PreprintPDF Available

Towards Safer Work Environments During the COVID-19 Crisis: A Study Of Different Floor Plan Layouts and Ventilation Strategies Coupling Open FOAM and Airborne Pathogen Data for Actionable, Simulation-based Feedback in Design

Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

As work environments struggle to reopen during the current COVID-19 pandemic, it is crucial to establish practical decision-aiding tools. While a strong emphasis has been placed on determining generic guidelines to reduce the risk of airborne viral spread, there is a lack of free and easy-to-use simulation workflows to quantify indoor air quality and the risk of airborne pathogens indoors at a spatial resolution that can take into account floor-plan layouts, furniture, and ventilation inlet-outlet positions. This paper describes the development of a new, free, early design tool that allows designers and other stakeholders to simulate and compare viral airborne concentration under different indoor conditions. The tool leverages OpenFOAM-based Computational Fluid Dynamics (CFD) and a passive scalar simulation approach to allow architects and interior designers to quantify airborne pathogens' exposure. The tool is integrated into the popular Rhino3d & Grasshopper CAD environment to facilitate its application in fast-paced design processes. We demonstrate good agreement compared to a CFD benchmark test. Further, we validate newly developed COVID-19 capabilities by comparing our results to an existing restaurant case study that included tracer gas measurements and validation using Fluent (Ansys). We demonstrate applications of the tool in a comparative study of a restaurant that investigates how plan and furniture layout interventions, ventilation strategies can impact the movement of airborne pathogens in indoor environments.
Content may be subject to copyright.
Towards Safer Work Environments During the COVID-19 Crisis: A Study of Different Floor
Plan Layouts and Ventilation Strategies Coupling Open FOAM and Airborne Pathogen Data
for Actionable, Simulation-based Feedback in Design
Zoe De Simone1, Patrick Kastner1, Timur Dogan1
1Cornell University, Environmental Systems Lab, Ithaca, USA
As work environments struggle to reopen during the
current COVID-19 pandemic, it is crucial to establish
practical decision-aiding tools. While a strong emphasis
has been placed on determining generic guidelines to
reduce the risk of airborne viral spread, there is a lack of
free and easy-to-use simulation workflows to quantify
indoor air quality and the risk of airborne pathogens
indoors at a spatial resolution that can take into account
floor-plan layouts, furniture, and ventilation inlet-outlet
positions. This paper describes the development of a new,
free, early design tool that allows designers and other
stakeholders to simulate and compare airborne viral
concentrations under different indoor conditions. The tool
leverages OpenFOAM-based Computational Fluid
Dynamics (CFD) and a passive scalar simulation
approach to allow architects and interior designers to
quantify airborne pathogens' exposure. The tool is
integrated into the popular Rhino3d & Grasshopper CAD
environment to facilitate its application in fast-paced
design processes. We demonstrate good agreement
compared to a CFD benchmark test. Further, we validate
newly developed COVID-19 capabilities by comparing
our results to an existing restaurant case study that
included tracer gas measurements and validation using
Fluent (Ansys).
We demonstrate applications of the tool in a comparative
study of a restaurant that investigates how plan and
furniture layout interventions, ventilation strategies can
impact the movement of airborne pathogens in indoor
Key Innovations
Definition of the Airborne Pathogen Concentration
(APC): a spatial and case-specific metric to evaluate
the concentration of small SARS-CoV-2 airborne
particles in indoor spaces.
Easy-to-use, early design simulation methodology,
and tool implemented in a popular CAD environment
to facilitate spatial indoor air quality (IAQ) and APC
analysis during the design of floor-plan and furniture
layouts and ventilation inlet-outlet positions.
A case study demonstrating the application Eddy3D
(Indoor) on redesigning a safer restaurant floor plan
Practical Implications
Always simulate IAQ and potential SARS-CoV-2
concentration of space on a case-to-case basis. Avoid
poorly ventilated rooms. Consider inlet and outlet
positions, furniture layout and allow modelers to consider
potential negative implications of multi-occupant zones
serviced by a single air conditioning unit.
As offices, businesses, and schools worldwide reopen and
COVID-19 restrictions are lifted, ensuring indoor spaces'
health and safety is more important than ever. Americans,
on average, spend approximately 90% of their time
indoors, where the concentrations of pollutants, ranging
from biological contaminants such as bacteria and viruses
to building materials, can be up to 2 to 5 times higher than
typical outdoor concentrations (Assessment, 2009; US
EPA, 2017), making indoor air quality a critical factor for
the design of safe and healthy buildings.
Multizone Airflow modeling tools such as COMIS
(Feustel, 1999), CONTAM (Dols and Polidoro, 2015),
and Building Energy Model (BEM) EnergyPlus (Gu,
2007) implement fast Airflow Network solvers for airflow
simulation in multizone BEMs. These models simulate air
movement through openings and rooms using a single air
node per room. While the resulting flows are accurate,
they do not provide the spatial resolution required to
evaluate the impact of design decisions on air quality
metrics. Due to the lack of geometric resolution, shape,
and location of openings, the design of inlets and outlets
of Heating, ventilation, and air conditioning (HVAC)
systems and interior furniture that strongly influence
indoor air movement is not considered in the simulations.
