Defence Technology 18 (2022) 1269e1280

Contents lists available at ScienceDirect

Defence Technology
journal homepage: www.keaipublishing.com/en/journals/defence-technology

Analysis of temporal and spatial distribution characteristics of

ammonium chloride smoke particles in confined spaces
Chen-youshi Xu, Bing-ting Zha*, Jia-qian Bao, He Zhang, Hong-xia Li
ZNDY of Ministerial Key Laboratory, Nanjing University of Science and Technology, Jiangsu, Nanjing, 210094, China

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

Article history: In response to the demand for short-range detection of anti-smoke environment interference by laser
Received 12 June 2021 fuzes, this study proposes a smoke environment simulation of non-uniform continuous point source
Received in revised form diffusion and investigates an experimental laboratory smoke environment using an ammonium chloride
9 September 2021
smoke agent. The particle size distribution, composition, and mass flow distribution of the smoke were
Accepted 26 September 2021
studied. Based on a discrete phase model and a k
ε turbulence model, a numerical simulation was
Available online 29 September 2021
developed to model the smoke generation and diffusion processes of the smoke agent in a confined
space. The temporal and spatial distribution characteristics of the smoke mass concentration, velocity,
Keywords:
Laser fuze
and temperature in the space after smoke generation were analyzed, and the motion law governing the
Smoke environment smoke diffusion throughout the entire space was summarized. Combined with the experimental veri-
Temporal distribution fication of the smoke environment laboratory, the results showed that the smoke plume changed from
Spatial distribution fan-shaped to umbrella-shaped during smoke generation, and then continued to spread around.
Confined space Meanwhile, the mass concentration of smoke in the space decreased from the middle outward; the
Smoke agent changes in temperature and velocity were small and stable. In the diffusion stage (after 900 s), the mass
concentration of smoke above 0.8 m was relatively uniform across an area of smoke that was 12 m thick.
The concentration decreased over time, following a consistent decreasing trend, and the attenuation was
negligible in a very short time. Therefore, this system was suitable for conducting experimental research
on laser fuzes in a smoke environment. Owing to the stability of the equipment and facilities, the setup
could reproduce the same experimental smoke environment by artificially controlling the smoke
emission of the smoke agent. Overall, this work provides a theoretical reference for subsequent research
efforts regarding the construction of uniform smoke environments and evaluating laser transmission
characteristics in smoky environments.
© 2021 China Ordnance Society. Publishing services by Elsevier B.V. on behalf of KeAi Communications
Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).

1. Introduction complex smoke environments that exist in real battlefields, leading

to false alarms and early explosions [3]. Such complex smoke en-
Laser detection is widely adopted in proximity fuzes owing to its vironments contain dust kicked up by the movement of tank armor,
high coherence, distance accuracy, rapid response, and anti- the hot smoke from gun muzzles after firing ammunition, the dust
electromagnetic interference [1,2]. Laser proximity fuzes have generated by the recoil after firing ammunition, the smoke from
played an important role in improving the hit rate and destructive ammunition explosions, and the smoke pot and smoke bombs used
effect of modern precision-guided weapons, especially ground-to- in the tactical combat of armored clusters. Therefore, the influence
air missiles, air-to-air missiles, and anti-radiation missiles. They of the smoke environment on laser detection cannot be ignored.
are widely used in conventional ammunitions and missiles; how- The interference effect on laser-guided ammunition is mainly
ever, short-range laser detection is susceptible to interference from related to the concentration and thickness of the smokescreen, as
well as the composition and size of the particles [4]. Therefore,
ongoing anti-environmental interference research efforts primarily
focus on evaluating the environmental smoke concentration in
* Corresponding author.
which laser proximity fuzes can work effectively.
E-mail address: [email protected] (B.-t. Zha).
Peer review under responsibility of China Ordnance Society
There are a limited number of studies and publications

https://doi.org/10.1016/j.dt.2021.09.017
2214-9147/© 2021 China Ordnance Society. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-
ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

