RESEARCH ARTICLE | OCTOBER 20 2020

Numerical investigation of aerosol transport in a classroom

with relevance to COVID-19 
Special Collection: Flow and the Virus , Flow and the Virus

Mohamed Abuhegazy; Khaled Talaat  ; Osman Anderoglu; Svetlana V. Poroseva

Physics of Fluids 32, 103311 (2020)

[Link]


View Export
Online Citation

30 April 2024 [Link]

 Physics of Fluids ARTICLE [Link]/journal/phf

Numerical investigation of aerosol transport

in a classroom with relevance to COVID-19
Cite as: Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118
Submitted: 10 September 2020 • Accepted: 16 September 2020 •
Published Online: 20 October 2020

Mohamed Abuhegazy,1 Khaled Talaat,2,a) Osman Anderoglu,2 and Svetlana V. Poroseva1

AFFILIATIONS
1
Mechanical Engineering Department, University of New Mexico, Albuquerque, New Mexico 87106, USA
2
Nuclear Engineering Department, University of New Mexico, Albuquerque, New Mexico 87106, USA

Note: This paper is part of the Special Topic, Flow and the Virus.
a)
Author to whom correspondence should be addressed: ktalaat@[Link]

ABSTRACT
The present study investigates aerosol transport and surface deposition in a realistic classroom environment using computational fluid-
particle dynamics simulations. Effects of particle size, aerosol source location, glass barriers, and windows are explored. While aerosol
transport in air exhibits some stochasticity, it is found that a significant fraction (24%–50%) of particles smaller than 15 μm exit the sys-

30 April 2024 [Link]

tem within 15 min through the air conditioning system. Particles larger than 20 μm almost entirely deposit on the ground, desks, and nearby
surfaces in the room. Source location strongly influences the trajectory and deposition distribution of the exhaled aerosol particles and affects
the effectiveness of mitigation measures such as glass barriers. Glass barriers are found to reduce the aerosol transmission of 1 μm particles
from the source individual to others separated by at least 2.4 m by ∼92%. By opening windows, the particle exit fraction can be increased by
∼38% compared to the case with closed windows and reduces aerosol deposition on people in the room. On average, ∼69% of 1 μm particles
exit the system when the windows are open.
Published under license by AIP Publishing. [Link] s

I. INTRODUCTION role in aerosol transport but becomes less important with increased
particle size.7,8 The velocity field of the fluid (air) under known
Transmission of COVID-19 occurs primarily through SARS- boundary conditions can in principle be estimated by numerically
CoV2-laden droplets and aerosol particles inhaled directly or trans- solving Navier–Stokes equations through direct numerical simula-
mitted from contaminated surfaces.1 Effective mitigation measures tions (DNS), or more practically by numerically solving Reynolds-
necessitate clear understanding of droplet and aerosol transport, sur- Averaged Navier–Stokes (RANS) equations with approximate tur-
face retention, and evaporation kinetics in different environments bulence closures such as k-ϵ and k-ω closures.9,10
and conditions.2 Aerosols are generated during exhalation, talking, As particle properties significantly affect aerosol and droplet
coughing, sneezing, and other activities.3,4 In indoor environments, transport within a system, it is necessary to consider accurate par-
some of the generated particles exit the system through ventilation, ticle shape, size, and evaporation kinetics. The distinction between
some deposit on surfaces in the room and may settle or re-enter aerosols and droplets is rather arbitrary with no general agree-
the air, and others may be directly inhaled. Of primary interest to ment on a particle size threshold or suspension time threshold.3
mitigation measures is maximizing the fraction of particles that exit However, droplets are typically considered to be larger particles
the system and minimizing aerosol deposition on people to reduce where evaporation kinetics is rapid leading to the production of
disease transmission.5,6 smaller aerosols with slow evaporation kinetics.3 Aerosol particles
Aerosol transport within a control volume is primarily affected and droplets released from activities such as exhalation, talking, or
by inertial forces due to airflow and drag on the particle, and grav- coughing are polydisperse in nature. Exhalation and talking release
itational sedimentation.7 The forces acting on a particle primar- particles mostly <1 μm,11 and coughing releases larger particles typ-
ily depend on particle size and its position in the flow field. For ically <10 μm,12 while sneezing was found in one study to release
smaller particles (<0.5 μm), Brownian force can play a significant particles characterized by a bimodal size distribution with peaks