CFD analyses are required at this spatial resolution,
increasing expertise and modeling effort needed from the
user. In early design stages, where design iterations
change quickly, computational overhead can further
hinder the integration of CFD in a design. In this scenario,
high-fidelity Large Eddy Simulations (LES) are too slow
for design workflows that require quick result turnaround
(Zhai et al., 2007; Blocken, 2018). This raises a
significant barrier to the consideration of ventilation and
indoor air quality assessment, especially in design where
decisions such as room layout, placement of vents, and
the configuration of operable windows and doors can
make or break (Chou et al., 1998) the indoor air quality
and thermal comfort of a building. Although the
integration of IAQ performance feedback in the design
process is increasingly important, CFD integration into
computational design workflows remains largely
unexplored. While researchers have successfully
integrated CFD in design workflows, creating tools such
as RhinoCFD plugin for Rhino (CHAM | RhinoCFD,
2020), Butterfly plugin for Rhino (Ladybug Tools |
Butterfly, 2020), Fast Fluid Dynamics (FFD) solver
validated for limited design problems (Zuo and Chen,
2007) and Envi-Met (‘ENVI-met Software Elements -
Wind and Sun’, 2021) limitations such as cost, fluid
dynamics and programming knowledge remain a barrier
to a broader implementation of CFD in design workflows.
Eddy3D (Kastner and Dogan, 2020) is an example of an
accessible and easy-to-use CFD design tool in
Rhino/Grasshopper but currently only supports outdoor
wind analyses. The goal of our research is to create an
equally easy-to-use and accessible tool for indoor
The need for indoor air assessment tools for design and
decision-making has become particularly evident in the
current global pandemic. The international call to action
on airborne transmission in early March (Morawska and
Milton, 2020) underscores the importance of simulating
the spread of airborne viral particles in indoor
environments, as airborne microdroplets released during
exhalation, talking, and coughing is small sufficient to
remain aloft in the air pose a risk of exposure at tens of
meters from an infected individual. The size of carrier
fluid droplets (0.2 μm -100 μm) is critical as it determines
settling velocity and time, distance of travel, and
deposition location in the respiratory system(Stilianakis
and Drossinos, 2010). While coughing and sneezing
produce larger airborne particles (100-1000 μm diameter)
that settle within 1 s (Somsen et al., 2020), small particles
(110 μm), produced when breathing, decay within 8 to
14 minutes (Stadnytskyi et al., 2020). Once airborne,
small droplets dehydrate, slowing their fall (Wells, 1934)
ranging between 30 to 177 min depending on
microclimate boundary conditions (Smither et al., 2020),
including ventilation rates (Somsen et al., 2020), thermal
and radiation environment. In addition to decay rate, other
factors such as airflow pattern direction make viral
transmission risk assessment a highly spatial and case-
specific modeling problem. Reported transmission routes
at a restaurant in Guangzhou, China (Lu et al., 2020; Li et
al., 2021) demonstrate the importance of simultaneously
simulating seating arrangements and ventilation systems.
Forensic analysis of the restaurant layout, seating
arrangements, and smear samples from air-conditioning
inlets and outlets led to the belief that the transmission
was likely due to a nearby air-conditioner and insufficient
outside Air Changes per Hour (ACH). As demonstrated in
the Guangzhou episode, source and distribution path of
ventilation play a key role in the potential for virulent air
circulation. This study presents three innovations that
facilitate the analysis of different design strategies and
their impact on potential indoor viral concentrations:
An optimized and accessible CFD workflow to simulate
Indoor Airflow on a case-to-case basis.
A methodology that allows unspecialized users to
visualize, quantify and compare viral air concentration
of design strategies leveraging OpenFOAM and
A case study demonstrating the application of the
workflow towards the redesign of a safer indoor space.
This paper implements IAQ, SARS-CoV-2 particle
spread simulation capabilities and workflows into the
existing tool. We implement steady-state interior
Reynolds-averaged Navier–Stokes (RANS) CFD
simulations with the 'buoyantSimpleFoam' solver
utilizing a passive scalar method to evaluate "Age-of-Air"
and derived particle concentration fields such as CO2. The
Age-of-Air is the mean time a particle takes to travel from
an inlet to the simulation domain's measurement point. It
is commonly used to calculate air-change effectiveness
and identify 'dead ventilation zones' in buildings. Age-of-
Air is a variable that is highly variable in 3D space and
provides information that cannot always be elicited by
evaluating simple velocity fields.
Since exhaled breath is the vehicle for the airborne release
of SARS-CoV-2 and other infectious particles as well as
the primary internal source of CO2 in buildings, we
leverage CO2 concentration tracers as a proxy for exhaled-
breath exposure in buildings to measure potential viral
particle accumulation and the Risk of Viral Airborne
Infection (Rudnick and Milton, 2003). Furthermore,
droplet nuclei smaller than (5–10 μm) have been
simulated with tracer gas such as “CO2 or N2O, because
the settling velocity is very low(Tang et al., 2011).
We focus on simulating small airborne particles in the
range of (0.1–2 μm), also referred to as accumulation
particles, (Nazaroff, 2004) which make up a large portion
of indoor particle concentration and have minimum
deposition rate (Gao and Niu, 2007) that can be neglected
in simulation (Zhang and Chen, 2006) and have the most
similar behavior to tracer gas (Bivolarova et al., 2017).
