone angle, droplet
L Nozzle length (m) velocity, and temperature of R134a by using the transparent nozzle.
D Nozzle diameter (m) Relations between the internal flow patterns and characteristics of
LDR Length-diameter ratio (-) R134a flashing spray were investigated.
Q Mass outflow rate (kg/s) Experimental studies on the pressurized liquefied gases releas-
Cd Discharge coefficient (-) ing from vessels are increasing. However, studies on the relations
A Area of the leakage nozzle (m2 ) between the thermodynamic state in the tank and flow character-
 Mixed density (kg/m3 ) istics outside are rare. The aim of this work is to investigate the
l Liquid density (kg/m3 ) interactions between the internal thermodynamics and the evo-
v Vapor density (kg/m3 ) lution of outside jet flow. In this study, a small-scale liquefied
P0 Tank pressure (Pa) gas leakage experiment has been established with leakage holes
Pc Critical pressure (Pa) in different sizes. Evolutions of jet expansion angles have been
Patm Atmospheric pressure (Pa) obtained by high-speed camera, and jet velocities have been mea-
Fv Percentage of vapor (-) sured by particle image velocimetry (PIV). The depressurization
Cp Specific heat (kJ/kg·K) process, variation of temperature in the tank and mass outflow rate
T0 Temperature (K) were obtained. Thereafter, an empirical two-phase model, which is
Tc Critical temperature (K) related to the geometric parameter and the upstream pressure is
H Latent heat of vaporization (kJ/kg) developed. The experimental data could be used for safety assess-
 Specific heat ratio (-) ment and theoretical model validation in the liquefied gas release.
M Molar mass (kg/mol)
R Gas constant (J/K/mol)
2. Experiments

2.1. Experimental setup


two-phase and liquid release. The model is suitable for the various-
type ruptures including sharp-edged and “pipe-like” types. Based
Combing the security and environmental protection, R134a was
on the van der Waals equation of state, Zhou et al. (2018a,b) built
chosen as working fluid in the experiment. With the characteristics
a high-pressure gas leakage model to calculate the gas state prop-
of non-toxic, non-reactive, non-flammable, and non-ozone deplet-
erties, flow parameters and its subsequent jet flame height in the
ing (Mustapa et al., 2011), R134a has been widely used (Ohm et al.,
hydrogen release.
2018). Besides, R134a has a high volatility and at a boiling point of
To improve the accuracies of mass outflow rate, many
-26.2 ◦ C under the atmospheric pressure, which can behavior flash-
researchers have conducted near-source region experiments to
ing process well. Therefore, it could be used as a substitution of
obtain more essential release parameters. A laboratory scales
liquefied gas. Schematic of the experimental setup has been shown
super-critical carbon dioxide pipelines leakage experiment has
as Fig. 1. The experimental system was constituted by three sub-
been conducted by Li et al. (2016). The mass outflow rates of
systems: (1) leakage system; (2) 2D-PIV system; and (3) recovery
leakage holes with different sizes are obtained. Compared with
system. The subsystems will be introduced in the following sec-
two typical accidental gas release mathematical models, a mod-
tions.
ified model has been addressed. Likewise, a near-field carbon
dioxide vessel leakage experiment was carried out by Teng et al.
(2018). The under-expanded jet structure, jet velocity and other 2.1.1. Leakage system
parameters were measured by varying the release orifices. There- Leakage system mainly contains five parts, R134a gas cylinder,
after, the relations between the under-expanded jet structure and filling machine, leakage tank, sensors and release nozzle. Liquid
thermo-hydraulics characteristics were analyzed. Tian et al. (2021) R134a was provided by R134a gas cylinder. With the filling machine
also conducted a near-filed region carbon dioxide release experi- (CM200), liquid R134a can be filled into the leakage tank without
ment, and the transient pressure, mass flow rate, wall temperature, backflow. The leakage tank was in a cubage of 2000 mL, which the
external jet structure and phase transition were studied. Mocellin working temperature was -20−200 ◦ C, and in a working pressure
et al. (2018) conducted an experimental campaign performed on range of 0−5 MPa. Two pressure sensors (CYB-20S-DT), in the mea-
carbon dioxide releases to investigate useful parameters used in surement range of 0−5 MPa, were installed in the upper region in
the modeling procedures. Thereafter, a source model for multi- the tank. Temperature sensor (SBWZ-P) in the measurement range
phase CO2 release is illustrated. Zhu et al. (2019) conducted a of -20−50 ◦ C, was fixed vertically in the tank. Besides, weight sensor
laboratory scale superheated liquid release experiment by water, (CYB-602ST) in the measurement range of 0−30 kg, was installed
pressure and temperature inside the tank were measured with below of leakage tank. Response frequencies of all of the sensors
different operating conditions. The characteristics of flashing jets were 10 kHz, and in the accuracies of 0.3 % FS.
were obtained by measuring the evolution of droplet diameters Release nozzle was the core part of the system. According to the
and three-dimensional velocities. However, since liquefied gases risk based inspection (RBI) (API581−2008) (American Petroleum
are also highly pressurized as carbon dioxide or water but their Institute, 2000), five different hole sizes were used in this study.
properties are quite different. The parameters for measuring the Ratio of hole length to hole diameter (LDR) was chosen to charac-
accidental release of carbon dioxide or water may not be suitable terize the leakage nozzle, which is defined as:
for liquefied gases. Besides, to understand the evolution of release
and dispersion better, it’s of great significances to study the rela- LDR = L/D
tions between the internal thermal state and external flashing jet.
Therefore, studies on the release and dispersion of pressurized Where, L is the hole length. For all of the holes in the test, hole
liquefied gases are also ongoing. Based on the open source tool length is 4.00 mm, which is equal to the shell thickness of release
OpenFOAM-2.4.0, Zhou et al. (2018a,b) conducted a systematically tank. D is the hole diameter. The details of leakage nozzle are shown
numerical study on the spray and thermal characteristics of R404A as Table 1.