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-1

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

of ∼386 μm and 72 μm and the corresponding geometric standard hospitals.6 No chairs are considered in the model due to the exten-
deviation of 1.8 and 1.5, respectively.13 sive variability in chair sizes and shapes. Students are assumed to be
Computational fluid dynamics has been used in many stud- exposed to aerosols in order not to underestimate deposition on stu-
ies to investigate aerosol transport in outdoor conditions,14 indoor dents. An instructor is defined in the front, as shown in Fig. 1(a), and
conditions such as hospitals,6,15 and even inside the human airway is assumed to be 1.7 m in height. Independent surfaces are defined
system with good agreement with the experimental data.16,17 Dur- in the model for each object for tracking the aerosol deposition on
ing the COVID-19 pandemic, significant efforts have been made objects and students, respectively.
to develop computational fluid dynamics models of the human Air conditioning of the classroom follows ASHRAE 62.1 ven-
sneeze,18 investigate mask mechanics,19 and study aerosol transport tilation standards for acceptable indoor air quality.26 The air con-
and air flow in different environments and conditions such as air- ditioning system consists of five supply diffusers and four return
crafts,20 vehicular cabins,1 urinals and toilets,21,22 public spaces,23 air diffusers distributed as shown in Fig. 1(a). The Cubic Feet per
and indoor spaces.24,25 Despite these efforts, to the authors’ knowl- Minute (CFM) required for adequate ventilation was found to be
edge, no studies have investigated aerosol transport in a classroom ∼1230 CFM. The supply diffusers (1, 3, 5, 7, and 9) supply air at
environment although classroom sizes, the air conditioning lay- a 37○ angle from the horizontal surface with an inlet flow area of
out, and aerosol source distribution are characteristically different 0.294 m2 and a diffuser inlet vertical air velocity of 0.395 m/s based
than hospital care units and other indoor spaces discussed in the on ASHRAE recommendations.27 In the present work, the effect of
literature. opening windows while the air conditioning system is running on
While a typical 900 sq. ft classroom can fit 18 students and particle removal is explored. For this purpose, the model includes
an instructor, guidelines for re-opening schools have restricted the 3 windows (2.2 × 1.3 m2 ) that can be opened up to 50% in 10%
number of students to less than 10 students with 6 ft minimum increments.
spacing between the students. The effectiveness of these measures An unstructured, tetrahedral mesh is used, as shown in Fig. 1.
is dependent in part on aerosol transport within the classroom’s The mesh was generated using ANSYS ICEM 19.1. The mesh con-
air conditioned environment, which remains under-characterized. sists of 3.3 × 106 mesh cells with a minimum cell size of 0.5 cm
Other strategies for COVID-19 mitigation may include the use of and maximum cell size of 10 cm with gradual transition, maximum
glass screens as barriers to reduce aerosol transport between peo- skewness of 0.823 (a mean value of 0.593), and maximum aspect
ple in the room, opening windows, and redistributing students in ratio of 3.21 (a mean value of 1.43). The grid is refined near sur-
classrooms, but the ability of these measures to reduce aerosol faces to maintain a wall y + <10 during the simulations. Each case of
the 20 cases simulated in this work consumed ∼9 h running on four

30 April 2024 [Link]

transmission from one person to another needs to be carefully
evaluated. computer cores.
The objective of the present work is to investigate aerosol trans-
port and surface deposition in a model classroom environment using
B. Airflow and particle dynamics
computational fluid-particle dynamics (CFPD) simulations. Partic-
ularly, it is of interest to estimate the fraction of particles that exit the The present study uses the commercial CFD code, ANSYS
system, deposit on students, and deposit on surfaces such as desks, FLUENT 19.2, to simulate the airflow and particle transport. The
ground, walls, and ceiling. The effects of particle size, aerosol source continuity and momentum equations of the continuum phase (air)
location, glass barriers, and windows are investigated. Aerosol depo- are solved in the beginning independent of the discrete phase using
sition on different students from different sources is compared to the steady state Reynolds Averaged Navier–Stokes (RANS) incom-
qualitatively explore the risk posed to individuals in the room due to pressible solver. The present simulations use the Re-Normalization
their position with respect to an infected student. Group (RNG) k-ε model.28 The choice of the RNG k-ε model is
motivated by the work of Ramponi and Blocken who investigated
the influence of turbulence models on cross-ventilation for a generic
isolated building, and it was found that the RNG k-ε model was suit-
II. METHODS able for their application and operation conditions, which, in part,
resemble the current application.29
A. Classroom model and spatial mesh The SIMPLE algorithm implemented in ANSYS FLUENT is
A three-dimensional model of a classroom consisting of nine applied for pressure velocity coupling with pressure interpolation
students and an instructor was developed. The model uses realistic of first order. The convection and viscous terms of the governing
classroom dimensions and air conditioning. The classroom shown equations were discretized utilizing the second-order discretization
in Fig. 1 is 9 × 9 m2 in area and 3 m in height. The distance scheme. The solution is assumed to be converged when all the scaled
between each student is 2.4 m (7′ 10′′ ), which is greater than the residuals stabilize and approach a minimum of 10−5 for k, ε, x, y,
recommended 6 ft separation distance for COVID-19 mitigation. and z momentum equations as well as 10−4 for the continuity equa-
The model includes desks (with glass screens and without them) tion. Once the continuum phase solution converges, the flow field
and windows. All students are represented similarly and have the is then frozen and is used to transport the discrete phase (aerosol
same dimensions. Each student consists of a cuboid body (0.5 × 0.25 particles). The effect of the particles on the flow of air is negligible.
× 1 m3 ) and a cuboid head (0.16 × 0.15 × 0.2 m3 ) with a rectangu- One-way coupling between the continuum phase and the discrete
lar mouth surface (0.06 × 0.03 m2 ) through which particles and air phase is used given the low concentration of the aerosol particles
are injected into the system. The simplified human model is inspired in air. The particle trajectory is determined by solving the equa-
by models used in a numerical investigation of cross-transmission in tion of motion for the particle in a Lagrangian framework. The

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-2

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

30 April 2024 [Link]

FIG. 1. Illustration of (a) the classroom model and (b) the computational mesh used in the CFD simulations.