Our focus on small particles is further supported by the
correlation between viral airborne transmission paths,
infectious dose, and severity of disease. While large
droplets are sprayed onto the body, a form of contact
transmission, aerosols are inhaled into the respiratory
system (Roy and Milton, 2004) (Wells, 1934) (Xie et al.,
2007) (Tellier et al., 2019) and can be more harmful
(Morawska et al., 2009).
Existing guidelines suggest that increasing ventilation
rates and air distribution can often be cost-effective means
of reducing indoor pollutant levels. (Stewart et al., 2020)
However, it is unclear how the orientation and
arrangement of ventilation systems are most beneficial
and the impact of furniture layout and other larger scale
objects in the space on airflow patterns. In this paper, we
demonstrate and deploy our fast IAQ tool for COVID-19
safety to study the spread at the Guangzhou restaurant and
evaluate the impact of spatial and case-specific designs to
answer questions as:
How can restaurants increase occupant density while
ensuring healthy air quality?
How can the design and arrangement of seating and
furniture promote air quality?
Are plastic partitions effective, and how can their
efficacy be maximized?
How can AC units be configured to ventilate a space
Airborne SARS-CoV-2 Design Metric
The early design decision-making tool and workflow
allows designers to evaluate and compare potential
distribution and concentration of small viral SARS-CoV-
2 airborne particles in indoor spaces. The metric,
Airborne Pathogen Concentration (APC), describes how
much space can be considered "well-ventilated" based on
Age-of-Air and CO2 concentration simulations. To ensure
accessibility and readability of the simulations, the
authors developed a viral concentration scalar field in
Eddy3D that allows users to gain visual feedback to easily
locate underperforming and poorly ventilated areas of the
design when viewing the OpenFOAM simulation results
in Paraview. Thus, the tool allows an iterative process,
where relationship between localized air quality and floor
plan design can be studied.
This metric has the advantage of defining regions unsafe
for occupancy and comparing floor plan designs based on
ventilation potential at the same time. As infectious dose
for SARS-CoV-2 is still under debate, the boundaries of
the metric are not absolute. Instead, the tool is used to
study space and ventilation iterations relative to one
another. To demonstrate the applicability and define
limitations of the metric in this study, we conduct a series
of restaurant design iterations.
Simulation Environment and Tools
Most indoor air quality simulation tools using CFD are
only accessible to highly specialized users. This paper
aims to create a simple and easy-to-use tool integrated
within popular 3D modeling software and design
workflows. To this end, the presented research builds on
four tools for simulation and geometric modeling:
OpenFOAM, ParaView, the McNeel Rhino Platform, and
Eddy3D for Grasshopper. OpenFOAM is open-source
software for CFD that is free and has been validated
thoroughly. Paraview is an open-source visualization
application that allows to view 3D CFD simulations
(‘ParaView’, 2021). The Rhino3d/Grasshopper CAD
environment was chosen due to its popularity amongst
architects allowing for easy integration into existing
design workflows. Eddy3D, an existing Computational
Fluid Dynamics modeling plugin for Grasshopper using
the OpenFOAM solver, is expanded to have Indoor CFD
Indoor CFD Setup Validation
We implement steady-state interior RANS CFD
simulations with OpenFOAM's 'buoyantSimpleFoam', a
thoroughly validated CFD code (Limane, Fellouah and
Galanis, 2015). To ensure a valid set-up of the indoor case
using OpenFOAM, we simulate a CFD Benchmark Test
case(Nielsen et al., 2003), a Mixing Validation
simulation, in the Rhino/Grasshopper environment. The
test consists of a unidirectional flow field, with a wind
tunnel of 2.44 m x 1.20 m x 2.46 m, circular exhaust
openings 0.25 m in diameter, and located 0.6 m from the
floor and the ceiling, temperature of 22 ºC and 76 W heat
Particle Emissions Data
We implement experimental exhalation emissions to add
capabilities to assess SARS-CoV-2 particle
concentration. Breathing and speaking emission data
including breathing and speaking flow, gas flow rate
(L/s), flow speed (m/s), breathing cone angle (˚), breath
frequency (1/s), particles per breath (corresponding to
peak of particle size distribution in breathing
experiments) for the simulation of an infected emitter is
based on in situ experimental data from the University of
Minnesota (Table1) (Shao et al., 2021).
Average flow
Rate (L/s)
Cone Angle
Speed (m/s)
30mm from mouth
Particles/L of exhaled gas
Peak particle size (µm)
Table 1: Breathing and Violent Expiratory Behavior
flow data (Shao et al., 2021) averaged over one breath
per 4 seconds.
The overall structure of the workflow is as follows: the
Rhino3d/Grasshopper CAD environment is used to model
the Indoor Space, Inlets and Outlets, Interior partitions
and furniture, and people in the space. The boundary
space is modeled as a closed surface with openings
corresponding to Inlet and Outlet locations, and the
corresponding temperature is reported. Inlets and Outlets
are modeled as surfaces with corresponding velocity
vectors and are subtracted from the boundary geometry
surface. Indoor partitions, furniture, and people are
modeled as closed surfaces in the space and connected to
the geometrical boundary. Emitter geometry is simplified
and modeled as two parallelepipeds with an 8 mm
diameter aperture corresponding to the mouth and nose
region's location, and associated heat flux is specified.