describing quantitative evaluations of the smokescreen effect for wind direction and particle mass flow rate on dust concentration
laser-guided weapons interference. In particular, the interference and temporal and spatial distribution [22]. Because field test con-
ability of smokescreens in actual combat is not well understood. ditions are harsh and uncontrollable, there is still a lack of experi-
Most reports do not mention the smoke particle composition or the mental data to support such studies. Ma et al. used an
particle size distribution of the smoke agent [5e10]. However, instantaneous point source to simulate explosion smoke based on
different smoke particle compositions and particle sizes have the Gaussian smoke model. They obtained the smoke concentration
distinct light reflectivity characteristics, which also lead to differ- distribution law but did not consider the sedimentation effect of
ences in laser penetration performance [11,12]. smoke particles [23]. Based on the mechanism of gas-solid two-
The currently available strategies for evaluating the interference phase flow, Ding et al. studied the spatial diffusion of smoke in
effect of smokescreens include the experimental test evaluation microgravity and high-vacuum environments [4]. Zeng et al. con-
method (ETEM) and the theoretical calculation evaluation method structed a vertical smokescreen spatial scale calculation model
(TCEM). Based on the environmental testing conditions, the ETEM through analytical geometric analysis, which was demonstrated to
can be classified as a laboratory smokescreen box evaluation be suitable for infrared shielding smokescreens [24]. Xiao et al.
method, a small- and medium-sized wind tunnel evaluation proposed a method that combines the wind field results calculated
method, or a field evaluation method. The laboratory smokescreen by the computational fluid dynamics method with the particle
box evaluation method is most commonly used because of its system to simulate the diffusion of aerosols in a complex envi-
controllable test conditions, reproducible test environment, and ronment. However, this method did not provide a specific analysis
safety compared with the other methods. Shen et al. created a of the particle forces, and the study finally obtained a graph of the
smokebox that could discharge a burning smoke agent, spraying transparency of the particle system, which could not get the spe-
powder, and spraying laser chaff, but the authors did not describe cific particle concentration distribution [38]. HE et al. established a
the smoke concentration distribution in the box [13]. Xie et al. used geometric model of the flow field and derived the correlation law of
a closed glass box filled with smoke to simulate a smoke environ- wind speed and wind direction on smoke diffusion through
ment [14]. Although an air quality detector was placed inside the simulation. But the study did not describe the physical character-
box to measure the concentration of particles in the air, its small istic parameters of smoke particle and external field experimental
size limited the thickness of the simulated smoke. Zhang et al. verification, and the correctness needs to be considered [39].
established a 10-m-long pipe-type device to study the relationship The aforementioned studies using ETEMs do not consider the
between the laser echo signal and the thickness of the smoke. temporal or spatial distribution of smoke in the smoke environ-
However, because of the random distribution of smoke in the pipe, ment, nor do they investigate the diffusion process of smoke; these
the influence of smoke concentration on their conclusions was oversights mean that the smoke concentration parameters at a
ignored [15]. In another study a quantified amount of smoke par- specified location cannot be determined. Additionally, because
ticles was blown into a closed test box [16]. However, in these cases, TCEMs are impacted by the calculation speed and the selected type
they either replaced the actual concentration with the average of model, calculation models that ignore some influencing factors
arithmetic concentration or replaced the smoke concentration with may lead to significant errors between the conclusions obtained by
the subjective visibility value, neither of which is a sufficiently the TCEM and the actual experimental data. Therefore, to under-
rigorous approach. The real battlefield smoking environment is stand whether laser proximity fuzes can work reliably in a smoky
very complicated and the space occupied by the smoke is quite environment, it is necessary to combine ETEM with TCEM and then
large, so it is difficult to carry out data measurement. And the set up a smoke environment laboratory to study the temporal and
smoking environment generated under the same conditions is spatial distribution characteristics of smoke.
inconsistent every time. It is impossible to carry out related ex- The rest of this article is organized as follows. Section 2 estab-
periments with laser fuzes in such an uncontrollable smoky envi- lishes a mathematical model of smoke diffusion and deduces the
ronment. Therefore, it is necessary to establish an environmental force equation that governs the investigated smoke particles. Sec-
laboratory with adjustable smoking concentration and provide an tion 3 analyzes the smoke particle composition, particle size dis-
experimental environment for laser anti-smoke experiments. tribution, and combustion mass loss distribution, and describes
Considering the complexity of the battlefield smoke environ- numerical simulations of the temporal and spatial characteristics of
ment, conventional TCEMs are not mature, and most of them adopt the smoke concentration, velocity, and temperature. Section 4
computer numerical simulation methods. The diffusion of smoke- verifies the accuracy of the smoke diffusion model and numerical
screens can be categorized as continuous point source diffusion simulation by using a smoke environment laboratory and con-
(e.g., smoke agents) or instantaneous body source diffusion (e.g., ducting experimental research and analysis. The described simu-
smoke bombs) [17]. Xu et al. proposed a smoke laboratory design lation supports the experimental results. Finally, Section 5 presents
scheme to determine the distribution of smoke concentration and the conclusions drawn from the results obtained in this work.
the concentration over time for different cross-sections in the
laboratory using the computational fluid dynamics (CFD) software, 2. Smoke diffusion mathematical model
FLUENT. However, since the boundary condition was a slender
rectangular inlet, this was difficult to achieve [18,19]. Hao et al. There are two methods for studying the propagation and
combined a CFD method with a discrete phase model to simulate deposition of aerosols: the Euler method and the Lagrange method.
the diffusion of smoke particles, and they deduced the temporal The Euler method regards the particle phase as a pseudo-fluid that
and spatial distribution of the mass concentration of an interpenetrates the fluid; this approach uses the Euler equation for
explosively-dispersed smoke bomb. However, this approach lacked both the gas phase and the particle phase. It comprehensively
experimental verification [20]. Zhou et al. studied the influence of considers the speed slip between the two phases, particle diffusion,
ambient wind speed, ventilation position, and fire source on the phase coupling, and the particles' effect on the fluid [25]. The
pressure distribution and smoke movement in single- and multi- Lagrange method emphasizes the individual behavior of each par-
chamber fires through a series of numerical simulations [21]. ticle, and it can determine the particle trajectory according to the
Yang et al. proposed a method for determining the temporal and equation of motion. This method regards the particle phase as a
spatial distribution of dust during tank movement based on CFD discrete phase and only the gas phase as a continuous phase. In
discrete phase dispersion theory. They studied the influence of addition to the interactions between particles and gas, the
1270
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

