Explosion Propagation in A Dust Removal Pipeline Under Dust Collector Explosion

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Journal of Loss Prevention in the Process Industries 74 (2022) 104662

Contents lists available at ScienceDirect

Journal of Loss Prevention in the Process Industries


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

Explosion propagation in a dust removal pipeline under dust


collector explosion
Lei Pang a, b, Jiaojiao Cao a, Qiuping Xiao c, Kai Yang a, b, *, Long Shi d, **
a
School of Safety Engineering, Beijing Institute of Petrochemical Technology, Beijing, 102617, China
b
Beijing Academy of Safety Engineering and Technology, Beijing, 102617, China
c
Shanghai Research Institute of Chemical Industry, Shanghai, 200062, China
d
Civil and Infrastructure Engineering, School of Engineering, RMIT University, Melbourne, 3004, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: The shockwave and high-temperature flame of dust explosion in the dust collector are often transmitted through
Dust explosion the dust removal pipeline, and then cause secondary harm. In this study, the dust explosion propagation in a
Dust collector typical dry dust collector connected with dust removal pipelines was studied by CFD. The impacts of static
Venting
activation pressure of the pressure relief panels on the explosion propagation were also addressed. The dust
Dust removal pipeline
Explosion shockwave
explosion propagation in the dust removal pipeline showed multiple accelerated combustion processes and se­
Flame propagation vere pressure shocks. The explosion pressure and temperature first increase and then decrease along with the
pipeline. Under a higher static activation pressure, the maximum explosion pressure along with the pipeline
increases by 144%, where the average velocity of the dust propagation, dust consumption, pressure shock wave,
and flame propagation increase by 82.2%, 32.1%, 25.3%, and 70.1%, respectively. When the static activation
pressure increases from 0.01 MPa by more than 25%, the changes of the dust explosion pressure and flame
propagation in the pipeline are observed limited. The explosion-proof valve should be set between 5 m and 10 m
in the dust removal pipeline. And the lower the static activation pressure, the closer the explosion-proof valve.
This research provides a scientific basis for the effective control of dust explosion disasters in dust removal
systems.

1. Introduction have a complete explosion venting and explosion-proof device, the


high-temperature gas and combustion products generated by the ex­
Dust explosions in industrial dust collectors often show serious plosion were ejected from the dust suction ports through the pipeline,
consequences (Taveau, 2010), which account for more than 40% of the endangering all the operators in the workshop (Li et al., 2016). There­
total dust explosion accidents (CSB, 2006; Zalosh et al., 2005). Dust fore, it is necessary to choose proper explosion venting restraints and
explosions spread through the dust removal pipeline, which is one of the explosion-proof measures (Taveau et al., 2019).
main hazards for casualties. During the operation, one side of a dry dust The installation position of the explosion protection facility on the
collector is connected to the dust removal pipeline that receives the dust, dust removal pipeline is affected by the explosion flame and shock wave
while the other side is equipped with a pressure relief panel to reduce propagation. At the same time, the explosion flame and shock wave
the explosion risk. Although an explosion venting device can reduce the propagation are related to the characteristics of the explosion vent on
risk to a certain degree, the aerodynamic conditions in the the dust collector. The installation of pressure relief panels in powder-
dust-conveying pipeline can accelerate the dust explosion propagation, related equipment is one of the most economical and reliable mea­
which may cause serious brisance (Proust, 1996). In 2014, a cata­ sures (Song et al., 2019; Jiang et al., 2020). The size, configuration,
strophic aluminum dust explosion occurred in a polishing workshop in static activation pressure and material properties of the pressure relief
Jiangsu Province of China. This accident is a typical case induced by a panels affect the propagation of explosion venting (Wang et al., 2020;
dust explosion in the dust collector. As the dust removal pipeline did not Tascón et al., 2016; Yan and Yu, 2014). Among them, the static

* Corresponding author. School of Safety Engineering, Beijing Institute of Petrochemical Technology, Beijing, 102617, China.
** Corresponding author. Civil and Infrastructure Engineering, School of Engineering, RMIT University, Melbourne, 3004, Australia.
E-mail addresses: ycyangk@bipt.edu.cn (K. Yang), long.shi@rmit.edu.au (L. Shi).

https://doi.org/10.1016/j.jlp.2021.104662
Received 14 August 2021; Received in revised form 10 October 2021; Accepted 12 October 2021
Available online 15 October 2021
0950-4230/© 2021 Elsevier Ltd. All rights reserved.
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