Once the geometry, ventilation speed, temperature, and
heat flux are specified in the Rhino/Grasshopper
environment, the simulation is run. The simulation reports
data slices at user-selected probing locations and graphs
that can be viewed in the ParaView: Age-of-Air (AOA)
and SARS-CoV-2 concentration.
SarsCoV2 Simulation Validation and Model
The validation is carried out by comparing SARS-CoV-2
concentrations of our simulations with tracer gas
measurements in 8 points of the Guangzhou Restaurant
floor plan and 18 normalized predicted concentration
values from the respective analysis (Li et al., 2020). The
COVID-19 spreading event occurred on the third floor of
a five-floor, air-conditioned building without windows.
On the day of the event, 91 people, of whom 83
customers, sitting at 15 tables approximately 1 m apart,
and eight staff members were in the space. Ten out of 83
customers, including one infector, became ill with
COVID-19, all of whom were sitting at three tables (A, B,
C) located near patient A1 and within the flow of air
conditioner 1, located above table C. Family A was seated
for 1 hour and 22 minutes. Families A and B were each
seated for an overlapping period of 53 minutes and
families A and C for an overlapping period of 73 minutes.
The case study is recreated in the Rhino CAD
environment. Dimensions reported by the case study
include dimensions of the space, exhaust fans' position,
occupant locations, and furniture. The restaurant has a
near-rectangular plan with a restroom, elevator, and fire
stair located on the south. There are five fan coil AC units,
four along the east-facing wall, one near the fire stair.
Measured ventilation rates are not reported in the case
study. We extrapolate three possible rates (230, 320,400
m3/h) based on AC unit type and test them for the best
There are four exhaust fans located along the west-facing
wall and one located in the restroom. At the time of the
incident, only the fan in the restrooms was functioning.
Figure 1: Restaurant floor plan adapted from (Li et al.,
2020). Each table is numbered A through 18. Patient
A1(red) at table A is the index occupant, while A2A5,
B1B3, and C1C2 are the individuals who became
infected with COVID-19(orange).
Therefore, it is the only fan modeled in our simulation.
Tables are distinguished into four types based on shape
and size and modeled as extrusions with 5 cm thickness.
People and chairs are modeled as one and simplified into
two parallelopipeds: Body and chair (1.07 x 0.3 x 0.4),
and head (0.22 x 0.22 x 0.22).
Emitters are modelled as two parallelepipeds with an 8-
mm diameter aperture corresponding to the mouth and
nose region's location, and respective heat flux is
specified (Figure 1).
Restaurant Design Case Study
We conduct a case study of the restaurant, comparing the
influence of variations in furniture and partition layout,
table distancing, and occupancy on indoor air quality and
potential SARS-CoV-2 concentrations.
1. Influence of Plastic Partitions: Plastic Partitions are
placed between tables in their existing position in
order to analyze the effect of temporary barriers.
2. Influence of Furniture Layout and De-densification:
Viral accumulation in the base case is compared to
that of a dedensified space. The base case is de-
densified, and tables are removed from the space,
leaving only 47 occupiable seats compared to 83 in
the original case. The furniture is organized such that
there are two tables per AC unit.
Indoor CFD Setup Validation
The case setup used in the paper is validated by comparing
velocity measurements at eight grid points (Figure 3-4)
with corresponding velocities in the Mixing Validation
Benchmark test as well as through visual correspondence
(Figure 2). Velocities are measured along the plane at z=0
m and x=1.69 m with two inlet ventilation rates,
corresponding to high (0.5 m/s) and medium (0.2 m/s)
flow rate.
Figure 2: Simulation of the Mixing Ventilation
Benchmark Tests with 0.2 m/s inlet velocity.
The simulations show that our implementation in Eddy3D
(Indoor) has a Root Mean Square Error (RMSE) of
0.125m/s and Mean Absolute Percentage Error (MAPE)
of 28.5% (Figure 3). The plots show that the largest error
occurs in proximity to the mannequin. While this method
is not the most precise due to a simplification of the
mannequin mesh, the error is acceptable within the
bounds of its intended use: the early-design stage, where
meshes are not precisely defined or fixed beyond simple
Figure 3: Comparison of normalized velocity
measurements of the Mixing Ventilation Benchmark
Tests and our simulations (Eddy3D) at z(m)=0,
x(m)=1.69, y(m)= 0.275, 0.550, 0.825, 1.100, 1.375,
1.650, 1.925, 2.200; with 0.2m/s (left) and 0.5m/s (right)
inlet velocity.
SARS-CoV-2 Simulation Validation
The validation of the precision and efficacy of Eddy3D
(Indoor) viral capabilities is carried out by comparing
SARS-CoV-2 concentrations simulated using Eddy3D
with tracer gas measurements and predictions modelled
with Ansys, Fluent(Li et al., 2021). The study reported
tracer gas measurements collected in 8 points of the
Guangzhou Restaurant floor plan and 18 normalized
predicted concentration values (Lu et al., 2020). In the
experimental setup, ethane gas was released through an 8
mm inner diameter pipe at a speed of 1.5 m/s at 3234°C,
with the pipe outlet placed in proximity of the index
patient's nose, marked as A1 at table A. In our simulation,
the emitter is modeled as the experimental setup, with a
breathing flow of 1.5 m/s and an opening of 8 mm in
diameter. Passive Scalar simulations of SARS-CoV-2
concentrations, Age-of-Air, and Velocity Streamlines are
performed using Eddy3D (Indoor).