interactions between particles and particles are also considered in continuous flow of heat in each phase; p is the flow field pressure; m
this approach, which uses the Lagrange equation to describe par- is the viscosity coefficient of laminar flow; mt is the turbulence
ticle motion and the Euler equation for the gas component [26]. The viscosity coefficient (defined as mt ¼ rCm k2 =ε, where Cm ¼ 0:09); sk
movement of smoke in space is a typical gas-solid two-phase flow and sε are the Prandtl numbers corresponding to the turbulent
movement, which conforms to the discrete phase diffusion model kinetic energy (k) and the turbulent energy dissipation rate (ε),
in the Lagrange method [18]. respectively (sk ¼ 1, sε ¼ 1.3); Gk is the turbulent kinetic energy due
to the average velocity gradient; Gb is the turbulent kinetic energy
generated by buoyancy (for incompressible fluids, Gb ¼ 0); YM is
the fluctuation caused by excessive diffusion in compressible tur-
2.1. Mathematical model of the gas phase bulence (for incompressible fluids, YM ¼ 0); Sk and Sε are defined
turbulent kinetic energy; and C1ε C2ε , C3ε are empirical constants
During its ascent, the smoke is affected by the oncoming air (i.e., (C1ε ¼ 1.44, C2ε ¼ 1.92, C3ε ¼ 1).
air resistance and energy exchange) and the diffusion of the up-
ward flow itself. The diffusion process of smoke can be regarded as
an incompressible ideal gas. Therefore, the appropriate equation for 2.2. Mathematical model of the particle phase
describing the incompressible viscous fluid adopts the Navier-
Stokes equation (NeS) combined with an energy conservation The discrete phase diffusion model regards the gas phase as a
equation [27]. The turbulence state of smoke can be described by continuous phase and the particle phase as a discrete phase. It also
the Reynolds time-averaged Navier-Stokes equation (RANS); the considers the particle-gas and particle-particle interactions and
follows the Euler-Lagrange equation. The force equation on the
standard k
ε model is therefore selected. This model has high
particle is shown in Eq. (4) [31],
calculation accuracy and is mainly used to calculate complex tur-
bulent flows. It is distinct from the single equation because it  
dup   gx , rp
r
combines the turbulent dissipation rate ε with the turbulent kinetic ¼ FD , ua
up þ þ Fx (4)
energy k. The mathematical model [28e30] involves the mass, dt rp
energy, and momentum conservation equations (Eq. (1)),
where up is the particle velocity, ua is the gas velocity, FD ,ðua
up Þ
8
> vr v is the unit mass drag force of particles (FD ¼ r18dm2 ,C24
D Re
), rp is the
>
> þ ðrui Þ ¼ Sm
>
p p
>
> vt vxi a2
particle density, dp is the particle diameter, CD ¼ a1 þ Re þ Re a3
> 2 , Re is
>
>    
>
< v  v   v vT the Reynolds number (Re ¼ rdp ua
up =m), r is the particle den-
rcp T þ rui cp T ¼ l þ Sh
> vt vxi vxi vxi sity, and gx ,ðrp
rÞ=rp is the difference between the unit mass
>
> " !#
>
> gravity and buoyancy of particles. Within a certain range of Rey-
>
> v v   vp v vuj vui
>
> ðrui Þ þ rui uj ¼
þ ðm þ mt Þ þ nolds numbers, a1 , a2 , and a3 of the spherical particles are con-
>
: vt vxj vxi vxj vxi vxj stants. The term Fx represents the additional forces on the smoke
(1) particles per unit mass, which generally include the pressure
gradient force, Basset force, virtual mass force, Saffman lift force,
the turbulent kinetic energy equation (Eq. (2)), and Magnus lift force, among others [32,40]. The present study only
considered solid particles, and assumed that the particle shape was
"  # a simplified sphere, so the pressure gradient force was not
v v v m vk
ðrkÞ þ ðrkui Þ ¼ mþ t þ Gk þ Gb
rε
YM þ Sk considered. Because the Basset force and the virtual mass force
vt vxi vxj sk vxj
were smaller in magnitude than the others, they were not
(2) considered. Since the particle rotation was ignored, the Magnus lift
force was not considered [31]. Therefore, only the drag force, the
and the turbulent kinetic energy dissipation rate equation (Eq. (3)), gravity, the buoyancy, and the Saffman lift were considered in this
study, and incorporating these factors into Eq. (4), gives Eq. (5):

         
dup r 3 ua
up m2 a3 þ ua
up ,rdp ma2 þ ua
up ,a1 rdp
¼ gx 1
þ   2 þ Fx (5)
dt rp 4ua
up ,d r rdp p p

"  #
v v v mt vε
ðrεÞ þ ðrεui Þ ¼ mþ The diameter dp of the smoke particles is not uniform and the
vt vxi vxj sε vxj particle size distribution satisfies a certain distribution function
(3)
ε ε2 f ðdÞ; and the particle composition also has an effect on the calcu-
þ C1ε ðGk þ C3ε Gb Þ
C2ε þ Sε lation. The particle size distribution of particulate systems typically
k k
conforms to a specific distribution law, most of which are two-
In these equations, r is the gas density; xi is the tensor repre- parameter distribution functions [33,34]. Commonly used two-
sentation of the turbulence model; i is the index sign of the tensor parameter distribution functions include the Rosin-Rammler (R-
(i ¼ 1, 2, or 3, which represent the x, y, and z directions, respec- R) function, the normal function, the normal logarithmic function,
tively); ui is the component of the flow velocity in direction i; Sm is and the upper limit logarithmic normal function. The specific
the particle mass of continuous inflow; cp is the isobaric specific analysis is studied in Section 3.2.
heat capacity; T is the flow field temperature; Sh is the sum of the In addition, the discrete random walk (DRW) model was used to
1271
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

Fig. 1. Physical model of the confined space smoke laboratory.

describe the random pulsating velocity in the flow field [29]. The
DRW model assumes that the random pulsating speed follows a
Gaussian probability density distribution. The pulsating velocity Fig. 3. Size distribution of smoke particles.

component, u0i , is given by Eq. (6), where xi is a random number that

obeys a normal distribution.
3. Numerical methods
qffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffi
u0i ¼ xi u0 2i ¼ xi 2k=3 (6) 3.1. Numerical model