dvi f
equation of motion for the particles is given in the following equa- = (ui − vi ) + gi (1 − α) + fi,Brownian + fi,lift , (2)
tion [Eq. (1)]: dt τp Cc
dÐ→
vi ÐÐ→ Ð → Ð →
m = Fdrag + Fg + Fa , (1) where ui is the velocity of the flow, f is the drag factor,31 τ p is the par-
dt
ticle reaction time, and Cc is the Cunningham correction factor.32
where vi is the velocity of the particle, m is the mass of the parti-
ÐÐ→ Ð→ The present simulations use 96 000 particles, which is a reasonable
cle, Fdrag is the drag force between the air and the particle, Fg is the
Ð
→
number of particles for sound statistics and is greater than those used
gravity force, and Fa represents the other additional forces includ- in another study of aerosol removal in hospital care units.15 The tur-
ing the pressure force, virtual mass force, Basset force, Brownian bulent dispersion of particles and the random effects of turbulence
force, and Saffman’s lift force. The particles used in the present work on particle dispersion were taken into account using the discrete
are sufficiently small to neglect pressure and virtual mass forces and random walk method implemented in ANSYS FLUENT. Since the
sufficiently large to neglect Brownian force.7,8,30 As the particles are particles are small enough to stick to surfaces, the trap boundary
much smaller than the mesh elements, it is necessary to use drag condition is used for the particles over all solid surfaces. In real-
models. The present work uses the Stokes–Cunningham drag model. ity, some of the particles will be reflected and others may re-enter
Therefore, the equation of motion of the particles could be written the air after deposition. However, re-entry and reflection are diffi-
more explicitly as follows [Eq. (2)]: cult to account for as they are affected by particle properties, surface

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-3

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

TABLE I. List of parameter combinations investigated in the present work.

Investigation Source Particle size (μm) Screens Windows

Student 5 1 No Closed
Student 5 4 No Closed
Student 5 10 No Closed
Effect of particle size
Student 5 15 No Closed
Student 5 20 No Closed
Student 5 50 No Closed

Student 1 1 No Closed
Student 2 1 No Closed
Effect of source location Student 5 1 No Closed
Student 8 1 No Closed
Student 9 1 No Closed

Student 1 1 Yes Closed

Student 2 1 Yes Closed
Effect of glass barriers/screens Student 5 1 Yes Closed
Student 8 1 Yes Closed
Student 9 1 Yes Closed

Student 5 1 No 10% open

Student 5 1 No 20% open
Effect of windows Student 5 1 No 30% open
Student 5 1 No 40% open
Student 5 1 N0 50% open

30 April 2024 [Link]

properties, and flow conditions.33 An escape boundary condition is efficiency of the air conditioning system, the present work is con-
employed for the diffusers and mouths. Air flow from mouths is cerned with the effect on particle removal. Table I summarizes the
assumed to be exhaled at 20 l/min specified as a velocity inlet bound- parameter combinations investigated in the present work.
ary condition (0.185 m/s) for a mouth inlet area of 0.0018 m2 . The
particles are released with the same velocity normal to the mouth
surface. III. RESULTS AND DISCUSSION
A. Airflow and particle dynamics
C. Study design The velocity field of the continuum phase and the distribu-
The base case uses 1 μm particles, student 5 as the source, no tion of turbulent kinetic energy and vorticity are of fundamental
glass barriers, and windows closed. The choice of 1 μm particles is in importance to aerosol transport. Figure 2 shows the turbulent kinetic
the range of the particle size of aerosol particles released in exha- energy distribution, velocity magnitude distribution, and velocity
lation and talking.11 Student 5 is used as the source for the base vectors of air across a two-dimensional slice going through students
case due to their location far away from vortices at the edges of 2, 5, and 8. In this slice, air is injected into the system through the
the room. supply diffuser in the middle (inlet 5) at a 37○ angle with the ceil-
The present study investigates the effects of particle size, source ing. Return diffusers 2 and 8 are shown at the sides. The turbulent
position, glass barriers, and windows on the fate of the exhaled kinetic energy is more significant at the edges of the room (especially
aerosol particles. Parameters of the base case are varied to inves- at the outlets) and close to student 8 by the virtue of their location
tigate these effects. Particle sizes studied are 1 μm, 4 μm, 10 μm, with respect to air conditioning [Fig. 2(a)]. The velocity magnitude
15 μm, 20 μm, and 50 μm. Aerosol sources considered are students is strongest at the inlets and outlets, but the air is not stagnant in
1, 2, 5, 8, and 9 and are studied with and without 70 cm high glass the rest of the room due to air conditioning [Fig. 2(b)]. The veloc-
barriers/screens placed on top of the desks. The effect of windows is ity vectors [Fig. 2(c)] demonstrate the recirculation near the edges
explored by comparing aerosol deposition and transfer in the class- of the room and near student 8’s head. Vortices can partially trap
room with 0%, 10%, 20%, 30%, 40%, and 50% open windows. As aerosol particles that are transported to those regions and increase
three windows are available in the room, 10% open windows implies deposition on neighboring surfaces.
that 10% of each of the three windows is open. The effect of win- Particle transport in the classroom environment due to an
dows is explored while the air conditioning system is running. While impulse aerosol source is a transient process. For the purposes of
this can increase the cooling/heating load and decrease the energy characterizing the dynamics and the fate of exhaled aerosol particles,

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-4

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

FIG. 2. (a) Turbulent kinetic energy, (b)

velocity magnitude distribution, and (c)
velocity vectors across a slice going
through students 2, 5, and 8.