21
X. Guo et al. Process Safety and Environmental Protection 151 (2021) 20–27

Fig. 1. Schematic of the experimental setup.

Table 1 stable, 1000 g liquid R134a were filled into the leakage tank. Open-
Geometry parameters of leakage nozzle.
ing the solenoid valve, the release would begin. When the digital
D (mm) 3.00 2.00 1.50 1.00 0.80 data of tank pressure attained to 0.01 MPa or the data of mass was
LDR 1.33 2.00 2.67 4.00 5.00 constant, the release was completed.

3. Mass outflow rate model


2.1.2. 2D-PIV system
To obtain the flow flied near the leakage nozzle, a two- In a leakage accident, there are several different release stages
dimension particle image velocimetry (2D-PIV) system if the liquid level is higher than the leakage nozzle. The evolution
(TM4−5M380) has been established. In this study, particles of release in the storage tank is shown as Fig. 2, it depicted that the
were the R134a liquid droplets. Double-pulsed Nd:YAG laser, whole release process can be divided into five stages: before the
which was in a 532 nm wavelength and 5 Hz operation frequency, release, the liquid level was higher than the release nozzle; as leak-
was used to illuminate the flow flied by a 200 mJ/pulse. Capturing age began, internal liquid released from the tank and liquid level
the light scattered from the particles by a CCD camera (CLB- went down gradually, namely stage I; when leakage proceeded, the
B2520 M), which was in a resolution of 2456 × 2056 pixels, 16 Hz liquid level was close to the height of leakage nozzle, the leakage
data acquisition rates, 14 bit gray-level outputs. The cooperation medium changed to liquid and gas phase, namely transition stage;
between laser and CCD camera was controlled by a synchronous when the liquid level was below the nozzle, the leakage medium
controller (MicroPulse725 V5). The interval time between two transformed into single gas phase, namely stage II; finally, the leak-
laser light pulses, dt, was in a great significance for acquiring age remained in gas release until completed. In the following, the
accurate flow field vector information during the image capture release process of stage I and II will be analyzed in details.
(Liu et al., 2019). Setting the interval time as 0.002 ms, and the
image magnification is 0.03278 mm/pixel. A 32 × 32 pixels inter-
3.1. Two phase release model
rogation windows size was used as the interrogation windows
during the particles capturing and recording. Simultaneously, a
In a leakage accident, mass outflow rate is a significant factor to
high-speed camera (FASTCAM) with the resolution of 1024 × 1024
evaluate the severity. Therefore, selecting an appropriate numerical
pixels, 20000 fps data acquisition rates was used to capture the
model is crucial. In the following, two different mass outflow rate
morphology of the flashing. MicroVec and Tecplot software were
models are analyzed based on the stage I and II mentioned in the
employed for data collection and particle image post-processing.
Fig. 2.
Compared with single liquid or gas leakage, the release case of
2.1.3. Recovery system
liquefied gas in the stage I is more complicated, which involves
To eliminate the interferential background light, the experi-
phase transition. Thus, it needs to be considered as two-phase
ment was conducted in a darkroom. Considering that discharging
release. From Chen et al. (2002), the mass outflow rate can be
too much R134a continuously might dilute the fresh air indoors, a
calculated by Eq. (1):
recovery system was set behind the leakage system.