The simulations show that Eddy3D (Indoor) has results
comparable to the measured concentration and a Mean
Absolute Percentage Error (MAPE) of 14.7 % (Table 2).
The concentration at A1, the index patient at table A, A2
A5, B1B3, and C1–C2 is higher than concentrations in
proximity to other restaurant occupants (Figure 4).
Furthermore, we can see that Velocity Streamlines from
Eddy3D (Indoor) (Figure 4) are comparable to those
computed using CFD software package Fluent (Ansys
Fluent, USA)(Li et al., 2021).
Velocity plots (Figure 5) support the notion that small
aerosol particles emitted from the index patient were
transported to contiguous tables B and C by the nearby
A/C unit. We note that both our simulations and the
referenced simulations assume rapid droplet evaporation,
approximate the exhaled droplet nuclei as a passive scalar,
and the deposition effect is neglected.
(Li et al.,
(Li et al.,
Table 2: Comparison of normalized tracer gas
concentration measurements (Column 2)(Li et al., 2021),
Fluent simulation (Column 3)(Li et al., 2021) and
Eddy3D(Indoor) simulation(Column 4) at each table.
Figure 4: SARS-CoV-2 concentration distribution of the
Base case computed using Eddy3D(Indoor).
Figure 5: Velocity of the Base case computed using
Restaurant Design Case Study
The Guangzhou restaurant validation simulation
demonstrates AC's role in SARS-CoV-2 particle transport
from the infected table to surrounding tables. To
demonstrate the use and efficacy of Eddy3D (Indoor) as a
tool for the design of safer indoor spaces, we compare the
base case with two interior design alternatives to improve
the air quality and reduce SARS-CoV-2 concentrations
inside the space.
The first design iteration considers the influence of
Furniture Layout and Plastic Partitions. Plastic partitions
are laid out between tables without considering movement
and table accessibility. Partitions are placed such that
tables with occupants closer than 6ft are separated by
partitions, while tables more than 6ft apart are not
separated by partitions.
Velocity (Figure 7) and SARS-CoV-2 concentration
(Figure 6) plots show that the air movement is
significantly altered by adding plastic partitions. While
virus concentrations are lower in proximity of table B, as
airflow from the air conditioner is blocked by the
partition, concentrations in proximity of tables 17, 16 and
13 are higher when partitions are added to the space. This
simulation demonstrates that plastic partitions do not
always ameliorate IAQ and SARS-CoV-2 concentrations,
and supports the need to model spaces on a case-to-case
The third case study analyses the influence of Furniture
Layout and De-densification. The base case is de-
densified, and tables are removed from the space, leaving
only 47 occupiable seats compared to 83 in the original
case. The furniture is organized such that multiple tables
are not aligned along the same AC unit ventilation
direction. We compute SARS-CoV-2 concentration
(Figure 8) and Velocity Streamlines (Figure 9) for each
design iteration using Eddy3D (Indoor) and compare
them to the Base Case and Partition restaurant design
While the high SARS-CoV-2 concentrations in the de-
densified case are primary located near table A and
gradually decrease with the distance, high-concentrations
in the de-densified case are not uniform and less
predictable (Figure 8).
Velocity in the Base case and De-densified case are
similar (Figure 10). However, concentrations are
significantly different. The results suggest that furniture
plays a large role in air distribution and the spreading of
the virus.
Furthermore, SARS-CoV-2 concentrations are overall
lowest in the Base case and higher in De-densified case,
as there are fewer obstacles to the diffusion of the virus
(Figure 10). From the Partition case is it clear that plastic
barriers increase concentration in proximity of the
emitter, with exception of Table B, which is effectively
blocked from the AC units’ airflow.
Figure 6: SARS-CoV-2 concentration distribution of the
Partitioned case computed using Eddy3D(Indoor).
Figure 7: Velocity of the Partitioned case computed
using Eddy3D (Indoor)
Figure 8: SARS-CoV-2 concentration distribution of the
De-densified case computed using Eddy3D(Indoor).
Figure 9: Velocity of the De-densified case computed
using Eddy3D (Indoor).
Figure 10: SARS-CoV-2 concentration of the Base case,
Partition and De-densified restaurant floor plan.
The results have shown that Eddy3D (Indoor) yields
reliable results for indoor air studies. The precision with
a 0.125m/s RMSE for the velocity values is caused by the
simplification of the mannequin mesh and is acceptable
within the bounds of the precision of the underlying
simulation engine OpenFOAM. The results of the
restaurant case study yield reliable results for the
concentration of airborne infection. Concentration values
yield a MAPE of 14.7%, which is acceptable within the
bounds of OpenFOAM. As expected, Air Stagnation and
Airborne Infection Concentration are identified in the
proximity of infected occupants.
Case Studies
The case studies show that furniture and people have a
large influence on the diffusion of SARS-CoV-2 and can
be effective means of reducing diffusion of the virus to
other spaces in the floor plan. Results show that plastic
partitions help block the diffusion of the virus from table
A to contiguous table B, but direct air towards other parts
of the floor plan and result in higher diffusion to tables
17,16,13. The comparison of the three layout shows that
it is necessary to precisely simulate floor plans for viral
Future Improvement
The authors plan to consider natural ventilation in
addition to mechanical ventilation in the future
development of the tool.