activation pressure of the pressure relief panel is an important influ­ due to sub-grid obstructions in the i-direction; and ρ0 is the initial
encing factor. Lunn, Crowhurst and Hey (1988) found that the vent density.
pressure along the pipeline continued to increase with the increase in The energy transfer equation is:
the vent opening pressure. A higher static activation pressure will cause ( )
∂ ∂ ( ) ∂ μeff ∂h Dp Q̇
a strong jet flame at the explosion vent, and cause higher turbulence, (βυ ρh) + βj ρuj h = βj + βυ + (3)
which is likely to increase the possibility of a second explosion (Taveau, ∂t ∂xj ∂xi σ h ∂xj Dt V
2010). A static activation pressure of up to several bars can be selected
where h is the specific enthalpy; μeff is the effective viscosity; σ h is the
for small containers with high pressure resistance, while a smaller static
Prandtl–Schmidt number of enthalpy; Dp is the diffusion coefficient of
activation pressure should be selected for large-scale equipment such as
dust collectors, silos, and hoppers. When the static activation pressure particle; Dt is the diffusion coefficient of turbulence; and Q̇ is the heat
increases to a certain degree, the explosion overpressure could no longer rate.
increase significantly (Yan and Yu, 2014). However, there are few Fuel mass transfer equation:
studies on the effect of the static activation pressure of the explosion ∂ ( ) ∂ ( ) ∂
(
μeff ∂Yfuel
)
vent of the dust collector on the explosion propagation in the dust βυ ρYfuel + βj ρuj Yfuel = βj + Rfuel (4)
∂t ∂xj ∂xi σfuel ∂xj
removal pipeline. Taveau et al. (2019) established a relatively real dust
collector explosion scene, but its volume is relatively small, and the where Yfuel is the mass fraction of the fuel; σfuel is the Prandtl–Schmidt
research content is relatively shallow, which cannot meet the re­ number of the fuel; and Rfuel is the reaction rate of the fuel.
quirements of modern industrial explosion-proof and controlled explo­ Turbulence energy transfer equation is given by:
sion technology. Therefore, it is of great significance to further study the ( )
∂ ∂ ( ) ∂ μeff ∂k
influence of explosion venting restrictions of dust collectors on the (βυ ρk) + βj ρuj k = βj + βυ Pk − βυ ρε (5)
spread of dust explosion disasters in dust removal pipeline. ∂t ∂xj ∂xi σk ∂xj
The aforementioned analysis shows that research on the propagation
where k is the turbulent kinetic energy; σ k is the Prandtl–Schmidt
of dust explosion hazards in dust removal pipeline of typical dust
number of turbulent kinetic energy; Pk is the turbulent kinetic energy
removal equipment is very limited. Therefore, this study addressed the
production; and ε is the turbulent kinetic energy dissipation.
influences of the static activation pressure of pressure relief panels on
Turbulent flow energy dissipation rate transmission equation is:
the dust explosions propagation in dust removal pipelines by computa­
( )
tional fluid dynamics (CFD). It pioneered a practical research based on ∂ ∂ ( ) ∂ μeff ∂ε ε2
the structure of a typical dust collector and provided a scientific basis for (βυ ρε) + βj ρuj ε = βj + β υ Pε − C 2 ε β υ ρ (6)
∂t ∂xj ∂xi σε ∂xj k
the design of explosion-proof and controlled explosions in the dust
removal pipeline. where σ ε is the Prandtl-Schmidt number of the turbulent kinetic energy
dissipation; Pε is the production of dissipation of turbulent kinetic en­
2. Numerical methodologies ergy; and C2ε is the constant in the k-ε equation, typically C2ε = 1.92.
A key feature of FLACS is the distributed porosity concept to repre­
2.1. FLACS-DustEx model sent complex geometric figures on a relatively coarse computational
grid. The large simulation objects and walls are represented on the grid
FLame ACceleration Simulator (FLACS) is a numerical tool using this method, while smaller objects are represented on the sub-grid.
frequently used to deal with industrial safety issues. The FLACS-DustEx The processor analyses the volume distribution and area porosity of each
is a toolkit for dust explosion and impact dynamics in FLACS, which rectangular grid cell. During the simulation processes, the porosity
shows a great potential to deal with severe losses caused by major in­ represents the local congestion and limitation, which helps the sub-mesh
dustrial and social disasters (Skjold, 2014; Gexcon, 2018). FLACS uses object forming strong flow resistance, turbulence generation, and flame
the finite volume method to solve compressible conservation equations folding (Tascón and Aguado, 2017; Gexcon, 2018).
on the three-dimensional Descartes grid, including the conservation of
mass, momentum, energy, and mass fraction equations of each compo­ 2.2. Combustion model
nent defined by the ideal gas law. The FLACS relies on the turbulence
model of the Reynolds-Averaged Navier–Stokes (RANS) equation, such The premixed fuel and oxidant are ignited in a static state to create a
as the k-ε model (Launder and Spalding, 1974). This eddy viscosity laminar flame at the initial stage. The flame propagation is entirely
model solves the two transport equations of the turbulent kinetic energy controlled by the thermal diffusion and molecular diffusion process, and
and the turbulent kinetic energy dissipation (Skjold, 2007). These the flame front is smooth. Subsequently, the flame wrinkles, the flame
equations are shown as follow. speed increases, and a turbulent flame are formed through the transition
The mass conservation equation can be given by: zone under the action of fluid dynamics and Rayleigh–Taylor flame
∂ ∂ ( ) ṁ instability (Skjold, 2007; Gexcon, 2018).
(β ρ) + β ρu = (1) The flame combustion process in FLACS-DustEx is calculated by
∂t v ∂xj j j V
combining the classic β-flame model shown in Eq. (7) (Abdel-Gayed
where uj is the velocity component in the xj coordinate direction; p is the et al., 1987) and the experimental results of the basic explosion of a 20 L
pressure; ρ is the density; βv is volume porosity, a dimensionless ball. Assuming that the kinematic viscosity is constant and the turbu­
parameter; βj is the area porosity in the j direction; ṁ is the mass rate; lence integral length scale (lI) is introduced, Eq. (7) can be simplified to
and V is the volume. Eq. (8) (Popat et al., 1996),
The momentum transfer equation is expressed as follows: ′
ST 0.392 urms
= 0.875K − (7)
∂ ∂ ( ) ∂p ∂ ( ) SL SL
(β ρu ) + β ρu u = − β υ + β σ + Fo,i + Fw,i + βυ (ρ − ρ0 )gi
∂t υ i ∂xj j i j ∂xi ∂xj j ij
(2) (8)
0.412
ST = 15.1SL0.784 u′ rms l0.196
I

where ui is the mean velocity in the i-direction; gi is the gravitational where ST is the turbulent burning velocity, SL is the laminar burning
acceleration in the xi direction; σ ij is the turbulent momentum flux at the velocity, u′rms is the root-mean square of turbulent velocity fluctuations,
cell surfaces; Fw,i is flow resistance due to walls; Fo,i is the flow resistance and K is the Karlovitz stretch factor (Arntzen, 1998).

2
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

To establish different combustible dust combustion models, it is area is 5 kg/m2. Different static activation pressures were then adopted.
necessary to have explosion experiments using a typical 20 L spherical When the pressure in the dust collector reaches the pre-set static acti­
explosion tank that determines the explosion pressure and pressure peak vation pressure, the porosity of the pressure relief panel increases from
time point for a series of dust concentrations. The experimental data in 0 to 1 immediately. The base static activation pressure was set as 0.01
the 20 L sphere was analyzed and calculated, and the turbulent com­ MPa, and the other static activation pressures were set to fluctuate
bustion thin flame velocity was obtained (Dahoe et al., 1996; Dahoe within 25%–75% up and down. The experimental static activation
et al., 2001). Then, Eq. (8) was substituted for the inverse calculation to pressures (Pstat) were 0.005, 0.0075, 0.01, 0.0125, 0.015, and 0.0175
obtain the estimated laminar combustion velocity (Skjold et al., 2005, MPa.
2006). Regular corn starch was selected as the dust fuel, with a particle
In the modelling processes, the dust cloud is defined as a dense gas density of 1180 kg/m3 and dust median diameter of 15 μm. The defla­
with a high molecular weight. The simplified two-phase fluid is a ho­ gration index (Kst) was 15 MPa m/s, and the maximum explosion pres­
mogeneous equilibrium model. It was assumed that the two-phase flows sure (Pmax) was 0.86 MPa. The dust concentrations were set as 750 g/m3
are in thermal and kinetic equilibrium with the continuous phase; that and 1000 g/m3, evenly arranged inside the dust collector. Table 1 lists
is, the heat and mass transport between the phases were infinitely fast. the variables.
For the modelling settings, only the particle volume fraction is required, The ignition position was set at the center of the dust collector, and
not the particle diameter and velocity (Ichard, 2012; Tascón and the ignition energy was 10 kJ. The initial temperature and pressure were
Aguado, 2017). set at 20 ◦ C and 0.1 MPa, respectively. A previous study (Gexcon, 2018)
showed that the Euler boundary is suitable for the numerical simulation
of most explosion scenarios; hence, each model in this study adopted
2.3. Geometry and scenario Euler boundary conditions.