For stage I : Q = Cd · A · 2 (P0 − P) (1)
2.2. Experimental procedure
where, Cd denotes discharge coefficient, which is 0.8; A denotes
The experiments were conducted under an ambient room tem- the area of the leakage nozzle, m2 ;  denotes the density of leaked
perature of 30 ◦ C and an atmospheric humidity of approximately medium, kg/m3 ; P0 denotes the tank pressure, Pa; Pc denotes the
70 %. All operation processes were in the same conditions except critical pressure, Pa, which can be calculated by Pc = 0.55P0 .
the LDR of leakage nozzle. Guaranteeing that the tank was tight and Since the violent pressure drop, pressurized liquefied gas will
all the indicators were on work, the test would start. After exam- flash and generate liquid-gas phase. Therefore, densities of released
ining the data acquisition of PIV system and leakage system were fluids need to be considered as the mixed density of liquid and gas

22
X. Guo et al. Process Safety and Environmental Protection 151 (2021) 20–27

Fig. 2. Evolution of leakage in the tank.

phase. The energy of flashing comes from fluid itself. Once the atmo- where, P0 denotes the tank pressure, Pa; Patm denotes the atmo-
spheric pressure and temperature are constant, the percentage of spheric pressure, Pa;  denotes the specific heat ratio of the vapor.
flashing is decided by the tank temperature. Percentage of vapor Fv 
can be calculated by Eq. (2). ·M
 2  +1
−1
Foe stage II : Q = Cd · A · P0 · (5)
Cp (T0 − Tc ) R · T0 +1
Fv = (2)
H
where, Cd denotes the discharge coefficient, which is 1.0 for circle;
where, Cp denotes the specific heat of the liquid, kJ/kg · K; T0 M denotes the molar mass of R134a vapor, 0.12 kg/mol; R denotes
denotes the temperature of the liquid, K; Tc denotes the critical the gas constant, 8.314 J/K/mol and T0 denotes the vapor tempera-
temperature under the critical pressure, K; which can be calcu- ture, K.
lated by property software (NIST-REFPROP software); H denotes If the ratio of tank pressure to atmospheric pressure satisfies Eq.
the liquid latent heat of vaporization, kJ/kg. (6), it is in the subsonic flow phases, and the mass outflow rate can
Thus, the uniformed two-phase density can be calculated by Eq. be calculated by Eq. (7).
(3).
P0
 2  −1

1 > (6)
= (3) Patm +1
Fv (1−Fv )
+ 
 2  
v l
   −1
Where, l denotes the density of liquid, kg/m3 ; v denotes the 2 · M Patm Patm 
For stage II : Q = Cd · A · P0 · 1− (7)
density of vapor, kg/m3 . ( − 1) R · T0 P0 P0

Combining simultaneous equation from Eqs. (1) to (3), the mass


outflow rate of R134a in the stage I can be calculated under given Combining simultaneous equation from Eqs. (4) to (7), the mass
upstream parameters and nozzle diameters. outflow rate of vapor in the stage II can be calculated.