A common drawback of design tools and corresponding
metrics is difficulty in judging and comparing computed
results. The authors plan to mitigate this inconvenience by
generating a database of viral assessments, ranging in
floor plan shape, layout, and ventilation strategy. This tool
could be accompanied by a guide to compare designs with
an optimum case.
How the tool should not be used
As data and research on SARS-CoV-2 evolve, this tool
should be considered work-in-progress and reflect current
data and knowledge as such a tool should not be
considered a perfect airborne transmission model.
Furthermore, as an infectious dose for SARS-CoV-2 is
still under debate, the metric's limitations are not absolute.
Instead, the tool is used to study space and ventilation
iterations relative to one another and to identify
improvements throughout the early design process. The
authors plan to integrate such boundaries when they
become available.
The novel approach to integrate indoor airflow pattern
and viral transmission risk assessment into CAD software
allows tackling highly spatial and case-specific modeling
problems. From a design standpoint, the tool
revolutionizes the way IAQ can be analyzed and
integrated into early design workflows. Furthermore, the
tool introduces airborne viral assessment into the design
repertoire, allowing designers to study the effect of floor
plan designs and layouts, ventilation strategies on indoor
air quality and health.
We thank Khaled Hashad (Cornell University) for
insightful discussions on particle deposition. We further
would like to thank the Hunter R. Rawlings III Cornell
Presidential Research Scholars Program and the Cornell
Atkinson Center for funding this research.
Assessment, U. E. N. C. for E. (2009) The total exposure
assessment methodology (TEAM) study: Summary and
analysis: Volume I. Available at:
erence_id/3004792 (Accessed: 26 October 2020).
Bivolarova, M. et al. (2017) ‘A comparison between
tracer gas and aerosol particles distribution indoors: The
impact of ventilation rate, interaction of airflows, and
presence of objects’, Indoor Air, 27(6), pp. 12011212.
CHAM | RhinoCFD (2020). Available at: (Accessed: 3 February 2021).
Chou, P. C. et al. (1998) ‘Effects of Window Position on
the Air Flow Distribution in a Cross-Ventilated
Residential Bedroom’, Indoor and Built Environment,
7(56), pp. 300307. doi:
Dols, W. S. and Polidoro, B. J. (2015) ‘CONTAM User
Guide and Program Documentation Version 3.2’, (1).
Available at:
(Accessed: 30 December 2020).
‘ENVI-met Software Elements - Wind and Sun’ (2021)
ENVI-met. Available at: https://www.envi- (Accessed: 3 February 2021).
Feustel, H. E. (1999) ‘COMISan international
multizone airflow and contaminant transport model’,
Energy and Buildings, 30(1), pp. 318. doi:
Gao, N. P. and Niu, J. L. (2007) ‘Modeling particle
dispersion and deposition in indoor environments’,
Atmospheric Environment, 41(18), pp. 3862–3876. doi:
Gu, L. (2007) ‘Airflow network modeling in energyplus’,
IBPSA 2007 - International Building Performance
Simulation Association 2007.
Kastner, P. and Dogan, T. (2020) ‘A cylindrical meshing
methodology for annual urban computational fluid
dynamics simulations’, Journal of Building Performance
Simulation, 13(1), pp. 59–68. doi:
Ladybug Tools | Butterfly (2020). Available at: (Accessed: 3
February 2021).
Li, Y. et al. (2021) ‘Probable airborne transmission of
SARS-CoV-2 in a poorly ventilated restaurant’, Building
and Environment, 196, p. 107788. doi:
Limane, A., Fellouah, H. and Galanis, N. (2015)
‘Thermo-ventilation study by OpenFOAM of the airflow
in a cavity with heated floor’, Building Simulation, 8(3),
pp. 271–283. doi: 10.1007/s12273-014-0205-4.
Lu, J. et al. (2020) ‘COVID-19 Outbreak Associated with
Air Conditioning in Restaurant, Guangzhou, China,
2020’, Emerging Infectious Diseases, 26(7), p. 4.
Morawska, L. et al. (2009) ‘Size distribution and sites of
origin of droplets expelled from the human respiratory
tract during expiratory activities’, Journal of Aerosol
Science, 40(3), pp. 256269. doi:
Morawska, L. and Milton, D. K. (2020) ‘It is Time to
Address Airborne Transmission of COVID-19’, Clinical
Infectious Diseases, p. ciaa939. doi: 10.1093/cid/ciaa939.
Nazaroff, W. W. (2004) ‘Indoor Particle Dynamics’,
Indoor Air, 14(Supplement 7), pp. 175183.
Nielsen, P. V. et al. (2003) ‘Benchmark Tests for a
Computer Simulated Person’, p. 8.
‘ParaView’ (2021), 24 January. Available at: (Accessed: 2 April 2021).
Roy, C. and Milton, D. (2004) ‘Airborne Transmission of
Communicable Infection - The Elusive Pathway’, The
New England journal of medicine, 350, pp. 17102. doi:
Rudnick, S. N. and Milton, D. K. (2003) ‘Risk of indoor
airborne infection transmission estimated from carbon
dioxide concentration’, Indoor Air, 13(3), pp. 237–245.
doi: 10.1034/j.1600-0668.2003.00189.x.