A simplified geometry model of the dust collector was established 2.4. Grid independence
based on the typical dry precipitator structure, as illustrated in Fig. 1(a).
The model includes a 4 × 4 × 5 m3 dust collector, where one side is In those previous studies, FLACS-DustEx was adopted to simulate the
connected to a dust removal pipeline with a cross-sectional diameter of dust explosion in large-scale geometry models, and the grid was set at
0.4 m and a length of 15 m. The dust removal pipeline and the dust 0.1 m (Castellanos et al., 2013; Tascón and Aguado, 2015, 2017). The
collector are in a connected state without any obstacles. As illustrated in grid resolution should be chosen to obtain a sufficiently accurate
Fig. 1(b), an explosion vent is on the other side of the dust collector, simulation result within an acceptable time frame. The pipes and con­
along with pressure relief panels. As illustrated in Fig. 1(c), the moni­ tainers must always be resolved by a minimum of five to six grid cells in
toring points were set at the center of the dust collector with an interval smallest direction, while the explosion vent should contain at least six to
of 1 m along with the pipeline. All monitoring points were used data eight grids (Skjold, 2014; Gexcon, 2018). Therefore, a 0.1 × 0.1 × 0.1
recordings, such as the pressure, flame, and dust concentration calcu­ m3 grid was selected as the main grid of the calculation model. Different
lated by the model. numbers of grids were set at the explosion vents and pipelines for the
According to ISO 6184-1 (1985), NFPA 68 (2018), VDI 3673 (2002) analysis, as shown in Table 2. At the same time, Yang et al. (2020b)
and GB 15605 (2008), the explosion venting area of the geometry model found that the range of the external flow field would have an impact on
is about 1.5 m2. A popout explosion vent was set up, and its mass per unit the explosion venting, so three types of simulation domains outside the
vent were selected to study their impact on the propagation of explosion
hazards.
The nine sets of simulation conditions were divided into four error
analysis blocks (A, B, C, and D), as illustrated in Fig. 2. The appropriate
simulation conditions were selected through average relative error
analysis and comparison of the simulation results to save the time and
economic cost of large-scale explosion experiments. After calculating the
physical model under nine sets of simulation conditions, the five typical
locations, including the center of the dust collector, pipe interface, 4 m
in the pipe, 1 m in the explosion vent, and 5 m in the explosion vent were
selected as the maximum explosion pressure dataset. According to the
error formula group (Eqs. (9-12)), the error analysis of the four blocks
was determined by,

Table 1
Analyzed variables of this study.
Parameter Variable

Dust concentration (g/ 750 1000


m3)
Static activation 0.005 0.0075 0.01 0.0125 0.015 0.0175
pressure (MPa)
Fig. 1. Geometry model of a typical industrial dust collector.

3
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

Table 2 and D represent the same.


Simulation conditions for different grid scenarios. The minimum value between A, B, C, and D was calculated as D. That
Simulation scenario The number of cells in smallest direction of the is, the smallest difference between four calculation scenarios (S22, S23,
dust removal pipeline and vent (Fine mesh S32, and S33). Hence, the most suitable calculation conditions were
size/m) selected among these four scenarios. Based on the principle of high ef­
Cn = 4 and Cn = 8 and Cn = 12 and ficiency, scene S22 with a small simulation domain and a large grid size
5 (0.1) 10 (0.05) 15 (0.033) was selected as the basis of the simulation in this research. This is in line
Simulation domain (X 29 × S11 S12 S13 with the condition that the simulation domain should be three to four
× Y × Z/m3) 10 × 10 times that of the main explosion container (Gexcon, 2018). A fine grid of
34 × S21 S22 S23 0.05 m was adopted to refine the calculation model locally.
14 × 12
39 × S31 S32 S33
18 × 14 2.5. Experimental validation

The corn dust experiment conducted by Tascón et al. (2016) in a


16.3 m3 silo was selected for the experimental validation. The reduced
explosion overpressure (Pred) was chosen for the comparison. Table 3
lists the simulation results obtained under different simulation condi­
tions. The experimental validation under 12 different grids and laminar
burning velocity conditions are denoted as N1 to N12.
A comparison between the numerical and experimental results was
undertaken at a burning rate of 1.5, a grid of 0.1 m. The local fine grid
showed an error of approximately 4%, which is within the acceptable
range (Pang et al., 2019). The results showed that the local fine grids
slightly increase the predicted pressure, but they were still within the
error range. Therefore, the laminar burning velocity enhancement factor
of 1.5 was confirmed as the simulation condition.

3. Results and discussion

3.1. Explosion parameter analysis method

Fig. 3 shows the time curve of the explosion parameters at 6 m in the


pipeline. The suddenly increased dust concentration is the moment
when the dust spreads to that location. The suddenly dropped dust
concentration is the moment when the dust starts being consumed. The
suddenly increased burning rate is the moment when the flame arrives
and this is consistent with the incubation period with sudden tempera­
ture rise and the instantaneous consumption of dust. The velocities at
various measuring points along the pipeline were obtained according to
Fig. 2. Schematic diagram of block division of simulation results error analysis. the relationship between arrival time and the distance (Jiang et al.,
2013; Ma et al., 2015). When the dust concentration is stable at 0.5 s,

⎪ dust concentration decreases rapidly after the pressure shock wave
⎪ |PS11 − PS12 |

⎪ ai =
⎪ sweeps. Then the flame reaches the dust, and it burns and consumes
⎪ Max(PS11 , PS12 )



⎪ quickly. The period between the dust consumption and flame arrival is



⎪ … determined by the airflow disturbance at the front of the flame.