30 April 2024 [Link]

a single-release impulse source is used. Figure 3 shows the distribu- student 8 due to air conditioning is strong compared to that near stu-
tion of 1 μm aerosol particles in the classroom at different points dent 5. Student 8 is also present near a region with recirculation and
in time since particle release. Figure 3(a) illustrates the transport of strong vortices compared to the rest of the room. The particles dis-
particles released from student 5. After 1 s of release, the aerosol perse slowly, and even after 5 min, most of the particles are present
particles exhibit a parabolic distribution at the front of the particle in the back half of the room.
swarm. The particles slowly disperse and rise up during the first 50 s.
Once the particles reach the downstream of the air conditioner, the
particles are rapidly transported to different parts of the room. As B. Effect of particle size
air flows from the supply diffusers to the return diffusers, the parti- Particle size is of fundamental importance to aerosol transport.7
cles that reach the downstream of the air tend to follow the flow and The present work considers spherical aerosol particles in the 1 μm–
exit the system. Overall, there are significantly more 1 μm particles 50 μm size range. Figure 4 shows the effect of particle size on the
in the upper half of the room than the bottom half due to the flow of fraction of aerosol particles released from student 5’s mouth that
air to the return diffusers that are located in the ceiling in the present deposit on different surfaces in the room, such as ground, ceiling
model. Figure 3(a) highlights the significance of the flow velocity dis- and walls, desks, and students, or escape from the outlet of the air
tribution on aerosol transport in the room. Therefore, the results of conditioning system. No significant difference is observed between
the present work are applicable to classrooms with comparable air 1 μm and 4 μm particles [Figs. 4(a) and 4(b)]. Nearly 50% of 1 μm
conditioning. and 4 μm aerosol particles exit the room through the air condition-
Figure 3(b) illustrates the transport of 1 μm particles released ing system after 15 min. Roughly 15% of the particles deposit on
from student 8. Aerosols released from student 8 [Fig. 3(b)] exhibit the ceiling, and ∼10% deposit on the walls of the classroom, which
a substantially different distribution than aerosols released from stu- is comparable to 14%–15% deposition on the ground [Figs. 4(a)
dent 5 [Fig. 3(a)]. At 1 s, the particle swarm curves downward and and 4(b)]. This suggests that gravity does not play a significant role
much of the particles deposit on the source student (student 8). This in the transport of 1 μm and 4 μm particles in the timescale of
is a result of the position of student 8 with respect to the air con- 15 min. In the case of 10 μm particles [Fig. 4(c)], deposition on the
ditioning system. As shown in Fig. 2, the velocity magnitude near ground increases to 27% compared to 14%–15% in 1 μm and 4 μm

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-5

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

30 April 2024 [Link]

FIG. 3. Distribution of 1 μm aerosol particles in the classroom at different points in time for the (a) student 5 source and (b) student 8 source.

particles [Figs. 4(a) and 4(b)] and the fraction of particles that exit and 0% at 50 μm [Fig. 4(f)]. On the other hand, the total fraction of
the system through air conditioning is reduced to 41%. The fraction particles that deposit on the ground, desks, and the source student
of aerosol particles that exit the system drops rapidly with the parti- increases significantly with increased particle size [Figs. 4(a)–4(f)].
cle size greater than 10 μm from 41% at 10 μm [Fig. 4(c)] to 24% at 15 For instance, ∼21% of 1 μm particles deposit on the ground, desks,
μm [Figs. 4(d)], 16% at 17 μm (not shown), 5% at 20 μm [Figs. 4(e)], and students, while ∼92% of 20 μm particles deposit on the ground,

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-6

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

30 April 2024 [Link]

FIG. 4. Effect of particle size on aerosol deposition and removal from the classroom model as a function of time since particle release from student 5’s mouth. This figure
shows the deposition fraction for (a) 1 μm particles, (b) 4 μm particles, (c) 10 μm particles, (d) 15 μm particles, (e) 20 μm particles, and (f) 50 μm particles.

desks, and students. The rest of the particles deposit on the ceiling compares the aerosol deposition on various surfaces originating
and walls, exit the room through the air conditioning system, or from different sources (students 1, 2, 8, and 9) using 1 μm particles.
remain in the air for longer than 15 min. The aerosol deposition for the student 5 source was shown earlier in
Figure 4 also shows that ∼15 min is adequate for 1 μm–50 μm Fig. 4(a).
particles to have at least one interaction with a surface or exit the The results in Fig. 5 show that the effect of source location on
room. In the case of 50 μm particles, the particles deposit rapidly in aerosol transport can be substantial as in the case for student 8. The
less than a minute and mostly on the source student. The extensive deposition distribution in the case of student 1 [Fig. 5(a)] is similar
deposition of 50 μm particles on the source student is due to gravi- to that of student 5 [Fig. 4(a)] except for very low aerosol depo-
tational settling and the simplified, rectangular geometry of the stu- sition on the ground compared to student 5 (3.9% vs 13.7%) and
dent modeled (Figs. 1 and 2). Much of these 50 μm particles would the increased aerosol deposition on the walls and ceiling (∼38% vs
deposit on the ground if not for the simplified student geometry. 26%). The deposition results for student 9, who is positioned in the
back corner, also suggest increased deposition on the wall and ceil-
ing to ∼44% of exhaled aerosol particles. In the case of student 1
C. Effect of source location and student 9, the increased deposition on walls and ceiling can be
The position of the initial aerosol source in the fluid flow field explained in part by proximity to walls and in part due to the vor-
affects the trajectory of the released particles [Eqs. (1) and (2)]. tex structures present near the edges of the room (Fig. 2). Student
The location of the student with respect to air conditioning influ- 2 who is positioned in the front-middle, far from walls, experiences
ences the local flow field and particle dynamics (Figs. 2 and 3). increased deposition on the walls and ceiling compared to student
It is, therefore, of interest to understand the extent of the effect 5 [Figs. 5(b) and 4(a)]. This increase in deposition on the wall may
of source location on the fate of the exhaled particles. Figure 5 be explained by the vortices present in the flow in front of student

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-7

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

30 April 2024 [Link]

FIG. 5. Effect of student location on aerosol deposition and removal from the classroom model using 1 μm particles. This figure shows four different student sources: (a)
student 1, (b) student 2, (c) student 8, and (d) student 9.