All walls in confined space (except the bottom surface and As shown in Fig. 1, the confined space was a cuboid with a length
smoking inlet) boundary conditions are assumed to be slip-free of 12 m, a width of 2 m, and a height of 2.5 m. The smoke agent was
walls, and the motion of particles in confined space boundary a cylinder with a diameter of 70 mm and a thickness of 15 mm,
conditions is elastic collision with collision recovery coefficients as placed on a bracket of the same diameter. The bracket ensured that
in Eq. (7), the upper surface of the smoker emit smoke normally and did not
interfere with the occurrence and diffusion of smoke. The con-
8 v1; structed confined space physical model comprised the computa-
>
> w ¼ n
>
< n v2; tional domain of the numerical simulations.
n
(7) The reference coordinate system O-XYZ is shown in Fig. 1; the
>
> v1; t
>
: wt ¼ smoke agent was placed in the center of the ground in the confined
v2; t space, and the bottom center of the smoke agent was located at the
coordinate origin, O. Tetrahedral meshing of the geometric model
where wn and wt are the normal and tangential recovery co- was performed. To meet the requirements in terms of high mesh
efficients for the collisional interaction of the particle with the quality and high simulation accuracy, the convergence speed of the
boundary, respectively, vn and vt are the normal and tangential calculation increased as much as possible. Since the velocity and
velocity components of the particle, the subscripts 1 and 2 temperature gradient at the smoke agent's outlet were more sig-
respectively indicate the quantities before and after the collision. A nificant than those in other areas, the nearby grids were encrypted.
normal (tangential) recovery factor equal to 1.0 means that the The area far from the smoke agent had relatively low requirements
particle has no momentum loss before and after the collision, i.e., a in terms of calculation accuracy, and the grid was gradually
fully elastic collision and equal to 0 means that all momentum is increased at a specific growth rate to reduce the calculation burden.
lost after the collision. The mesh model was converted to polyhedral elements and sub-
Based on thermal gravimetric analysis (TGA; SDTA851e), it is jected to a smoothing process. In Fig. 2, the grids of the computa-
determined that the smoking process of the smoke agent is non- tional domain are presented. The final number of high-quality grids
uniform, and the combustion mass loss (i.e., the mass flow rate) (154893) ensured the accuracy and convergence of the solution.
generally satisfies a Gaussian distribution. The mass flow rate of
particles goes from slow to faster and then gradually slows down
after reaching a certain peak. Therefore, the setting of mass flow 3.2. Boundary conditions
rate in this paper adopts Gaussian distribution instead of the
traditional uniform distribution, which makes the calculation The smoke particles were sampled and recovered, and then, a
closer to the actual value. laser particle size analyzer (HORIBA LA950) was used to test and

Fig. 2. Grids of the computational domain: (a) x ¼ 0 m section; (b) y ¼ 0 m section.

1272
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

largest particle size was 29.907 mm, and the average particle size
was 10.723 mm.
According to the test results and Fig. 3, the particle size distri-
bution of smoke particles should follow the R-R distribution. The
differential distribution function [35,36] of this function f ðdÞ is
described by Eq. (8),

 n
1  n
n d d
f ðdÞ ¼ exp
(8)
d d d

where n is the spread parameter, d is the average particle diameter,

and d is the particle diameter. The cumulative distribution function
[37] of smoke particles, FðdÞ, can be obtained by integrating the
differential distribution function using the expression in Eq. (9):

Fig. 4. Sample characterization diagram: XRD patterns and SEM image.

 n
d
FðdÞ ¼ 1
exp
(9)
d
Table 1 From the results shown in Fig. 3, the spread parameter
Simulation conditions and parameters.
(n ¼ 2.713) and the average particle diameter (d ¼ 10.723 mm) were
Condition Parameter Type or Value calculated.
Gas phase Composition Air X-ray diffraction (XRD) analysis is the primary method used to
Initial temperature/K 298.15 study the phase and crystal structure of materials. The recovered
Initial pressure/Pa) 101325
smoke particles were observed using a Bruker D8 ADVANCE X-ray
Discrete phase Bulk density/(kg∙m
3) 1527 diffractometer, and the composition of the smoke particles was
Specific heat/(J∙kg
1∙K
1) 1553
confirmed to be ammonium chloride (Fig. 4) based on the powder
Velocity/(m∙s
1) 4
Temperature/K 733.15
diffraction cards. A scanning electron microscope (SEM; QUANTA
Starting time/s 0 FEG250) was used to observe the morphology of the smoke parti-
End time/s 300 cles. In Fig. 4, the abscissa 2q is the diffraction angle, and the
Mass flow rate Gaussian distribution ordinate is the intensity.
Boundary Export of smoke agent Velocity-inlet According to the test results, the temperature of the smoke
Inlet velocity/(m∙s
1) 4 agent while it was smoking was approximately 460  C, the mass
Inlet Temperature/k 733.15
loss of ammonium chloride following the reaction was 15 g, and the
Wall of smoke agent Wall
Other walls Wall
smoke lasted for 5 min. Other simulation conditions and parame-
ters are shown in Table 1. Since the smoke velocity of ammonium
chloride smoke agent was obtained by high-speed photography
and calculation, the discrete phase velocity was set to 4 m/s. The
analyze the smoke particles. The particle size distribution results
above mathematical model established in this paper is valid for
are shown in Fig. 3. The smallest particle size was 2.976 mm, the
single-component smoke screens.

Fig. 5. Smoke concentration distribution over time (x ¼ 0 m cross-section).