⎪ Fig. 4 shows the curve of the dust cloud distribution velocity in the




… dust removal pipeline after the dust explosion in the dust collector under
(i = 1, 2, 3, 4, 5) (9) different dust concentrations. It was found that the dust cloud distri­
⎪ |PS21 − PS11 |


⎪ ci = bution after the dust cloud explodes in the dust removal pipeline shows a
⎪ Max(P S21 , PS11 )



⎪ similar pattern under different concentrations. However, both the



⎪ … propagation velocity and consumption velocity of the dust cloud at a




⎪ dust concentration of 750 g/m3 are greater than those at a dust con­
centration of 1000 g/m3. This is because the dust and air content reach




⎩ …
the optimal explosive reaction ratio at a dust concentration of 750 g/m3.
The rapid spread and consumption of the dust reflect the severity of the
ai + ci + di + f i explosion, which is consistent with the most dangerous corn dust ex­
Ai = (10)
4 plosion concentration of 750 g/m3 (Pang et al., 2020). And the air
∑5 mixture with lower dust concentration has lower inertia, which makes
A= i=1 Ai
(11) the explosion spread easier (Taveau, 2017). Therefore, the dust con­
5 centration of 750 g/m3 was selected as the control or the basic scenario
for further analysis. Table 4 lists the related scenario parameters.
δ = min{A, B, C, D} (12)
The ai is the relative error of the maximum pressure of the selected 3.2. The propagation of dust explosion
measuring point under the two sets of simulation conditions, namely S11
and S12. A represents the average error between the calculation results of The explosion processes in a typical dry dust collector were simu­
the five measuring points in four scenarios (S11, S12, S21, and S22); B, C, lated under the basic scenario. Fig. 5 shows explosion images of the

4
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

Table 3
Error between experiment and simulation results.
Experimental validation Simulated conditions Simulation result Experiment result Relative error

Cell size/m Modified laminar burning velocity Pred/MPa Pred/MPa

N1 0.05 1 0.0611 0.09 32.1%


N2 0.1 1 0.0391 56.6%
N3 0.2 1 0.0380 57.8%
N4 0.1 (Fine grid: 0.05) 1 0.0397 55.9%
N5 0.05 1.25 0.0970 7.78%
N6 0.1 1.25 0.0652 27.6%
N7 0.2 1.25 0.0684 24.0%
N8 0.1 (Fine grid: 0.05) 1.25 0.0654 27.3%
N9 0.05 1.5 0.1238 27.3%
N10 0.1 1.5 0.0935 3.89%
N11 0.2 1.5 0.1050 14.3%
N12 0.1 (Fine grid: 0.05) 1.5 0.0937 3.95%

Table 4
Basic scenario parameters.
Simulation parameters Value

Explosion vent size 1.5 m2


Static activation pressure 0.01 MPa
Ignition position Dust collector center
Ignition energy 10 kJ
Dust concentration 750 g/m3
Modified laminar burning velocity 1.5

time, the pressure piling effect caused by the shrinkage of the container
joint diameter pre-compresses the unburned dust medium (Reding and
Shiflett, 2020; Bjerketvedt et al., 1997; Zalosh, 2009; Eckhoff, 2003).
Thus, the first pressure shock wave (P1) was formed, which quickly
propagated toward the end of the pipeline. The pressure at 10 m, where
P1 propagates toward the pipeline, increases significantly because the
Fig. 3. Explosion curve of a monitoring point at 6 m of the dust
waves that are generated later will always catch up with the waves
removal pipeline.
generated earlier. And the P1 has spread to 10 m, while the flame has
spread to 5 m, indicating that the pressure wave is always faster than the
flame. The sweeping area of P1 formed a negative pressure zone under
the action of the expansion wave. Under the action of the negative
pressure vacuum, the explosion medium in the dust collector was sucked
into the dust removal pipeline, creating a small pressure fluctuation (P2).
The unburned dust that accumulated at the end of the pipeline was
ignited under the action of the jet flame. After 10 m, it is in the sec­
ondary flame acceleration stage, and the second strong pressure wave
(P3) is formed. After the strong compression wave P3 is swept, a large
negative pressure peak (P4) appears, and the maximum negative pres­
sure appears near 6 m of the pipeline. The negative pressure of explosion
propagation in the dust removal pipeline is lower than the positive
pressure, so there is no need to worry about the equipment failure
caused by the negative pressure. After the explosion at 0.66 s propagated
to the end of the pipeline, the pressure fluctuation gradually stabilised
(Rocourt, 2014). It only takes 0.11 s for the accelerate propagation of
explosion flame in the dust removal pipeline. This provides a reasonable
basis for the reaction time between the spark detector and the sprinkler
system in the pipeline.
Fig. 4. Distribution velocity of dust cloud in pipeline after explosion under At 0.512 s, the explosion reaches the static activation pressure of the
different dust concentrations.
pressure relief panels and begins to propagate in one direction from the
explosion vent. The Pb in the dust collector is 0.0102 MPa; at 0.525 s, the
flame (PROD), dust concentration (DUST), and pressure (P) changes in Pb decreases to 0.0051 MPa at 2 m in the dust removal pipeline, and the
the dust explosion. These three factors interact and communicate with explosion begin to propagate to both sides. The high pressure of the
each other (Abbasi and Abbasi, 2007; Zhang et al., 2017; Yang et al., explosion in the dust collector has a short propagation distance in the
2019). And Fig. 6 shows the explosion pressure propagation curve and dust removal pipeline, and the pressure potential energy is quickly
temperature change curve at every 2 m in the dust removal pipeline. It consumed. This is caused by the rapid shrinkage of the dust removal
was observed that the peak pressure of the explosion relief (Pb) first pipeline connection. Based on this, it is believed that adding a blocking
appeared in the dust removal pipeline. As the flame propagated into the device between the dust collector and the dust removal pipeline can
dust removal pipeline, the dust cloud within 6 m of the dust removal reduce the propagation of explosive kinetic energy into the pipeline.
pipeline is rapidly consumed, and a jet flame is formed. At the same The time at which the temperature suddenly increases is considered