2 (Fig. 2). The deposition on the ground appears somewhat stochas- the particles on the screens varies significantly from one source to
tic due to the vortices, but in general, it is <20% for 1 μm particles. another. The fraction of 1 μm particles deposited on the screens is
The fraction of particles that exit through the air conditioning sys- very small (<1.5%) for students 5, 8, and 9 [Figs. 6(c)–6(e)]. More
tem is consistently >30% except for student 8 [Fig. 5(c)]. The case of significant deposition on the screens is observed in the case of stu-
student 8 is special due to their unique position with respect to the dent 1 (9%) and student 2 (∼51%), as shown in Figs. 6(a) and 6(b).
air conditioning system (Fig. 2), which directs the particles down- Differences in the aerosol deposition compared to the case with no
ward and onto themselves (Fig. 3). Less than 10% of the particles barriers (Fig. 5) are also observed. The differences can be attributed
exhaled by student 8 exit the room through the air conditioning to the modulation of the local flow field as a result of the barriers,
system. which further depends on the position of the barrier in the flow field.
Notably, the inclusion of barriers decreases the total fraction of par-
ticles deposited on the students by ∼63% on average compared to the
D. Effect of glass barriers/screens case with no barriers. However, barriers appear to slow down aerosol
One of the commonly used measures to reduce COVID-19 removal and deposition. For instance, ∼20% of the particles remain
transmission is the use of sneeze guards in the form of glass or plas- in the air after 15 min in the case of student 9 when barriers are used,
tic barriers. The efficiency of barriers is not independent of the flow while only ∼3% of particles remain in the case with no barriers.
field where they are employed, which depends on air conditioning It is difficult to assess the effectiveness of glass barriers in reduc-
and the geometry of the surroundings. Therefore, it is necessary to ing aerosol transmission based on Figs. 5 and 6, which do not dis-
evaluate its effectiveness in the classroom environment especially criminate between the source student and receivers. For a clearer
for small particles such as 1 μm particles which can diffuse for long comparison, Fig. 7 shows source–receiver maps for 1 μm particles in
distances in the room. the absence and presence of screens. The sources considered are stu-
Figure 6 shows the deposition distribution of 1 μm particles dent 1, student 2, student 5, student 8, and student 9. Self-deposition
released from different student sources in the presence of 70 cm is indicated in a box next to each student, and the fraction of aerosol
tall glass barriers on top of the desks [Fig. 6(f)]. The deposition of deposited on other students is marked by arrows from the source

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-8

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

30 April 2024 [Link]

FIG. 6. Effect of glass barriers on aerosol deposition and removal from the classroom model using 1 μm particles for different student sources. This figure shows five different
student sources: (a) student 1, (b) student 2, (c) student 5, (d) student 8, and (e) student 9. The glass barriers are shown in (f).

FIG. 7. Effect of glass barriers on aerosol transmission between students. Sources considered are student 1, student 2, student 5, student 8, and student 9, and particle size
is 1 μm. Source–receiver maps are shown for cases with (a) no glass screens or sneeze guards and (b) glass screens employed.

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-9

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

to the receiver. A threshold of 0.01% (∼10 particles) is applied to E. Effect of windows

the maps. The use of a threshold is to ensure that only statistically
meaningful numbers are reported. The effect of opening windows while the air conditioning sys-
On average, the total fraction of aerosols transmitted from a tem is running is investigated in order to understand its impact on
source student to others in the classroom decreases by ∼92% in the particle removal compared to the case with windows closed. A typ-
case with screens. In the presence of screens, very few aerosol par- ical sliding window can be opened up to 50% of its total width.
ticles (<0.01%) are transmitted from student 1 to the others in the The present work considered cases with 0%, 10%, 20%, 30%, 40%,
room, and self-deposition is significantly reduced from 1.5% to 0.3%. and 50% open windows using 1 μm particles. The source student
In the cases of student 2 and student 5, aerosol transmission to oth- is assumed to be student 5. Figure 8 shows the effect of opening
ers and self is consistently reduced with the exception of increased windows on aerosol deposition and removal.
transmission from student 5 to student 9 (from 0.04% to 0.08%). In The total fraction of particles that exit the system through the
the case of student 8, self-deposition increases from ∼47% to ∼60% windows and air conditioning outlet is increased on average by
and deposition on students 6 and 9 increases, but deposition on ∼38% [Figs. 8(a)–8(f)]. The fraction of particles that exit the system
others decreases significantly. In the case of student 9, total aerosol through air conditioning is reduced by ∼60%. This is advantageous
transmitted to others is reduced by ∼74%. However, transmission as fewer particles may be able to transfer to other rooms bypassing
from student 9 to student 5 increases from 0.02% to 0.3% and that the air conditioning filters. The fraction of particles that exit through
from student 9 to student 7 increases from 0.05% to 0.18%. Over- the windows appears to be affected by the extent to which the win-
all, the addition of screens substantially reduces aerosol transmis- dows are open. The results shown in Figs. 8(a)–8(f) suggest that there
sion from one student to another, but it does not eliminate particle may be an optimal configuration such that the fraction of particles
transmission between students. that exit the system is maximized although no systematic trend is

30 April 2024 [Link]

FIG. 8. Effect of opening windows on aerosol deposition and removal using 1 μm particles and the student 5 source. This figure shows the deposition fractions for cases with
(a) 0%, (b) 10%, (c) 20%, (d) 30%, (e) 40%, and (f) 50% open windows.