1273
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

the positive direction of the z-axis in a fan shape, and the plume
spread to the top surface at 5 s. The space was closed, so there was
no lateral wind interference or smoke leakage. The smoke
continued to generate and accumulate at the upper surface, and the
plume changed from a fan shape to an umbrella shape at 10 s. The
outlet of the smoke agent had the highest concentration (5.7e6.0 g/
m3), and the concentration decreased as the height increased.
Additionally, the smoke sedimentation mainly occurred on both
sides of the smoke agent's outlet. As the smoke moved away from
the outlet, it began to sink gradually due to gravity. Some smoke
particles exchanged momentum with newly generated smoke
particles during the sedimentation process, which changed their
direction of movement. In the latter half of the smoking process, the
smoke had filled the entire cross-section of the space and the
thickness of smoke is 12 m. The state of smoke at the smoke agent
outlet was more complicated because it including the rising motion
of newly generated smoke and the secondary rising motion of
previously generated smoke.
The process of smoke filling the entire space is clearly shown in
Fig. 6. The same plume changed from a fan shape to an umbrella
shape, and then slowly spread to both sides of the x-axis. The mass
concentration of the entire space decreased from the middle out-
ward, and the smoke mass concentration distribution in the posi-
tive half axis of the x-axis was almost symmetrical with that in the
opposing half axis. Overall, the mass concentrations at other loca-
tions were the same, except just above the outlet of the smoke
agent.
The temperature field and velocity field distributions of the
smoke agent smoking process are shown in Fig. 7 and Fig. 8,
respectively. Because the space was closed, the temperature and
Fig. 6. Smoke mass concentration distribution over time (y ¼ 0 m cross-section).
velocity of the whole smoking process remained stable. The sec-
ondary or even later rising effects of a small amount of smoke
3.3. Analysis of simulation results caused the velocity field distribution in the latter half of the
smoking process to vary. However, its influence could be ignored by
3.3.1. The smoking process analyzing the concentration distribution presented in Figs. 5 and 6.
The smoking process lasted for 300 s, and throughout this
period, the smoke state was observed from the x ¼ 0 m cross-
3.3.2. The smoke diffusion process
section and the y ¼ 0 m cross-section. The smoke mass concentra-
At the end of the smoke generation (i.e., after 300 s), one plane
tion (i.e., the diffusion behavior of the smoke) distributions over
was monitored along the x-axis at 1-m intervals to observe the
time at these two cross-sections are shown in Fig. 5 and Fig. 6,
smoke mass concentration distribution (diagram in Fig. 9),
respectively. Specifically, Fig. 5 shows that the smoke moved along
comprising a total of 13 observation surfaces. The smoke mass

Fig. 7. Smoke velocity distribution over time (x ¼ 0 m cross-section).

1274
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

Fig. 8. Smoke temperature distribution over time (x ¼ 0 m cross-section).

Compared with other smoke components such as zinc oxide, there

are more research about ammonium chloride. And NH4Cl smoke
agent is safety and commonly used, so it is suitable for research in a
closed laboratory environment. Therefore, in this study, a specific
type of environmentally friendly and non-toxic white smoke agent
(NH4Cl smoke agent) was used. In future work, the transmission
characteristics of lasers in this smoke environment and the
parameter mapping relationship between this smoke and other
types of smoke will be studied. Additionally, the transmission
characteristics of lasers in other smoke environments will be
analyzed.
The required smoke environment was created by igniting a
specific dose (that could be adjusted manually) of the NH4Cl smoke

Fig. 9. Smoke mass concentration distribution along the x-axis.

concentration distributions on the positive and negative semi-axes

of the x-axis were similar, and the further from the x ¼ 0 m cross-
section, the lower the mass concentration. The concentrations at
other positions were very close, except at x ¼ 0.
The free diffusion process of smoke in the laboratory is shown in
Fig. 10, which describes the distribution of mass concentration at
times t ¼ 500, 1500, 2500, 3000, and 3500 s. The smoke mass
concentration in the space decreased over time, and it was rela-
tively uniform at heights greater than 0.8 m. The original position
where the smoke agent outlet was located had a higher mass
concentration than the surrounding positions. The thickness of the
smoke is continuously maintained at 12 m.

4. Experimental validation

4.1. Smoke environment laboratory

To create a relatively stable smoke environment, a smoke

environment laboratory with three functional zones (i.e., control
room, test room, and settling room (Fig. 11)) was designed and
constructed to verify the simulation model established in this
research.
Actually, smoke screen usually contains a variety of components,
such as NH4Cl and ZnO. Considering that most smoke bombs and
smoke pots are dangerous and toxic to humans and the environ-
ment, they do not meet standard laboratory safety regulations; Fig. 10. Smoke mass concentration distribution over time (y ¼ 0 m cross-section).

1275
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

Fig. 11. Layout of the smoke environment laboratory.