5
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

Fig. 5. Explosion propagation flame (PROD), dust concentration (DUST) and pressure (P) changes in the dust removal pipeline (Pstat: 0.01 MPa; Dust concentration:
750 g/m3).

as the flame arrival time (Jiang et al., 2013). The temperature in the proof valve should be installed 5 m after the pipeline to effectively block
pipeline stays above 2000 K until the flame reached the end of the the flame under high-pressure start. The arrival time of the pressure
pipeline. The material of the dust removal pipeline should be designed shock wave between 5 m and 9 m is the same as the dust consumption
based on the high temperature of the explosion to prevent the flame time. The pressure shock is the main driving force that drives the dust
from spraying out through the flange gasket. The temperature of the dust dispersion. The high-speed dust consumption stage involves the accu­
explosion along the pipeline showed a trend of increasing first and then mulation of energy after 10 m of pressure waves and dust propagation.
decreasing. The temperature reached the maximum value at 10 m in the The flame velocity peaks were observed at 4 m and 9 m, confirming the
pipeline. This is caused by the secondary accelerated combustion at the pressure accumulation effect caused by the sudden contraction of the
end of the pipeline. pipe and secondary combustion at the end of the pipe. Therefore, in
Fig. 7 shows the arrival time and velocity changes of the dust ex­ order to prevent the secondary acceleration of explosion propagation in
plosion propagation. According to the above analysis, the flame and the pipeline, the explosion-proof valve should be set 10 m in front of the
fluid media influence each other during the dust explosion propagation. pipeline.
At 0.293 s, the explosion airflow first carried dust into the dust removal
pipeline at a low speed, forming an initial explosion environment, and
3.3. Static activation pressures
the dust cloud started to spread before the flame entered the pipeline.
After the dust propagation experienced 0.232 s, compared with the
The turbulent flow energy in the dust collector represents the tur­
pressure shock wave, the flame first was transmitted into the pipeline to
ignite the explosive environment, and the P1 was formed and quickly bulent mixing capacity of different explosion media. The higher the
mixing capacity of air and dust is the larger the contact surface of air,
spread to the end of the pipeline with an interval of 0.0297 s. When the
explosion reaches 5 m at the pipeline, the flame propagation lagged dust, and flame will be (Zhang et al., 2019, 2020; Yang et al., 2020a).
Therefore, it results in a more complete explosion reaction. Thus, high
behind the pressure shock significantly. This is because the initial stage
of the explosion is in the laminar combustion stage, so the explosion turbulence energy can promote the spread of dust explosions in dust
removal pipelines.
pressure accumulates slower than the flame propagation. Then the ex­
plosion develops to a turbulent combustion state and quickly forms an Fig. 8 shows the trend of the turbulent flow energy in the dust col­
lector over time following a dust explosion at different static activation
explosion shock wave that surpasses the flame. Therefore, the explosion-
pressures. It was found that the greater the static activation pressure is,

6
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

Fig. 6. Explosion overpressure and temperature changes at different positions along with the pipeline (Pstat: 0.01 MPa; Dust concentration: 750 g/m3).

the greater the turbulent kinetic energy after the explosion in the dust
collector will be. And it increases by up to 2.66 times with the static
activation pressure from 0.005 MPa to 0.0175 MPa. The moment of the
sudden increase of turbulent kinetic energy advances with the increase
of static activation pressure. The lower static activation pressure can
ensure the explosion vent in the early stage of deflagration (Taveau
et al., 2019). This is because the static activation pressure is small, and
explosion venting occurs first before the violent explosion reaction, so
the sudden increase of turbulent kinetic energy is delayed and the peak
value is small. The greater the static activation pressure, the sufficient
time to complete the dust explosion in the dust collector, so a more vi­
olent explosion occurred earlier. At the same time, there will be more
unburned dust in the dust collector at a lower static activation pressure
to participate in the secondary explosion, resulting in a second peak of
turbulent energy.
Combined with the explosion propagation stage division shown in
Fig. 11, the peak turbulent kinetic energy occurs in the two-way prop­
agation process of the dust collector explosion when the static activation
Fig. 7. Dust explosion behaviors along the dust removal pipeline (Pstat: 0.01 pressure is less than 0.015 MPa. When the static activation pressure is
MPa; Dust concentration: 750 g/m3). 0.0175 MPa, the peak turbulent kinetic energy occurs before the

7
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

Fig. 8. The turbulent kinetic energy of dust explosion in the dust collector
under different static activation pressures.