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-10

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

observed. The fraction of particles that exit the system for 0%, 10%, between students and increases the fraction of particles that exit the
20%, 30%, 40%, and 50% open windows is 50%, 87%, 51%, 92%, 36%, system.
and 78%, respectively. On average, ∼69% of particles exit the sys- The present work is subject to many limitations. First, deposi-
tem when windows are open at all, compared to 50% with windows tion of aerosol particles on contact with solid surfaces is assumed.
closed. With the exception of 40% open windows, opening windows Reflection and re-entry are not considered. This is, however, jus-
increases the fraction of particles that exit the system. tified as most of the simulations conducted in this study are of
1 μm particles. Particles <50 μm in diameter can stick to surfaces
through van der Waals forces.34 Adhesion forces acting on 1 μm
F. Discussion
particles can exceed gravitational force acting on the particle by fac-
The results demonstrate that a large fraction (24%–50%) of tors greater than 1 × 106 .34 Adhesion forces, however, depend on
smaller particles (<15 μm) exit the room without interacting with particle properties, surface properties, and environmental factors.33
any surfaces in the room. This finding highlights the need for effi- Second, the present work does not investigate the synergy between
cient filtering in the air conditioning systems. The aerosol released the different factors considered. For instance, the effect of open-
from students disperses in the room, and its concentration decreases. ing windows on aerosol removal and deposition is not necessarily
The concentration of the aerosol particles increases again as they independent of particle size. Nevertheless, investigating the synergy
enter the air conditioning system. The transfer of a larger fraction between the different variables would necessitate extensive compu-
of exhaled particles to the air conditioning return diffuser, although tational resources not available to the current project. The current
beneficial to individuals in the room, may pose greater risk to indi- study is rather focused on identifying what factors are important for
viduals in other rooms as air conditioning systems often use recycled aerosol transport in a classroom in order to inform other studies
air. It is also found that a 2.4 m separation distance between students that may further investigate the interactions between the different
is inadequate to eliminate particle transmission between students factors. Third, the deposition fraction is assumed to be a single deter-
with the exception of 50 μm particles. ministic value. Statistical characterization of the deposition fraction
The fraction of particles that exit the system without interacting would be of interest especially because of the existence of recircu-
with any surfaces depends on the source location. Interestingly, stu- lation and vortices near the edges of the classroom. Fourth, class-
dents closer to the supply diffusers such as student 1, student 5, and rooms are subject to extensive variability in sizes, air conditioning,
student 9 are associated with greater particle exit fractions than stu- student distribution, and student age/, which would affect aerosol
dents closer to outlets such as student 2 and student 8. The position deposition and removal. Effective mitigation strategies should con-

30 April 2024 [Link]

of the student in the flow field significantly affects particle transport. sider multi-layer approaches including using masks, redistributing
Significant aerosol deposition (∼47%) on student 8 is observed due students, using glass barriers, opening windows, optimizing the air
to the aerosol they released. This is due to their unique position in conditioning system for maximum particle removal, and improving
the flow field near a vortex region close to the edge of the room and air conditioning filters.
close to an outlet. An important implication of this increased aerosol
deposition on student 8 is that it suggests the presence of mixing
hotspots in the room where aerosol deposition can increase by as IV. CONCLUSIONS AND RECOMMENDATIONS
much as tenfold. In such a hotspot, if two students are present, the Understanding aerosol transport in different environments
chances of aerosol transmission between the two will be significantly is of critical importance to COVID-19 mitigation measures. The
higher than elsewhere in the room. This highlights the need for thor- present study investigated aerosol removal and surface deposition
ough characterization of aerosol transport in different environments in a realistic classroom environment using computational fluid-
to identify and avoid hotspot areas. particle dynamics (CFPD) simulations. A model classroom that
Sneeze guards/glass barriers were found to effectively reduce included nine students and a teacher was constructed. Air condi-
the transmission of 1 μm aerosol between students by ∼92% on aver- tioning of the classroom followed ASHRAE 62.1 ventilation stan-
age. While the fraction of particles deposited on the screens directly dards for acceptable indoor air quality. Four different factors were
is small in most cases studied, the screens appear to modulate the considered: particle size (1 μm–50 μm), source location (students
local flow field resulting in less aerosol transmission between stu- 1, 2, 5, 8, and 9), presence of barriers/sneeze guards, and opening
dents. Screens, however, do not completely eliminate transmission windows (10%–50% of window width). The following points high-
of 1 μm particles between students and their effectiveness depends light the main findings of this work and the implications of these
on source location within the classroom with respect to the air findings:
conditioning system. Nevertheless, the 92% reduction in aerosol
transmission is highly beneficial. (a) Aerosol distribution in the room is not uniform and is
Opening windows was found to increase the fraction of parti- strongly influenced by the air conditioning layout.
cles that exit the system by ∼38% compared to the case with closed (b) Even with only 9 students in the room and 2.4 m distance
windows. The fraction of aerosol particles that deposit on students between students, the aerosol (1 μm–20 μm) is transmitted in
(including the source) decreased from 2.3% to an average of 0.45% significant quantities between students and from one student
when windows are open at all suggesting that opening windows to other students’ desks with aerosol transmission between
reduces aerosol deposition on students by ∼80%. The present study two neighboring students reaching 0.9% of exhaled parti-
only investigated one source (student 5) for cases with open win- cles in some 1 μm particle cases. Studies have estimated that
dows. However, the results suggest that opening windows while ∼20 000 particles in the 0.8 μm–5.5 μm range are released
the air conditioning system is running reduces aerosol transmission and that over 100 000 virions are emitted per minute of