Fig. 12. Schematic diagram of the smoke environment laboratory.

agent in the test room. Then, equipment control, experimental the smoke release process in each measurement was approxi-
phenomenon observations, and recording processes were carried mately 5 min. For convenience, these four measurements were
out in the control room. Finally, by turning on an exhaust fan in the known as Smoke1, Smoke2, Smoke3, and Smoke4. In the first three
settling room, the air in the laboratory was purified after the measurements (i.e., Smoke1, Smoke2, and Smoke3), the CEL-712
experiment. In addition, we could adjust the length (i.e., thickness moved steadily along the light path (xL-axis) at a speed of 8 cm/s
of smoke) of the test chamber as needed. The test room was a long for 5 min (to minimize changes in the smoke concentration caused
and narrow space with a length of L ¼ 12 m, a width of W ¼ 2 m, a by movement) and the mass concentrations were recorded at
height of H ¼ 2.5 m, and Rx ¼ 6 m (Fig. 12). specific moments. Simultaneously, the FCMs continuously
measured the mass concentrations at particular locations (x ¼ 1, 1.5,
4.2. Experimental setup and procedure and 2 m, which were selected considering the dust environment
tests to be carried out in future laboratory experiments) after the
The environmental mass concentration of smoke in the labo- smoke environment was relatively stable, as shown in Fig. 13. In
ratory was measured with a real-time dust monitor (CEL-712 these measurements, the CEL-712 recorded data every second, and
Microdust Pro, CASELLA) over the measurement range from 0.001 the FCMs recorded every 3 s.
to 250 g/m3. In addition, there were three fixed-concentration
measuring (FCM) instruments, labelled FCM1, FCM2, and FCM3 4.3. Results of mass concentration measurements
(MODEL 2030), which monitored the changes in mass concentra-
tion along the y-axis. The four devices were arranged based on the The raw data from the mass concentration measurements are
light scattering method to measure the smoke concentration (CEL- shown in Fig. 14, which indicates that the changes in the mass
712 is more widely recognized in the industry). To reduce the concentration of the smoke agent during the smoking process were
relative measurement error between the devices, all of the FCMs complicated. However, the overall trend was that the concentration
were calibrated based on CEL-712 to achieve a unified standard in increased over time, reaching its highest value near 300 s (at the
the same smoke environment before beginning the experiment. end of smoking) and maintaining it for a short period of time. After
Under identical experimental environments (i.e., same humid- about 800 s, the smoke environment was stable, and the concen-
ity, same temperature, no wind), the mass concentration of the tration decreased steadily with the deposition, diffusion, and
smoke environment was measured four times, and the duration of adsorption of smoke. Fig. 14a shows that although the CEL-172 was

Fig. 13. Experimental setup of mass concentration measurements (a) inside and (b) outside the laboratory.

1276
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

Fig. 14. Mass concentration data measured by (a) CEL-172 along the xL-axis, (b) FCM1, (c) FCM2, and (d) FCM3.

Fig. 15. Mass concentration measurement data in (a) Smoke1, (b) Smoke2, (c) Smoke3, and (d) Smoke4.

1277
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

moved at a slower rate, it was impossible to avoid fluctuations in experiments are shown in Fig. 15. It was clear that the mass con-
the smoke concentration caused by relative movement. However, centrations at the three FCMs were very similar, and they all
due to the fixed positions of the FCMs, the fluctuations of their exhibited a similar downward trend. In addition, the attenuation
concentrations were relatively small (Fig. 14bed). Because the over a short time period was ignored.
laboratory space was closed and the temperature, humidity, and Data from two simulations and experiments are presented in
smoke agent dosage could be adjusted and controlled manually, a Fig. 16. The Smoke5 experimental data were obtained by doubling
consistent smoke experimental environment could be obtained, the smoke agent dose, and the simulation (Simulation2) was
although the two-phase flow of smoke particles was complex and completed using the model established in the article. The concen-
uncertain. tration data from the original simulation and Simulation2 were
The variations in environmental smoke concentration measured extracted (see Section 3.3) at the same locations as the three FCMs.
by the three FCMs (representing three specific locations) in four The test data from the FCM1 device and the simulation data
collected from the same location are shown in Fig. 16a. The rising
trend of the simulated and experimental concentrations during the
smoking process were very similar, and the maximum concentra-
tion obtained in the experiments were generally greater than those
determined in the simulation. Since FCM1 was closer to the smoke
source than the other two FCMs, the diffusion effect of the smoke
was more obvious at FCM1. Once its concentration reached its peak,
that concentration was maintained for a shorter time than at the
other two locations. In the smoke diffusion stage (i.e., during the
steady decline of smoke concentration), the simulation results were
consistent with the experimental results. In addition, the concen-
tration of smoke during the smoking stage increased significantly
by increasing the dose of smoke agent. However, the particles’ own
gravity caused the concentration during the first half of the diffu-
sion stage (300e800 s) to decrease faster than in the second half of
this phase. In contrast to the significant difference in peak con-
centrations during the smoking process, the smoke concentration
following a double-dose of the smoke agent was only slightly
higher than that of the single-dose of smoke agent at the same time
(after 2500 s). The experiment paid more attention to the smooth
diffusion process, relative to the smoke phase. During the diffusion
process (800e3600 s), the average relative errors of the experiment
and simulation at the three fixed positions were all less than 10%
(9.85%, 6.35%, and 7.67%, respectively). Fig. 16bec describe the same
situation as Fig. 16a, showing the consistency between the three
detection devices in the same experimental environment, thus
verifying the stability of the smoke environment.
Therefore, it was considered that the calculation results of the
simulation model established in this report and the experimental
results were consistent, thus confirming that the simulation model
was correct and feasible. The cost of simulation calculations is
much lower than the cost of experiments, and the temporal and
spatial changes in the smoke concentration in the entire space can
be observed more intuitively through simulation. Therefore, we
propose that future experimental processes should be conducted as
follows: first, the temporal and spatial distribution of smoke and
dust concentrations should be analyzed through simulation

Fig. 16. Mass concentration data at position (a)x ¼ 1 m, (b) x ¼ 1.5 m, and (c) x ¼ 2 m. Fig. 17. Experimental scene of laser fuzes.