Shao, S. et al. (2021) ‘Risk assessment of airborne
transmission of COVID-19 by asymptomatic individuals
under different practical settings’, Journal of Aerosol
Science, 151, p. 105661. doi:
Smither, S. J. et al. (2020) ‘Experimental aerosol survival
of SARS-CoV-2 in artificial saliva and tissue culture
media at medium and high humidity’, Emerging Microbes
& Infections, 9(1), pp. 14151417. doi:
Somsen, G. A. et al. (2020) ‘Small droplet aerosols in
poorly ventilated spaces and SARS-CoV-2 transmission’,
The Lancet Respiratory Medicine, 8(7), pp. 658–659. doi:
Stadnytskyi, V. et al. (2020) ‘The airborne lifetime of
small speech droplets and their potential importance in
SARS-CoV-2 transmission’, Proceedings of the National
Academy of Sciences, 117(22), pp. 1187511877. doi:
Stewart, E. J. et al. (2020) ‘ASHRAE Position Document
on Infectious Aerosols’, p. 24.
Stilianakis, N. I. and Drossinos, Y. (2010) ‘Dynamics of
infectious disease transmission by inhalable respiratory
droplets’, Journal of the Royal Society Interface, 7(50),
pp. 1355–1366. doi: 10.1098/rsif.2010.0026.
Tang, J. W. et al. (2011) ‘Observing and quantifying
airflows in the infection control of aerosol- and airborne-
transmitted diseases: an overview of approaches’, Journal
of Hospital Infection, 77(3), pp. 213222. doi:
US EPA, O. (2017) Indoor Air Quality, US EPA.
Available at:
environment/indoor-air-quality (Accessed: 26 October
American journal of Epidemiology, 20(3), pp. 611618.
Zhang, Z. and Chen, Q. (2006) ‘Experimental
measurements and numerical simulations of particle
transport and distribution in ventilated rooms’,
Atmospheric Environment, 40(18), pp. 3396–3408. doi:
Zuo, W. and Chen, Q. Y. (2007) ‘Validation of fast fluid
dynamics for room airflow’, in. Building Simulation
2007, BS 2007. Available at:
n-of-fast-fluid-dynamics-for-room-airflow (Accessed: 3
February 2021).
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
SARS-CoV-2, the causative agent of the COVID-19 pandemic, may be transmitted via airborne droplets or contact with surfaces onto which droplets have deposited. In this study, the ability of SARS-CoV-2 to survive in the dark, at two different relative humidity values and within artificial saliva, a clinically relevant matrix, was investigated. SARS-CoV-2 was found to be stable, in the dark, in a dynamic small particle aerosol under the four experimental conditions we tested and viable virus could still be detected after 90 minutes. The decay rate and half-life was determined and decay rates ranged from 0.4 to 2.27 % per minute and the half lives ranged from 30 to 177 minutes for the different conditions. This information can be used for advice and modelling and potential mitigation strategies.
Full-text available
Speech droplets generated by asymptomatic carriers of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) are increasingly considered to be a likely mode of disease transmission. Highly sensitive laser light scattering observations have revealed that loud speech can emit thousands of oral fluid droplets per second. In a closed, stagnant air environment, they disappear from the window of view with time constants in the range of 8 to 14 min, which corresponds to droplet nuclei of ca. 4 μm diameter, or 12- to 21-μm droplets prior to dehydration. These observations confirm that there is a substantial probability that normal speaking causes airborne virus transmission in confined environments.
Full-text available
In some emergencies, such as fire or accidental release of chemical/biological agents in buildings, it is very useful to simulate the flow on real time or even faster than real time so that proper measures can be taken to minimize casualties. The traditional computational fluid dynamics (CFD) simulation of fire or transient contaminant transport in buildings is accurate but too time consuming, such as by using unsteady Reynolds averaged Navier-Stokes equations (URANS) and large eddy simulation (LES). On the other hand, multizone flow network modeling is fast, but its accuracy is poor. Therefore, a new CFD technology, named Fast Fluid Dynamics (FFD), was developed. The FFD is faster than traditional CFD, and more accurate than multizone modeling. This paper shows the validation of the FFD through three cases: (1) flow in a lid-driven cavity; (2) flow in a plane channel; and (3) flow in a ventilated room. The results conclude that the FFD method can simulate the flows faster than real time, although some discrepancies exist between the numerical results and experimental data. The discrepancies are acceptable for the emergency management.
Although airborne transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been recognized, the condition of ventilation for its occurrence is still being debated. We analyzed a coronavirus disease 2019 (COVID-19) outbreak involving three families in a restaurant in Guangzhou, China, assessed the possibility of airborne transmission, and characterized the associated environmental conditions. We collected epidemiological data, obtained a full video recording and seating records from the restaurant, and measured the dispersion of a warm tracer gas as a surrogate for exhaled droplets from the index case. Computer simulations were performed to simulate the spread of fine exhaled droplets. We compared the in-room location of subsequently infected cases and spread of the simulated virus-laden aerosol tracer. The ventilation rate was measured using the tracer gas concentration decay method. This outbreak involved ten infected persons in three families (A, B, C). All ten persons ate lunch at three neighboring tables at the same restaurant on January 24, 2020. None of the restaurant staff or the 68 patrons at the other 15 tables became infected. During this occasion, the measured ventilation rate was 0.9 L/s per person. No close contact or fomite contact was identified, aside from back-to-back sitting in some cases. Analysis of the airflow dynamics indicates that the infection distribution is consistent with a spread pattern representative of long-range transmission of exhaled virus-laden aerosols. Airborne transmission of the SARS-CoV-2 virus is possible in crowded space with a ventilation rate of 1 L/s per person.