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-11

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

speaking.35,36 Therefore, particles transmitted between neigh- DATA AVAILABILITY

boring students separated by a 2.4 m distance in a classroom
The data that support the findings of this study are available
may exceed 180 particles per minute. The transmission of par-
from the corresponding author upon reasonable request.
ticles from one student to other students’ desks highlights the
need for hand sanitization even without contact with other
students’ belongings. REFERENCES
(c) The effect of source location on aerosol transport is signif- 1
M. Jayaweera, H. Perera, B. Gunawardana, and J. Manatunge, “Transmission of
icant. Student 1 in the front corner transmitted ∼0.55% of COVID-19 virus by droplets and aerosols: A critical review on the unresolved
exhaled 1 μm aerosol particles to other students, while student dichotomy,” Environ. Res. 188, 109819 (2020).
5 in the middle transmitted ∼2.1% of exhaled particles to oth-
2
R. Mittal, R. Ni, and J.-H. Seo, “The flow physics of COVID-19,” J. Fluid Mech.
ers. Removing the middle student seat (student 5) may help 894, F2 (2020).
3
M. A. Kohanski, L. J. Lo, and M. S. Waring, “Review of indoor aerosol generation,
reduce the risk of infection to others. Furthermore, student
transport, and control in the context of COVID-19,” Int. Forum Allergy Rhinol.
position appears to affect the likelihood of receiving aerosol (2020).
particles from others. Students 7 and 9 in the back corners 4
P. Anfinrud, V. Stadnytskyi, C. E. Bax, and A. Bax, “Visualizing speech-generated
received 2 to 3 times less particles on average than most other oral fluid droplets with laser light scattering,” N. Engl. J. Med. 382, 2061 (2020).
5
students in the room. Therefore, students at risk of COVID- L. Morawska, J. W. Tang, W. Bahnfleth, P. M. Bluyssen, A. Boerstra, G.
19 complications may be placed in positions with a lower Buonanno, J. Cao, S. Dancer, A. Floto, F. Franchimon, C. Haworth, J. Hogeling,
chance of receiving particles. C. Isaxon, J. L. Jimenez, J. Kurnitski, Y. Li, M. Loomans, G. Marks, L. C. Marr,
L. Mazzarella, A. K. Melikov, S. Miller, D. K. Milton, W. Nazaroff, P. V. Nielsen,
(d) Opening windows while the air conditioning system is run- C. Noakes, J. Peccia, X. Querol, C. Sekhar, O. Seppänen, S.-i. Tanabe, R.
ning, while not recommended from an HVAC point of view, Tellier, K. W. Tham, P. Wargocki, A. Wierzbicka, and M. Yao, “How can airborne
significantly increases particle exit fraction by ∼38% and transmission of COVID-19 indoors be minimised?,” Environ. Int. 142, 105832
reduces transmission between students by ∼80%. (2020).
6
(e) Glass screens reduce aerosol transmission from one student J.-X. Wang, X. Cao, and Y.-P. Chen, “An air distribution optimization of hospital
to another and should be used. The extent of their effective- wards for minimizing cross-infection,” J. Clean. Prod. 279, 123431 (2021).
7
ness depends on the source location with respect to the air A. Tsuda, F. S. Henry, and J. P. Butler, Comprehensive Physiology (John Wiley &
Sons, Inc., Hoboken, NJ, USA, 2013).
conditioning system. 8
P. W. Longest and J. Xi, “Effectiveness of direct Lagrangian tracking models for
(f) Particles disperse in the room and re-concentrate at the return simulating nanoparticle deposition in the upper airways,” Aerosol Sci. Technol.

30 April 2024 [Link]

ducts of the air conditioning system. A large fraction of 41, 380 (2007).
exhaled particles end up in the air conditioning system, which 9
B. E. Launder and D. B. Spalding, “The numerical computation of turbulent
highlights the need for effective filtration and sterilization flows,” Comput. Methods Appl. Mech. Eng. 3, 269 (1974).
10
systems within air conditioners. F. R. Menter, “Two-equation eddy-viscosity turbulence models for engineering
applications,” AIAA J. 32, 1598 (1994).
11
F. Wurie, O. Le Polain de Waroux, M. Brande, W. DeHaan, K. Holdgate,
Finally, the results of this work should be interpreted under R. Mannan, D. Milton, D. Swerdlow, and A. Hayward, “Characteristics of exhaled
the context of the air conditioning layout and student distribution particle production in healthy volunteers: Possible implications for infectious
used. Other classrooms may employ different air conditioning stan- disease transmission,” F1000Research 2, 14 (2013).
12
dards and might necessitate aerosol transport investigations tailored W. G. Lindsley, T. A. Pearce, J. B. Hudnall, K. A. Davis, S. M. Davis, M. A. Fisher,
to the specific classroom. Each case of the 20 cases simulated in this R. Khakoo, J. E. Palmer, K. E. Clark, I. Celik, C. C. Coffey, F. M. Blachere, and
work consumed ∼9 h running on four computer cores. Notably, this D. H. Beezhold, “Quantity and size distribution of cough-generated aerosol par-
ticles produced by influenza patients during and after illness,” J. Occup. Environ.
runtime was enabled by freezing the continuum solver upon conver- Hyg. 9, 443 (2012).
gence of the residuals. Only the discrete phase transport is solved for 13
Z. Y. Han, W. G. Weng, and Q. Y. Huang, “Characterizations of particle size
a simulation time of 15 min. distribution of the droplets exhaled by sneeze,” J. R. Soc. Interface 10, 20130560
(2013).
14
G. Habilomatis and A. Chaloulakou, “A CFD modeling study in an urban
SUPPLEMENTARY MATERIAL street canyon for ultrafine particles and population exposure: The intake fraction
approach,” Sci. Total Environ. 530-531, 227 (2015).
See the supplementary material for a high resolution (7832 15
× 3168 pixel2 ) figure showing the velocity vectors of Fig. 2(c) for
H. Qian and Y. Li, “Removal of exhaled particles by ventilation and deposition
in a multibed airborne infection isolation room,” Indoor Air 20, 284 (2010).
greater clarity. 16
K. Talaat, J. Xi, P. Baldez, and A. Hecht, “Radiation dosimetry of inhaled
radioactive aerosols: CFPD and MCNP transport simulations of radionuclides in
the lung,” Sci. Rep. 9, 17450 (2019).
AUTHORS’ CONTRIBUTIONS 17
J. Xi, T. Yang, K. Talaat, T. Wen, Y. Zhang, S. Klozik, and S. Peters, “Visualiza-
tion of local deposition of nebulized aerosols in a human upper respiratory tract
M.A. and K.T. contributed equally to this work. model,” J. Vis. 21, 225 (2018).
18
G. Busco, S. R. Yang, J. Seo, and Y. A. Hassan, “Sneezing and asymptomatic
virus transmission,” Phys. Fluids 32, 073309 (2020).
19
ACKNOWLEDGMENTS T. Dbouk and D. Drikakis, “On respiratory droplets and face masks,” Phys.
Fluids 32, 063303 (2020).
Mr. Ibrahim El-Hagali at Pennsylvania State University is 20
P. S. Desai, N. Sawant, and A. Keene, medRxiv:2020.08.17.20176909 (2020).
gratefully acknowledged for providing computational resources 21
J.-X. Wang, Y.-Y. Li, X.-D. Liu, and X. Cao, “Virus transmission from urinals,”
used in this work. Phys. Fluids 32, 081703 (2020).