1278
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

calculations. Then, the experimental conditions (temperature, hu- for Basic Scientific Research (Grant No. 30918012201), the Foun-
midity, smoker dose, etc.) should be set according to the simulation dation of JWKJW Field (Grant 2020-JCJQ-JJ-392), and the Post-
parameters. Finally, the ammonium chloride smoke agent can be graduate Research & Practice Innovation Program of Jiangsu
ignited to generate an actual smoke environment and carry out Province (Grant No. KYCX20_0315).
experimental research and analysis. In general, it was beneficial to
use the y ¼ 0 section as the optical path plane in these experiments References
and to select a stable and uniform smoke environment at the
appropriate time for the laser fuze experiments (Fig. 17). [1] Affan AS, Mohsin M, Zubair ASM. Survey and technological analysis of laser
and its defense applications. Defence Technol 2021;17(2):583e92.
[2] Zhang H, Li HX, Ding LB, Zha BT. Modeling and simulation of multi-detection
5. Conclusions point optimal initiation of synchronous scanning panoramic pulse laser fuze.
Infrared Laser Eng 2020;49(4):9e15.
[3] Bao JQ, Zha BT, Zhang H, Xu CYS. Simulation method of pulse laser fuze echo in
This report detailed a numerical simulation study on the gas-
dust environment. Infrared Laser Eng 2020;50(5):1e10.
solid two-phase flow of smoke in a confined space based on a [4] Ding GZ, Zhang ZY, Zhou SY, Tang WL. Study on diffusion mechanism and
CFD discrete phase model. A method was proposed for creating a attenuation performance of space smoke-screen. Infrared Technol
smoke environment using a smoke agent. This approach enabled 2014;36(11):914e9.
[5] Thiruvengadam M, Zheng Y, Tien JC. DPM simulation in an underground en-
detailed investigations of the temporal and spatial distribution of try: comparison between particle and species models. Int J Mining Sci Technol
the smoke's mass concentration, velocity, and temperature during 2016;26(3):487e94.
the smoking process and the diffusion stage. Following experi- [6] Nebuloni R. Empirical relationships between extinction coefficient and visi-
bility in fog. Appl Opt 2005;44(18):3795e804.
mental verification of the smoke environment laboratory, the main [7] Zeng ZK, Zhu DS, Guo XD, Wang D, Gao XH. Research on anti-aircraft smoke
conclusions are as follows: equipment position disposition model based on infrared shielding. Acta
Armamentarii 2019;40(6):1244e51.
[8] Gao W, Sun YF, Li J, Dang DN, Wang SH. Field test of smokescreen jamming
(1) The key factors that interfere with laser detection include effect on TV seekers. Acta Photonica Ainica 2014;43(10):1011006.
smoke concentration, composition, particle size, and thick- [9] Wang ZG, Guo SX. Simulation of non-lethal efficiency of tear bomb aerosol
ness. Smoke particle collection and X-ray diffraction analysis smoke based on Gaussian diffusion model. Chin J Energetic Mater 2019;27(2):
113e8.
were used to verify that the smoke particles were mainly [10] Feng CG, Qiao XJ, Li WC. Research progress in smoke bombs. Sci Technol Rev
composed of ammonium chloride. The distribution function 2014;32(4/5):110e5.
of smoke particle size was obtained, and the thickness of the [11] Noh YM, Müller D, Mattis I, Lee H, Kim YJ. Vertically resolved light-absorption
characteristics and the influence of relative humidity on particle properties:
smoke environment created in this work was 12 m, which
multiwavelength Raman lidar observations of East Asian aerosol types over
could be expanded to 18 m in the future. Korea. J Geophys Res 2011;116(D6).
(2) This work enabled clear observations of the smoke's tem- [12] Kokorin AM. How moisture in the air affects the light-scattering and -ab-
poral and spatial characteristics in the laboratory space. The sorption characteristics of radially inhomogeneous aerosol particles in the
boundary layer over the sea. J Opt Technol 2012;79(12):748e53.
shape of the plume changed from a fan shape to an umbrella [13] Shen T, Yang WJ, Yao ML, Liu ZG. Measurement of jamming effect of smoke on
shape and then spread around. The concentration, speed, and laser guided weapon based on semi-physical simulation. Acta Armamentarii
temperature at the smoker outlet were the highest, reaching 2015;36(11):2122e7.
[14] Xie RJ, Li ZQ, Gu ED, Huang QQ. Signal attenuation of visible light communi-
5.7e6 g/m3, 4.0e4.5 m/s, and 657e700 K, respectively, and cation in smoke environment. Opt Eng 2019;58(11):114102.
they all decreased as the height of the plume rose. There was [15] Zhang PD, Wang CY, Song WH, Yan W, Lai JC, et al. The effect of smoke spatial
almost no change in temperature or speed during the entire distribution on pulse laser echo characteristics. SPIE 2018;10827:108273I.
[16] Zhong HC, Zhou J, Du ZX, Li X. A laboratory experimental study on laser at-
smoking process, so the smoking state was considered rela- tenuations by dust/sand storms. J Aerosol Sci 2018;121:31e7.
tively stable. [17] Hua C, Liao SY, Zhang ZY, Li R, Chen HY. Simulation research on jamming
(3) The simulation law was verified through multiple experi- efficiency about the smoke screen against infrared seeker. Laser Infrared
2019;49(2):217e21.
ments, which confirmed that the smoke environment labo-
[18] Xu CYS, Zhang H, Zha BT, Zheng Z, Chen JY. Numerical Simulation of dust
ratory could generate a consistent smoky environment. After concentration distribution in confined space. Acta Armamentarii 2020;41(3):
the smoke diffusion stabilized (after ~900 s), the mass con- 618e24.
[19] Bao JQ, Zha BT, Xu CYS, Zhang H. Study of a laser echo in an inhomogeneous
centration was very similar, and the downward trend was
dust environment with a continuous field Monte Carlo radiative transfer
nearly consistent at positions higher than 0.8 m. It was also model. Opt Express 2021;29(12):17976.
determined that the attenuation in a very short time could be [20] Xu CL, Hao XY, Xiao KT, Song WW, Chen CS. Simulation study of screening
ignored. efficiency of explosive smoke bomb. Acta Armamentarii 2020;41(7):
1299e306.
[21] Zhou Y, Bu RW, Gong JH, Geng ZY, Fu HA, Yi L. Effect of ambient wind speed on
The results presented herein are of great military engineering pressure distribution and smoke movement in single and multiple compart-
significance, in particular, the analysis of the temporal and spatial ment fires. Combust Sci Technol 2019;191(8):1354e79.
[22] Yang SX, Chen HM, Gao LJ, Qi B, Guo PY, Deng JH. Study of spatial distribution
distribution of mass smoke concentration in a confined space, characteristics for dust raised by vehicles in battlefield environments using
creation of a stable and uniform smoke environment in a smoke CFD. IEEE Access 2021;9:48023e38.
laboratory, and detailed investigations of the functional charac- [23] Ma Chao, Chen HM, Wang FJ, Liu WB. Simulation of explosive smoke con-
centration distribution characteristic based on Gaussian puff model. Guidance
teristics of laser fuzes in such smoke environments. These aspects Fuze 2017;38(3):4e9.
also represent future research directions ongoing in our laboratory. [24] Zeng ZK, Zhu DS, Guo XD, Wang D, Gao XH. Research on antiaircraft smoke
equipment position disposition model based on infrared shielding. Acta
Armamentarii 2019;40(6):1244e51.
Declaration of competing interest [25] Zhang Z, Chen Q. Comparison of the Eulerian and Lagrangian methods for
predicting particle transport in enclosed spaces. Atmos Environ 2007;41(25):
The authors declare that they have no known competing 5236e48.
[26] Zhang Z, Chen Q. Experimental measurements and numerical simulations of
financial interests or personal relationships that could have
particle transport and distribution in ventilated rooms. Atmos Environ
appeared to influence the work reported in this paper. 2006;40(18):3396e408.
[27] Zhou G, Zhang Q, Bai RN, Fan T, Wang G. The diffusion behavior law of
Acknowledgements respirable dust at fully mechanized caving face in coal mine: CFD numerical
simulation and engineering application. Process Saf Environ Protect
2017;106:117e28.
This work is supported by the Central University Special Funding [28] Rashidi S, Esfahani J, Ellahi R. Convective heat transfer and particle motion in