3.2. Gas phase release model 4. Results and discussion

In the stage II, the leakage medium is single vapor. It’s of great 4.1. Evolution of pressure and temperature in the tank
significance to estimate the flow states, namely, the sonic and the
subsonic flow phases of vapor. If the ratio of tank pressure to atmo- In the leakage accidents, liquefied gases will jet into the
spheric pressure satisfies Eq. (4) (Xie and Xiang, 2020), it is in the atmosphere sharply due to pressure drops. Thus, variations of
sonic flow phase, and the mass outflow rate can be calculated by depressurization in the tank have significant effects to the severity
Eq. (4). of the release. The evolution of tank pressure with different leakage
 2  −1
 nozzle is shown as Fig. 3. From Fig. 3, given a release time, the tank
P0 pressure drops faster as the LDR decreases. However, the tenden-
≤ (4)
Patm +1 cies of pressure drops are in the same, which are compose of three

23
X. Guo et al. Process Safety and Environmental Protection 151 (2021) 20–27

As the release proceeds, the liquid level decreases and goes below
to the height of sensor. At this time, the temperature undergoes a
rapid decrease due to the transformation from the liquid phase to
the vapor phase, the sensor records the temperature of the vapor.
At this stage, the vapor temperature decrease due to pressure drop.

4.2. Evolution of flashing jet

For leakage nozzles with different LDR, the phase transforma-


tion process is in the same tendency (Zhu et al., 2019). Therefore, the
case of nozzle with a 4.00 LDR was chosen as an example to analyze
characteristics of flashing jet in this study. The energy of pressur-
ized liquefied gas will transfer into the kinetic energy mostly when
it discharges to the atmosphere, then a high-speed jet will gener-
ate. Due to Joule-Thomson (J–T) effect, the volume of liquid will
expand and cause a violent temperature drop. When the tempera-
ture decreases to the condensation point, frost tiers will accumulate
in the leakage nozzle. The process of phase change in the stage
Fig. 3. Evolution of tank pressures for different leakage nozzles. I is more obvious than other two stages, and the flashing jet can
be observed more distinctly. Thus, the evolutions of jet expansion
angles and velocities in the stage I were analyzed.

4.2.1. Jet expansion angle


Jet expansion angle is defined as the angle between two tangent
lines of distinct plume boundaries (Doungthip et al., 2002), shown
as  in Fig. 5. From the evolution of jet expansion angle in the stage I,
expansion angle increases in the initial and maintains in a relatively
stable value, then decreases in the final. In the stage I-1, due to
combined actions of evaporation heat effects and pressure drops,
the liquids are broken and expansion angles are generated. In the
stage I-2, influenced by the relatively stable pressure drops, there
is little change for the value of expansion angle and it keeps at
11◦ approximately. As the liquid level closes to the leakage nozzle
gradually, the residual liquid in the tank can’t support great release
intensity and the angle begins to decrease.