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-12

Published under license by AIP Publishing
 Physics of Fluids ARTICLE [Link]/journal/phf

22 29
Y.-y. Li, J.-X. Wang, and X. Chen, “Can a toilet promote virus transmission? R. Ramponi and B. Blocken, “CFD simulation of cross-ventilation for a generic
From a fluid dynamics perspective,” Phys. Fluids 32, 065107 (2020). isolated building: Impact of computational parameters,” Build. Environ. 53, 34
23 (2012).
A. Adwibowo, MedRxiv 2020.07.02.20145219 (2020).
24 30
V. Vuorinen, M. Aarnio, M. Alava, V. Alopaeus, N. Atanasova, M. Auvinen, B. Zhao, Y. Zhang, X. Li, X. Yang, and D. Huang, “Comparison of indoor aerosol
N. Balasubramanian, H. Bordbar, P. Erästö, R. Grande, N. Hayward, A. Hellsten, particle concentration and deposition in different ventilated rooms by numerical
S. Hostikka, J. Hokkanen, O. Kaario, A. Karvinen, I. Kivistö, M. Korhonen, method,” Build. Environ. 39, 1 (2004).
31
R. Kosonen, J. Kuusela, S. Lestinen, E. Laurila, H. J. Nieminen, P. Peltonen, S. A. Morsi and A. J. Alexander, “An investigation of particle trajectories in two-
J. Pokki, A. Puisto, P. Råback, H. Salmenjoki, T. Sironen, and M. Österberg, “Mod- phase flow systems,” J. Fluid Mech. 55, 193 (1972).
32
elling aerosol transport and virus exposure with numerical simulations in relation M. D. Allen and O. G. Raabe, “Slip correction measurements of spherical solid
to SARS-CoV-2 transmission by inhalation indoors,” Saf. Sci. 130, 104866 (2020). aerosol particles in an improved millikan apparatus,” Aerosol Sci. Technol. 4, 269
25
B. Zhang, G. Guo, C. Zhu, Z. Ji, and C.-H. Lin, “Transport and trajectory (1985).
33
of cough-induced bimodal aerosol in an air-conditioned space,” Indoor Built S. G. Yiantsios and A. J. Karabelas, “Deposition of micron-sized particles on
Environ. (2020). flat surfaces: Effects of hydrodynamic and physicochemical conditions on particle
26 attachment efficiency,” Chem. Eng. Sci. 58, 3105 (2003).
American Society of Heating, Refrigerating, and Air-Conditioning Engineers,
34
in ASHRAE Standard 62.1-2007: Ventilation and acceptable indoor air quality R. A. Bowling, Surfaces, Part 1 (Springer US, Boston, MA, 1988), pp. 129–142.
(ASHRAE, Inc., Atlanta, GA, 2007). 35
V. Stadnytskyi, C. E. Bax, A. Bax, and P. Anfinrud, “The airborne lifetime of
27
American Society of Heating, Refrigerating, and Air-Conditioning Engineers, small speech droplets and their potential importance in SARS-CoV-2 transmis-
in 2012 ASHRAE Handbook - HVAC Systems and Equipment (ASHRAE, Inc., sion,” Proc. Natl. Acad. Sci. U. S. A. 117, 011875 (2020).
Atlanta, GA, 2012). 36
L. Morawska, G. R. Johnson, Z. D. Ristovski, M. Hargreaves, K. Mengersen,
28
V. Yakhot, S. A. Orszag, S. Thangam, T. B. Gatski, and C. G. Speziale, “Devel- S. Corbett, C. Y. H. Chao, Y. Li, and D. Katoshevski, “Size distribution and sites
opment of turbulence models for shear flows by a double expansion technique,” of origin of droplets expelled from the human respiratory tract during expiratory
Phys. Fluids A 4, 1510 (1992). activities,” J. Aerosol Sci. 40, 256 (2009).

30 April 2024 [Link]

Phys. Fluids 32, 103311 (2020); doi: 10.1063/5.0029118 32, 103311-13

Published under license by AIP Publishing