The lack of quantitative risk assessment of airborne transmission of COVID-19 under practical settings leads to large uncertainties and inconsistencies in our preventive measures. Combining in situ measurements and computational fluid dynamics simulations, we quantify the exhaled particles from normal respiratory behaviors and their transport under elevator, small 15 classroom, and supermarket settings to evaluate the risk of inhaling potentially virus-containing particles. Our results show that the design of ventilation is critical for reducing the risk of particle encounters. Inappropriate design can significantly limit the efficiency of particle removal, create local hot spots with orders of magnitude higher risks, and enhance particle deposition causing surface contamination. Additionally, our measurements reveal the presence of 20 a substantial fraction of faceted particles from normal breathing and its strong correlation with breathing depth.
For urban CFD simulations, it is considered a best practice to use a box-shaped simulation domain. Box-shaped domains, however, show drawbacks for airflow from several wind directions as remeshing and additional preprocessing steps become necessary. We introduce a routine to create a cylindrical mesh that expedites the simulation of arbitrary wind directions using OpenFOAM. Results computed with the cylindrical domain are validated against wind tunnel data. We report that the cylindrical method yields comparable results in terms of accuracy and convergence behaviour. Further, run time comparisons in a real-world scenario are conducted to discuss its advantages and limitations. Based on the findings, we recommend using the cylindrical approach if at least eight wind directions are analyzed for which we report 18% run time savings. The cylindrical domain along with automated best practice boundary conditions has been implemented in Eddy3D – a plugin for Rhinoceros.
Results of an investigation of the effects of window position on the airflow characteristics for a typical bedroom setting in Taiwan are presented. Four different window positions were examined in the experiment which used a full-scale laboratory bedroom model with a single bed. A three-dimensional ultrasonic anemometer was used to measure airflow distribution and the results of flow measurements at two height levels are presented. Computer simulation of the airflow distribution was performed using the standard k-ε turbulence model. The measurements and the computer calculations resulted in similar airflow distributions for all positions of window openings. Close congruence between the results of calculations and those of the measurements shows the validity of using such a computer simulation in the airflow design of a residential bedroom. The results also show that the positions of window openings have appreciable effects on the airflow distribution. Proper window position is therefore an important factor in the design of ventilation for a cross-ventilated bedroom.
Technical Report
This manual describes the computer program CONTAM version 3.2, developed by NIST. CONTAM is a multizone indoor air quality and ventilation analysis program designed to help determine airflows, contaminant concentrations, and personal exposure in buildings. Airflows include infiltration, exfiltration, and room-to-room airflow rates and pressure differences in building systems, and can be driven by mechanical means, wind pressures acting on the exterior of the building, and buoyancy effects induced by temperature differences between zones, including the outdoors. Contaminant concentrations include the transport and fate of airborne contaminants, due to airflow, chemical and radio-chemical transformation, adsorption and desorption to building materials, filtration, and deposition to and resuspension from building surfaces. Personal exposure includes the exposure of building occupants to airborne contaminants, for eventual risk assessment. CONTAM can be useful in a variety of applications. Its ability to calculate building airflow rates and relative pressures between zones of the building is useful for assessing the adequacy of ventilation rates in a building, for determining the variation in ventilation rates over time, for determining the distribution of ventilation air within a building, for estimating the impact of envelope air-tightening efforts on infiltration rates, and for evaluating the energy impacts of building airflows. The program has also been used extensively for the design and analysis of smoke management systems. The prediction of contaminant concentrations can be used to determine the indoor air quality performance of buildings before they are constructed and occupied, to investigate the impacts of various design decisions related to ventilation system design and building material selection, to evaluate indoor air quality control technologies, and to assess the indoor air quality performance of existing buildings. Predicted contaminant concentrations can also be used to estimate personal exposure based on occupancy patterns. Version 2.0 contained several new features including: non-trace contaminants, practically unlimited number of contaminants, contaminant-related libraries, separate weather and ambient contaminant files, building controls, scheduled zone temperatures, improved solver to reduce simulation times and several user interface related features to improve usability. Version 2.1 introduced more new features including the ability to account for spatially varying external contaminants and wind pressures at the building envelope, more new control elements, particle-specific contaminant properties, total mass released calculations and detailed program documentation. Version 2.4 introduced two new deposition sink models, a one-dimensional convection/diffusion contaminant model for ducts and user-selectable zones, new contaminant filter models, control super nodes, super filters, a duct balancing tool, building pressurization and model validity tests and several other usability enhancements. Version 3.0 added a deposition with resuspension source/sink model, a self-regulating vent airflow model, and integrated coupling between Computational Fluid Dynamics (CFD) and multizone modeling. Version 3.1 introduced a variable time step ordinary differential equation solver (VODE) for solving the contaminant equations and coupling between the TRNSYS energy simulation program and CONTAM to enable combined energy, airflow and contaminant transport analysis. Version 3.2 added an ultra-violet germicidal irradiation (UVGI) filter model, contaminant exfiltration result files, a SQLite database result file, and co-simulation between the EnergyPlus whole building energy analysis program and CONTAM.