1279
C.-y. Xu, B.-t. Zha, J.-q. Bao et al. Defence Technology 18 (2022) 1269e1280

an obstructed duct with two side by side obstacles by means of DPM model. [35] Gonza lez-Tello P, Camacho F, Vicaria JM, Gonzalez PA. Analysis of the mean
Appl Sci 2017;7(4):431. diameters and particle-Size distribution in powders. Part Part Syst Char
[29] Chao CYH, Wan MP. A study of the dispersion of expiratory aerosols in uni- 2010;27(5e6):158e64.
directional downward and ceiling-return type airflows using a multiphase [36] Gonza lez-Tello P, Camacho F, Vicaria JM, Gonz alez PA. A modified
approach. Indoor Air 2006;16(4):296e312. NukiyamaeTanasawa distribution function and a RosineRammler model for
[30] Cheng C, Zhang XB, Wang C, Lu W. Numerical investigation on cooling per- the particle-size-distribution analysis. Powder Technol 2008;186(3):278e81.
formance of filter in a pyrotechnic gas generator. Defence Technol 2021;17(2): [37] Farizhandi AAK, Zhao H, Lau R. Modeling the change in particle size distri-
343-351. bution in a gas-solid fluidized bed due to particle attrition using a hybrid
[31] Beghein C, Jiang Y, Chen QY. Using large eddy simulation to study particle artificial neural network-genetic algorithm approach. Chem Eng Sci 2016;155:
motions in a room. Indoor Air 2005;15(4):281e90. 210e20.
[32] Bivolarova M, Ondr a 
cek J, Melikov A, Zdímal V. A comparison between tracer [38] Xiao KT, Xu LC, Li HH. Implement of particle system in complex wind field and
gas and aerosol particles distribution indoors: the impact of ventilation rate, turbulence field. J Beijing Jiaot Univ 2015;39(2):13e21.
interaction of airflows, and presence of objects. Indoor Air 2017;27(6): [39] He F, He KK, Huang D, Wang H, Zhu YJ. Analysis of the effect of wind on smoke
1201e12. spread properties of a type explosive tear-gas grenade. J Ordnance Eng Coll
[33] Tang H. Study of inversion algorithm of particle size distribution using total 2016;28(5):25e9.
light scattering method. Harbin: Harbin Institute of Technology; 2008. [40] Xu LC, Hao XY, Xiao KT, Song WW, Chen CS. Simulation study of screening
[34] Bu XN, Chen YR, Ma GX, Sun YJ, Ni C, Xie GY. Wet and dry grinding of coal in a efficiency of explosive smoke bomb. Acta Armamentarii 2020;41(7):
laboratory-scale ball mill: particle-size distributions. Powder Technol 1299e306.
2020;359:305e13.

1280