4.2.2. Jet velocity


To obtain the evolution of jet velocity, the flow field closing to
Fig. 4. Evolution of temperature in the tank for different leakage nozzles. the leakage nozzle was measured by PIV. Due to the maximum field
of view limitations and desired high spatial resolution, the region
where was 60 mm away from the leakage nozzle was determined
parts mainly. In the first part P1 , the pressure drops slowly, but as tested flow field. The evolution of jet velocity depends on time
decays rapidly in the second part P2 and third part P3 . The reason is shown as Fig. 6. From Fig. 6, an interesting phenomenon can be
of this is that tank pressure decreases sharply when the leakage observed: an obvious breach exists in the front 2 mm of X-axis. This
begins, which makes the liquid in the tank evaporate instanta- phenomenon steams from freeze of frosts. Much frosts stick in the
neously. With the generated vapor, the depressurization caused nozzle and move with high-speed jet, which blocks partial laser
by release is balanced and the rates of pressure drops are slowly. light and velocities in these regions are missing. From the contours
As release proceeds, the fluid is decreasing and vaporization rate of velocity, decrement rate in the central region has attained to
can’t support the pressure difference. Thus, the tank pressure is 43.53 % in the stage I-1. Then, the velocity maintains in a relatively
controlled by pressure drops again. When reaching to the vapor stable status in the stage I-2. At stage I-2, pressure drops caused
release stage, reduction rate of pressure increases again. by release is balanced by the flashed vapor, which makes the sta-
Evolution of temperature in the tank for different leakage noz- ble velocity. Selecting the point 30 mm away from the nozzle in
zles is shown as Fig. 4. From Fig. 4, reduction rates of temperature the center line of jet as an analysis point, velocity in the balanced
are related to the LDR of leakage nozzle: with an increasing LDR, pressure stage was 25 m/s, and the max jet velocity was 42 m/s.
the reduction rates are going up. However, with the tempera-
ture variation curve, an interesting phenomenon can be observed. 4.3. Mass outflow rate model
The temperature increases slightly at the beginning of the release,
which is approximately 1 ◦ C. Then, the temperature decreases grad- Selecting upstream thermal parameters under different leakage
ually and goes to be equilibrium. The reason for the temperature nozzles as inputs, the mass outflow rates calculated by “two-phase
goes up a bit in the initial is caused by the slight thermal stratifica- model” and “vapor model” are shown as Fig. 7. Results show that
tion in the liquid R134a. Due to the violent pressure drop, flashing the data calculated by “two-phase model” in the stage I is less than
and evaporation occur in the tank at the moment release begins. the experiment data. However, “vapor model” can illustrate the
The initial calm liquid level was destroyed and changed the den- mass outflow rate suitably in the stage II. Therefore, the “two-phase
sity of the liquid in the upper level. Then, there is slight thermal model” in the stage I will be focused in the following.
stratification in the liquid R134a. As hotter R134a moved down- In the two-phase model, values of discharge coefficient and crit-
ward, resulting in temperature increase of the temperature sensor. ical pressure in the Eq. (1) have significant effects to the mass

24
X. Guo et al. Process Safety and Environmental Protection 151 (2021) 20–27

Fig. 5. The evolution of jet expansion angle in the stage I (LDR = 4.00).

Fig. 6. The evolution of jet velocity in the stage I (LDR = 4.00).

outflow rate. The flow rate ratio of actual to ideal is expressed by upstream parameters and nozzle LDR. Mass outflow rates calcu-
discharge coefficient Cd , which is related to the upstream pressure lated by empirical two-phase model are shown as Fig. 7. Results
and geometric parameters of leakage nozzle. In the liquid model, show that “empirical two-phase model” in the stage I can illus-
the value of Cd is simplified (Chen et al., 2009), but it is for the trate the time-dependent mass outflow rate suitably. Comparison
incompressible flow. For the liquefied gas leakage, since the easy- between the experiment and empirical two phase model is shown
flashing properties, incompressible assumption is not practicable. in Fig. 8. The results show that 90 % of the experimental data are
Correspondingly, the value of critical pressure Pc is influenced by within ±12 % of the prediction. However, for the nozzle which LDR
geometric parameters, especially the length of nozzles. Given a is 1.33, 30 % of the tested data are out of prediction, the main reason
diameter, pressure losses will decrease for the shorter leakage is, for the nozzle that LDR is null, it’s very fast for liquid R134a to
length. In this study, although the hole length was constant, the flow through the nozzle. In this sense, there is not enough time for
diameter was varying, and the LDR was from 1.33 to 5. There- liquid to flash in the nozzle, and the release can be considered liq-
fore, fluids release to the ambient with an in-develop fully flowing uid. Therefore, the liquid model can be used to calculate the mass
(Ward-Smith, 1979), and relations between the critical pressure outflow rate if the LDR is less than 2.00 in our experiment. The
and specific LDR need to be investigated. It’s essential to estab- empirical two-phase model should be used with caution for con-
lish a supplement model in the release cases that fluid flowing ditions outside the present range of variables (Fan et al., 2018).
in-developed. Therefore, an empirical two phase model for release Overall, “empirical two-phase model” can predict the mass out-
cases that LDR is smaller than 12 was investigated. flow rate well when the LDR of release hole is smaller than 5.00
The non-dimensional parameter LDR = DL is used to evaluate but larger than 2.00, which can be supported in the prediction of
the effect of geometric parameters. Using a non-linear curve fitting leakage accidents and theoretical model validation.
method, relationships between the discharge coefficient and criti-
cal pressure can be obtained as Eq. (8), and relationships between
the critical pressure Pc and LDR of leakage nozzle are Eq. (9). 5. Concluding remarks