explosion venting. And when the static activation pressure reaches


0.0125 MPa, the trend in the turbulent kinetic energy changes is almost
the same. Therefore, the explosion venting effect is considered reaching
the limit.
The greater the turbulence kinetic energy in the dust collector after
the explosion is, the more severe the impact of the explosion in dust
removal pipeline will be. Fig. 9(a) shows the distribution of the
maximum explosion pressure at each monitoring point of the dust ex­
plosion spreading in the dust removal pipeline at different static acti­
vation pressures of explosion venting. The maximum explosion pressure
at each monitoring point showed a trend of first increasing and then
decreasing. The maximum explosion pressure appeared at 3–4 m in the
dust removal pipeline, which is the weak position of the dust removal
equipment explosion.
Fig. 9(b) shows the average maximum explosion pressure and pres­
sure distribution range at each monitoring point in the dust removal
pipeline. As the static activation pressure increases, the average
Fig. 9. The maximum explosion pressure in the pipeline under different static
maximum explosion pressure at each monitoring point in the dust
activation pressures.
removal pipeline increases. And the maximum explosion pressure dis­
tribution floating range also increases. In addition, the average
should be installed before 10 m. Fig. 10(b) compares the average
maximum explosion pressure increases by 144% with the static activa­
propagation velocities of the two fluids at different static activation
tion pressure from 0.005 MPa to 0.0175 MPa, and the maximum pres­
pressures. It is found that the average propagation velocity of the P1 is
sure fluctuation range of each monitoring point in the dust removal
between 370 and 480 m/s; the average velocity of dust cloud propaga­
pipeline expands by 2.82 times. However, when the static activation
tion is between 40 and 90 m/s; and both increase with the increase in
pressure reaches 0.0125 MPa, there is almost no change of the maximum
static activation pressure. Based on this, it is believed that a smaller
explosion pressure distribution at various positions in the pipeline. The
static activation pressure should be selected. And it is found that the
maximum explosion pressure in the pipeline is lower than 1.5 MPa,
static activation pressure of 0.0125 MPa is the dust explosion venting
which has met the requirements of the mechanical explosion-proof valve
limit of the dust collector model, and the explosion indicators will no
to isolate the explosion separately, and there is no need to add chemical
longer change.
explosion-proof technology (Taveau et al., 2018).
After a dust explosion occurs in the dust collector, the explosion fluid
propagation can be classified into two types: pressure wave propagation 3.4. Flame propagation
and dust cloud propagation. Fig. 10(a) shows the propagation velocity of
the explosion shock wave P1 and dust diffusion along the dust removal Fig. 11 shows the time nodes of the dust explosion propagation in the
pipeline at different static activation pressures. It was observed that the dust collector at different static activation pressures. It was found that
pressure shock wave P1 first enters the dust removal pipeline at a low the increase in the static activation pressure increases the time it takes to
speed, reaches a peak velocity at 2 m, and then decreases and propagates open the vent. Therefore, the larger static activation pressure allows
4–8 m in the pipeline at an almost uniform velocity. After 10 m, the sufficient time for the explosion reaction in the dust collector, which
pressure wave velocity gradually increases and exceeds the peak ve­ causes a more violent explosion flame to quickly enter the dust removal
locity at 2 m. The dust diffusion velocity shows a rapidly increasing pipeline and shorten the propagation time. Among them, the time
trend after 11 m. After the explosion spreads to 10 m, the explosion required for flame propagation is reduced by up to 38.3% with the static
propagation speed is obviously accelerated. In order to reduce the dy­ activation pressure from 0.005 MPa to 0.0175 MPa. The shorter the
namic impact caused by the increase speed, an explosion-proof valve flame propagation time, the more violent the explosion. Therefore, the
reaction time of spark detection and sprinkler system should be shorter.

8
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

In addition, since the flame spreads slowly when the static activation
pressure is lower, the installation distance of the explosion-proof valve
in the dust removal pipeline should be a little forward. When the static
activation pressure increases to 0.0125 MPa, the explosion flame first
enters the dust removal pipeline before it is released from the explosion
vent. The accelerated propagation time of the explosion flame in the
pipeline hardly changed, indicating that the explosion reaction inside
the dust collector is fully completed.
Combustion velocity is an attribute of dust, while the flame propa­
gation velocity is a practical parameter for explosion prevention and
control. The flame propagation velocity is relative to a static pipe, which
depends on the disturbance of the airflow in front of the flame array (Bi
et al., 2017). Flame propagation velocities under different static acti­
vation pressures are shown in Fig. 12(a). It was found that the flame
propagated at a fluctuating speed. The flame combustion speed fluctu­
ates more frequently at a lower explosion vent opening pressure. When
the explosion pressure release occurs at a lower static activation pres­
sure, the unburned dust in the dust collector is relatively large, while the
turbulence velocity in the pipeline is relatively low. Therefore, the

Fig. 10. Velocity of pressure shock wave (P1) and dust diffusion under different
static activation pressures.

Fig. 11. The dust explosion propagation stage in the dust collector under
different static activation pressures.
Fig. 12. Velocity of flame propagation and dust consumption under different
static activation pressures.