Cd = 0.273+0.968 · P01.333 (8)


A small scale liquefied gas release experiment has been con-
Pc
L ducted with a 2000 mL storage tank and five leakage holes with
= 0.068 + 0.194 · ln (9) different length-diameter ratio (LDR). With R134a as experimental
P0 D
fluid, morphology evolutions of R134a flashing jet have been inves-
Combining the Eqa. (8) and (9) into the Eq. (1), the mass out- tigated with a high-speed camera and particle image velocimetry
flow rate of R134a in the stage I can be calculated under given (PIV). Thermodynamic state in the tank has been monitored, includ-

25
X. Guo et al. Process Safety and Environmental Protection 151 (2021) 20–27

Fig. 7. Mass outflow rate in the experiment and the calculation results from the “empirical two phase leakage model” and “vapor model”.

26
X. Guo et al. Process Safety and Environmental Protection 151 (2021) 20–27

Daniel, A.C., Joseph, F.L., 1990. Chemical Process Safety Fundamentals With Appli-
cation. Prentice-Hall, New Jersey, pp. 109–113.
Deaves, D.M., Gilham, S., Mitchell, B.H., et al., 2001. Modelling of catastrophic
flashing releases. J. Hazard. Mater. 88, 1–32, http://dx.doi.org/10.1016/S0304-
3894(01)00284-9.
Doungthip, T., Ervin, J.S., Williams, T.F., et al., 2002. Studies of injection of jet fuel at
supercritical conditions. Ind. Eng. Chem. Res. 41, 5856–5866, http://dx.doi.org/
10.1021/ie0109915.
Fan, X., Wang, Y., Zhou, Y., et al., 2018. Experimental study of supercritical CO2 leak-
age behavior from pressurized vessels. Energy 150 (May 1), 342–350, http://dx.
doi.org/10.1016/j.energy.2018.02.147.
Galeev, A.D., Starovoytova, E.V., Ponikarov, S.I., 2013. Numerical simulation of the
consequences of liquefied ammonia instantaneous release using FLUENT soft-
ware. Process Saf. Environ. Prot. 91, 191–201, http://dx.doi.org/10.1016/j.psep.
2012.05.002.
Horvat, A., 2018. CFD methodology for simulation of LNG spills and rapid phase
transition (RPT). Process Saf. Environ. Prot. 120, 358–369, http://dx.doi.org/10.
1016/j.psep.2018.09.025.
Johnson, D.W., Cornwell, J.B., 2007. Modeling the release, spreading, and burning of
LNG, LPG, and gasoline on water. J. Hazard. Mater. 140, 535–540, http://dx.doi.
org/10.1016/j.jhazmat.2006.10.022.
Khajehnajafi, S., Shinde, A., 1994. Prediction of discharge rate from pressurized vessel
blowdown through sheared pipe. Process Saf. Prog. 13, 75–82, http://dx.doi.org/
10.1002/prs.680130210.
Fig. 8. Comparison between experiment and empirical two phase model.
Li, K., Zhou, X., Tu, R., et al., 2016. An experimental investigation of supercritical CO2
accidental release from a pressurized pipeline. J. Supercrit. Fluids 107, 298–306,
http://dx.doi.org/10.1016/j.supflu.2015.09.024.
ing depressurization process, variation of temperature in the tank
Liu, Y.P., Wang, X.S., Zhu, P., et al., 2019. Experimental study on gas jet suppressed by
and mass outflow rates. The results are as follows: water mist: a clean control technique in natural gas leakage incidents. J. Clean.
Results show that, despite the LDR is varying, expansion angles Prod. 223, 163–175, http://dx.doi.org/10.1016/j.jclepro.2019.03.107.
Luo, J.H., Zheng, M., Zhao, X.W., et al., 2006. Simplified expression for estimating