9
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

accumulated dust in the dust removal pipeline can achieve multiple determined between 5 m and 10 m of the dust removal pipeline. And
accelerated combustion and form an unstable flame within the same when the static activation pressure is lower, the explosion-proof valve
propagation distance. It is consistent with the fact that the flame prop­ should be installed closer to the dust collector. The larger static acti­
agation velocity fluctuates in the experiment (Pang et al., 2021). As the vation pressure causes the explosion to spread more violently, so the
static activation pressure increases, the number of flame velocity peaks reaction time between the spark detector and the sprinkler system
decreases from three to two. Multiple combustion accelerations are not should be reduced. In the process of explosion disaster propagation
good for explosion propagation, and based on this, the static activation control involving typical dry dust collectors, attention should be paid to
pressure should not be too small. When the static activation pressure the characteristics of explosion propagation inside the dust removal
increases to 0.0125 MPa, the flame propagation velocity at the front pipeline.
section of the pipeline is almost the same, and there is a slight difference
in the propagation to the end of the pipeline. Author statement
Fig. 12(a) also shows a rapid change in the dust consumption rate
when the explosion spreads along the dust removal pipeline at different Lei Pang: Conceptualization, Resources, Supervision, Project
static activation pressures. The greater the dust consumption rate is, the administration, Funding acquisition. Jiaojiao Cao: Formal analysis,
greater the degree of disturbance of the explosion fluid at the front edge Writing – original draft, Writing – review & editing, Visualization,
of the flame will be. The dust explosion spreads along the dust removal Investigation. Qiuping Xiao: Methodology, Formal analysis, Investiga­
pipeline, and the dust consumption velocity showed the three stages of tion. Kai Yang: Supervision, Writing – original draft, Writing – review &
increasing, decreasing, and then increasing again. The dust consumption editing. Long Shi: Writing – original draft, Investigation, Visualization.
velocity is at a relatively high level between 6 m and 9 m, indicating that
the explosion disturbance is the most severe in the middle section of the
Declaration of competing interest
dust removal pipeline.
The average flame propagation velocity and average dust con­
The authors declare that they have no known competing financial
sumption velocity at different static activation pressures are compared,
interests or personal relationships that could have appeared to influence
as illustrated in Fig. 12(b). Both increase with a higher static activation
the work reported in this paper.
pressure. The flame propagation and dust consumption velocity increase
by 70.1% and 32.1%, respectively, with the static activation pressure
Acknowledgements
from 0.005 MPa to 0.0175 MPa. The airflow disturbance zone at the
front edge of the flame has a faster propagation velocity than the flame,
The authors thank Elsevier for their English translation service. The
resulting in an increase in the average dust consumption velocity of
authors appreciate the financial support from the Beijing Science and
approximately 80 m/s compared with the average velocity of flame
Technology Nova Program (No. Z181100006218092), the Shanghai
propagation.
Science and Technology Development Funds (No. 19QB1402800), the
Training Funded Project of the Beijing Youth Top-Notch Talents of China
4. Conclusions
(No. 2016000026833ZK05), and Innovation and Entrepreneurship
Project of China Occupational Safety and Health Association (No. CXCY-
The effects of static activation pressure on the dust explosion prop­
2021-04).
agation were investigated in a dust collector by CFD under the coupling
effects of the explosion venting constraint and the dust removal pipeline.
References
The explosion flame enters the dust removal pipeline to achieve multiple
accelerations and generate multiple explosion shockwaves. The explo­ Abdel-Gayed, R.G., Bradley, D., Lawes, M., 1987. Turbulent burning velocities: a general
sion pressure and temperature at various locations in the pipeline show a correlation in terms of straining rates. Proc. Roy. Soc. Lond. 414, 389–413.
trend of first increasing and then decreasing. And the maximum explo­ Arntzen, B.J., 1998. Modelling of turbulence and combustion for simulation of gas
explosions in complex geometries. Dr. Ing. Thesis. NTNU, Trondheim, Norway.
sion pressure and temperature occurred at 4 m and 10 m of the dust Abbasi, T., Abbasi, S.A., 2007. Dust explosions–Cases, causes, consequences, and control.
removal pipeline, respectively. J. Hazard Mater. 140 (1–2), 7–44.
The greater the static activation pressure of the dust collector is, the Bjerketvedt, D., Bakke, J.R., Wingerden, K.V., 1997. Gas explosion handbook. J. Hazard
Mater. 52 (1), 1–150.
more violent the dust explosion spread in the dust removal pipeline will
Bi, M.S., Li, G., Chen, X.F., Yang, G.G., 2017. Gas and Dust Explosion Prevention
be. With a higher static activation pressure, the maximum explosion Engineering, second ed. Chemical Industry Press, Beijing.
pressure at each position in the dust removal pipeline increases, and the CSB, 2006. Investigation Report. Combustible dust hazard study. No. 2006-H-1.
propagation velocities of the explosion shock wave and dust cloud both Castellanos, D., Skjold, T., Wingerden, K.V., Eckhoff, R.F., Mannan, M.S., 2013.
Validation of the DESC code in simulating the effect of vent ducts on dust explosions.
rise. At the same time, the average maximum explosion pressure in­ Ind. Eng. Chem. Res. 52, 6057–6067.
creases by 144%, and the pressure fluctuation range also expands by Dahoe, A.E., Zavenbergen, J.F., Lemkowitz, S.M., Scarlett, B., 1996. Dust explosions in
2.82 times. When the static activation pressure increases by 25% from spherical vessels: the role of flame thickness in the validity of the ‘cuberoot law.
J. Loss Prev. Process. Ind. 9 (1), 33–44.
0.01 MPa to 0.0125 MPa, the explosion propagation velocity and in­ Dahoe, A.E., Cant, R.S., Scarlett, B., 2001. On the decay of turbulence in the 20-litre
tensity hardly change. The static activation pressure of the pressure explosion sphere. Flow, Turbul. Combust. 67, 159–184.
relief panels for industrial dust explosions should not be greater than Eckhoff, R.K., 2003. Dust Explosions in the Process Industries, third ed. Gulf Professional
Publishing, Amsterdam.
0.0125 MPa. GB 15605, 2008. Guide for Pressure Venting of Dust Explosions. Standardization
The increase in the static activation pressure promoted the explosion Administration of the People’s Republic of China, Beijing.
flame propagation in the dust removal pipeline. When the static acti­ Gexcon, A.S., 2018. FLACS v10.7 User’s Manual, Bergen, Norway.
ISO 6184-1, 1985. Explosion Protection Systems-Part I: Determination of Explosion
vation pressure increases, the explosion reaction in the dust collector is Indices of Combustible Dusts in Air. International Organization for Standardization,
more complete and violent. The time required for flame propagation in Geneva.
the dust removal pipeline is reduced by 38.3%. At a relatively low static Ichard, M., 2012. Numerical Computations of Pressurized Liquefied Gas Releases into the
Atmosphere. PhD thesis. Department of Physics and Technology, University of
activation pressure, the flame propagation velocity fluctuates more
Bergen. February 2012. ISBN978-82-308-2010-0.
frequently. The fluid turbulence at the flame front causes the dust cloud Jiang, B.Y., Lin, B.Q., Zhu, C.J., Zhai, C., Liu, Q., 2013. Premixed methane-air
disappearance velocity to be faster than the flame propagation velocity deflagrations in a completely adiabatic pipe and the effect of the condition of the
by about 80 m/s. pipe wall. J. Loss Prev. Process. Ind. 26 (4), 782–791.
Jiang, Z.Y., Li, W., Liu, X.M., Jiang, F., Li, B.B., 2020. Research on different venting
According to the relative position of the flame and the pressure shock performance between flat and curved bursting panels. J. Loss Prev. Process. Ind. 65
wave, the arrangement point of the explosion-proof valve can be https://doi.org/10.1016/j.jlp.2020.104134.