and jet velocities behavior in the same tendency: decreases in the
release rate of hazardous gas from a hole on high-pressure pipelines. J. Loss
initial and maintains in a stable value for a period. The stable veloc- Prev. Process Ind. 19, 362–366, http://dx.doi.org/10.1016/j.jlp.2005.06.029.
ity status was worked by the balanced pressure drop. Thereafter, Mocellin, P., Vianello, C., Salzano, E., Maschio, G., 2018. Pressurized CO2 releases in
the applications of a typical two phase mass outflow rates predic- the framework of carbon sequestration and enhanced oil recovery safety analy-
sis: experiments and model. Process Saf. Environ. Prot. 116, 433–449, http://dx.
tion model and a vapor model have been discussed based on the doi.org/10.1016/j.psep.2018.03.018.
experimental data. Results show that the data calculated by typi- Mustapa, A.N., Manan, Z.A., Azizi, C.Y.M., et al., 2011. Extraction of ␤-carotenes from
cal two phase model is less than the experiment data but “vapor palm oil mesocarp using sub-critical R134a. Food Chem. 125, 262–267, http://
dx.doi.org/10.1016/j.foodchem.2010.08.042.
model” can illustrate the mass outflow rate suitably. Subsequently, Ohm, T.I., Chae, J.S., Zhang, M.Y., Moon, S.H., 2018. Effect of burner types and steam
an empirical two-phase model related to the length-diameter ratio injection methods on thermal destruction of waste refrigerants (hfc-134a). Pro-
(LDR) of leakage nozzle and the upstream pressure has been estab- cess Saf. Environ. Prot. 121, http://dx.doi.org/10.1016/j.psep.2018.10.001.
Stawczyk, J., 2003. Experimental evaluation of LPG tank explosion hazards. J. Hazard.
lished. In the empirical two-phase model, 90 % of the experimental Mater. 96, 189–200, http://dx.doi.org/10.1016/S0304-3894(02)00198-X.
data are within ±12 % of the prediction. Therefore, for release case Teng, L., Li, Y.X., Hu, Q.H., et al., 2018. Experimental study of near-field structure and
that LDR is less than 5.00 the empirical two-phase model can be thermo-hydraulics of supercritical CO2 releases. Energy 157, 806–814, http://
dx.doi.org/10.1016/j.energy.2018.04.195.
supported for the theoretical model validation. However, if the LDR
Tian, G.Y., Zhou, Y., Huang, Y.P., et al., 2021. Experimental study of accidental release
is decrease to 2.00, it’s recommended to evaluate the mass outflow behavior of high-pressurized CO2 vessel. Process Saf. Environ. Prot. 145, 83–93,
rate with liquid model. http://dx.doi.org/10.1016/j.psep.2020.07.051.
Wang, X.S., Chen, B., Wang, R., et al., 2017. Experimental study on the relation
between internal flow and flashing spray characteristics of R134a using straight
Declaration of Competing Interest tube nozzles. Int. J. Heat Mass Transf. 115, 524–536, http://dx.doi.org/10.1016/
j.ijheatmasstransfer.2017.08.040.
Ward-Smith, A.J., 1979. Critical flowmetering: the characteristics of cylindrical noz-
The authors report no declarations of interest. zles with sharp upstream edges. Int. J. Heat Fluid Flow 1 (3), 123–132, http://dx.
doi.org/10.1016/0142-727X(79)90028-6.
Acknowledgement Woodward, J.L., 2009. Validation of two models for discharge rate. J. Hazard. Mater.
170 (1), 219–229, http://dx.doi.org/10.1016/j.jhazmat.2009.04.126.
Xie, Q.Y., Xiang, L., et al., 2020. Mutual effects between dynamic leakage behavior