10
L. Pang et al. Journal of Loss Prevention in the Process Industries 74 (2022) 104662

Launder, B.E., Spalding, D.P., 1974. The numerical computation of turbulent flows. Taveau, J., 2010. Correlations for blast effects from vented dust explosions. J. Loss Prev.
Comput. Methods Appl. Mech. Eng. 3, 269–289. Process. Ind. 23, 15–29.
Lunn, G., Crowhurst, D., Hey, M., 1988. The effect of vent ducts on the reduced explosion Tascón, A., Aguado, P.J., 2015. CFD simulations to study parameters affecting dust
pressures of vented dust explosions. J. Loss Prev. Process. Ind. 1, 182–196. explosion venting in silos. Powder Technol. 272, 132–141.
Li, G., Yang, H.X., Yuan, C.M., Eckhoff, R.K., 2016. A catastrophic aluminium-alloy dust Tascón, A., Ramírez-Gómez, Á., Aguado, P.J., 2016. Dust explosions in an experimental
explosion in China. J. Loss Prev. Process. Ind. 39, 121–130. test silo: influence of length/diameter ratio on vent area sizes. Biosyst. Eng. 148,
Ma, Q.J., Zhang, Q., Li, D., Chen, J.C., Ren, S.Y., Shen, S.L., 2015. Effects of premixed 18–33.
methane concentration on distribution of flame region and hazard effects in a tube Tascón, A., Aguado, P.J., 2017. Simulations of vented dust explosions in a 5 m3 vessel.
and a tunnel gas explosion. J. Loss Prev. Process. Ind. 34, 30–38. Powder Technol. 321, 409–418.
NFPA 68, 2018. Standard on Explosion Protection by Deflagration Venting. National Fire Taveau, J., 2017. Dust explosion propagation and isolation. J. Loss Prev. Process. Ind. 48,
Protection Association, Quincy. 320–330.
Popat, N.R., Catlin, C.A., Arntzen, B.J., Lindstedt, R.P., Hjertager, B.H., Solberg, T., Taveau, J., Hochgreb, S., Lemkowitz, S., Roekaerts, D., 2018. Explosion hazards of
Saeter, T., Van den Berg, A.C., 1996. Investigation to improve and assess the aluminum finishing operations. J. Loss Prev. Process. Ind. 51, 84–93.
accuracy of computational fluid dynamics based explosion models. J. Hazard Mater. Taveau, J., Lemkowitz, S., Hochgreb, S., Roekaerts, D., 2019. Metal dusts explosion
45 (1), 1–25. hazards and protection. Chem. Eng. Trans. 77, 7–12.
Proust, C., 1996. Dust explosions in pipes: a review. J. Loss Prev. Process. Ind. 9, VDI 3673, 2002. Pressure Venting of Dust Explosions, Part. 1. Verein Deutscher
267–277. Ingenieure, Berlin.
Pang, L., Hu, Q.R., Zhao, J.J., Lv, P.F., Sun, S.H., Yang, K., 2019. Numerical study of the Wang, S.M., Li, G.Q., Guo, H., Li, X.D., Pu, X.Y., Ren, H.Y., Wu, D.J., 2020. Experimental
effects of vent opening time on hydrogen explosions. Int. J. Hydrogen Energy 44, study on vented deflagration of hydrocarbon fuel-air mixtures in a 20-L semi-
15689–15701. confined cylindrical vessel with a slight static activation pressure. J. Loss Prev.
Pang, L., Zhang, Z.W., Cui, S.Q., Sun, S.H., 2020. Experimental study of the venting Process. Ind. 64 https://doi.org/10.1016/j.jlp.2020.104091.
characteristics of dust explosion through a vent duct. J. Loss Prev. Process. Ind. 65 Yan, X.Q., Yu, J.L., 2014. Dust explosion venting of small vessels at the elevated static
https://doi.org/10.1016/j.jlp.2020.104144. activation overpressure. Powder Technol. 261, 250–256.
Pang, L., Cao, J.J., Zhao, Y., Yuan, C.M., Yang, K., Zhang, Z.W., 2021. Flame propagation Yang, K., Hu, Q.R., Sun, S.H., Lv, P.F., Pang, L., 2019. Research progress on multi-
behaviours and overpressure characteristics of LDPE dust/ethylene hybrid mixture overpressure peak structures of vented gas explosions in confined spaces. J. Loss
explosions. Combust. Sci. Technol. https://doi.org/10.1080/ Prev. Process. Ind. 62 https://doi.org/10.1016/j.jlp.2019.103969.
00102202.2021.1909006. Yang, K., Liu, W.S., Ren, J.Q., Li, S.G., Zhao, Y., Hu, Q.R., Pang, L., 2020a. Effects of
Rocourt, X., Awamat, S., Sochet, I., Jallais, S., 2014. Vented hydrogeneair deflagration in multiple annular obstacles on flame propagation of local corn starch dust in a
a small enclosed volume. Int. J. Hydrogen Energy 39, 20462–20466. vertical pipe. J. Loss Prev. Process. Ind. 64 https://doi.org/10.1016/j.
Reding, N.S., Shiflett, M.B., 2020. Consequence prediction for dust explosions involving jlp.2020.104077.
interconnected vessels using computational fluid dynamics modeling. J. Loss Prev. Yang, K., Lv, P.F., Gao, J.C., Pang, L., 2020b. Influence of the region outside a vent on the
Process. Ind. 65 https://doi.org/10.1016/j.jlp.2020.104149. explosion of an indoor gas. J. Eng. Phys. Thermophys. 93 (2), 466–473.
Skjold, T., Arntzen, B.J., Hansen, O.R., Taraldset, O.J., Storvik, I.E., Eckhoff, R.K., 2005. Zalosh, R., Grossel, S., Kahn, R., Sliva, D., 2005. Dust Explosion Scenarios and Case
Simulating dust explosions with the first version of DESC. Process Saf. Environ. Histories in the CCPS Guidelines for Safe Handling of Powders and Bulk Solids. 39th
Protect. 83, 151–160. AIChE. Loss Prevention Symposium Session on Dust Explosions. Atlanta, April 2005.
Skjold, T., Arntzen, B.J., Hansen, O.R., Storvik, I.E., Eckhoff, R.K., 2006. Simulation of Zalosh, R., 2009. Dust Explosion Fundamentals: Ignition Criteria and Pressure
dust explosions in complex geometries with experimental input from standardized Development. The Fire Protection Research Foundation, Wellesley, MA.
tests. J. Loss Prev. Process. Ind. 19, 210–217. Zhang, X.Y., Yu, J.L., Gao, W., Zhang, D.W., Sun, J.H., Guo, S., Dobashi, R., 2017. Effects
Skjold, T., 2007. Review of the DESC project. J. Loss Prev. Process. Ind. 20, 291–302. of particle size distributions on PMMA dust flame propagation behaviors. Powder
Skjold, T., 2014. Flame Propagation in Dust Clouds: Numerical Simulation and Technol. 317, 197–208.
Experimental Investigation. Ph.D. Dissertation. University of Bergen, Norway. Zhang, Q., Yan, Q., 2019. Experimental study on explosion process of flour deposits/air
Song, Y.F., Zhang, Q., Wang, B., Li, J.C., 2019. Numerical study on the internal and mixture in horizontal pipelines. Powder Technol. 346, 273–282.
external flow field of dust explosion venting. Int. J. Therm. Sci. 145 https://doi.org/ Zhang, S.L., Bi, M.S., Yang, M.R., Gan, B., Jiang, H.P., Gao, W., 2020. Flame propagation
10.1016/j.ijthermalsci.2019.106008. characteristics and explosion behaviors of aluminum dust explosions in a horizontal
pipeline. Powder Technol. 359, 172–180.

11

You might also like