The authors are grateful for Internal Project of China Special and the pressure/temperature in a LNG tank with external heat fluxes. J. Loss
Equipment Inspection and Research Institute (2019Youth12). Prev. Process Ind. 63, 104029, http://dx.doi.org/10.1016/j.jlp.2019.104029.
Zhang, Y.X., Zhu, J.L., Peng, Y.M., et al., 2020. Experimental investigation of LNG
release underwater and combustion behavior under crosswinds. Process Saf.
References Environ. Prot. 134, 239–246, http://dx.doi.org/10.1016/j.psep.2019.12.019.
Zhou, K., Liu, J., Wang, Y., et al., 2018a. Prediction of state property, flow parameter
American Petroleum Institute, 2000. Risk Based Resource Document API., pp. 581. and jet flame size during transient releases from hydrogen storage systems. Int.
Bellegoni, M., Ovidi, F., Landucci, G., Tognotti, L., Galletti, C., 2021. CFD analysis of the J. Hydrog. Energy. 43, 12565–12573, http://dx.doi.org/10.1016/j.ijhydene.2018.
influence of a perimeter wall on the natural gas dispersion from an LNG pool. 04.141.
Process. Saf. Environ. Prot. 148, 751–764, http://dx.doi.org/10.1016/j.psep.2021. Zhou, Z.F., Lu, G.Y., Chen, B., 2018b. Numerical study on the spray and thermal char-
01.048. acteristics of R404A flashing spray using OpenFOAM. Int. J. Heat Mass Transf.
Britter, R., Weil, J., Leung, J., et al., 2011. Toxic industrial chemical (TIC) source emis- 117, 1312–1321, http://dx.doi.org/10.1016/j.ijheatmasstransfer.2017.10.095.
sions modeling for pressurized liquefied gases. Atmos. Environ. 45, 1–25, http:// Zhu, D.Z., 2014. Example of simulating analysis on LNG leakage and dispersion.
dx.doi.org/10.1016/j.atmosenv.2010.09.021. Procedia. Eng. 71, 220–229, http://dx.doi.org/10.1016/j.proeng.2014.04.032.
Chen, T.K., Xu, J.L., Luo, Y.S., 2002. Experiment study on two-phase critical flow. J. Zhu, X.L., Song, Z., Pan, X., et al., 2019. Pressure-decay and thermodynamic charac-
Eng. Therm. 23 (5), 623–626. teristics of subcooled liquid in the tank and their interaction with flashing jets. J.
Chen, Q.S., Wegrzyn, J., Prasad, V., 2004. Analysis of temperature and pressure Hazard. Mater. 378, 120578, http://dx.doi.org/10.1016/j.jhazmat.2019.04.061.
changes in liquefied natural gas (LNG) cryogenic tanks. Cryogenics 44, 701–709, Zhu, J.L., Zhang, Y.X., Liu, S.N., et al., 2020a. Experimental research on natural gas leak-
http://dx.doi.org/10.1016/j.cryogenics.2004.03.020. age underwater and burning flame on the water surface. Process Saf. Environ.
Chen, Y.Z., Yang, C.S., Shuming, Zhang, et al., 2009. Experimental study of choked Prot. 139, 161–170, http://dx.doi.org/10.1016/j.psep.2020.03.038.
flow of water at supercritical pressure. Front Energy Power Eng. China 3 (2), Zhu, G.R., Guo, X.Y., Yi, Y., et al., 2020b. Experiment and simulation research of evolu-
175–180, http://dx.doi.org/10.1007/s11708-009-0029-6. tion process for LNG leakage and diffusion. J. Loss Prev. Process Ind. 64, 104041,
Dan, S., Lee, C.J., et al., 2014. Quantitative risk analysis of fire and explosion on the http://dx.doi.org/10.1016/j.jlp.2019.104041.
top-side lng-liquefaction process of lng-fpso. Process Saf. Environ. Prot. 92 (5),
430–441, http://dx.doi.org/10.1016/j.psep.2014.04.011.

27

You might also like