Predicting Contamination Spread Inside a Hospital Breakroom with Multiple Occupants Using High Fidelity Computational Fluid Dynamics Simulation on a Virtual Twin

Abstract

Mitigating the rise and spread of contaminants is a major challenge faced during any contagious disease outbreak. In densely occupied areas, such as a breakroom, the risk of cross-contamination between healthy and infected individuals is significantly higher, thereby increasing the risk of further spread of infectious diseases. In this study, a high fidelity transient fluid solver and Lagrangian particle-based method were used to predict the airflow distribution and contaminant transmission inside a detailed 3D virtual twin of an emergency hospital breakroom. The solver efficiently captured the contaminants emitted simultaneously from multiple talking occupants as well as their propagation inside the breakroom. The influence of airflow distribution on the aerosol spread inside the breakroom for two different air conditioning vent positions was demonstrated with all occupants and with reduced occupants. The baseline simulation with all occupants in the breakroom showed a higher risk of contamination overall as well as between adjacent occupants. It was observed that there was a 26% reduction in the contaminants received by the occupants with the proposed modified vent arrangement and a 70% reduction with the scenarios considering a reduced number of occupants. Furthermore, the fomite deposition and cross-contamination between adjacent humans significantly changed with different ventilation layouts. Based on the simulation results, areas with higher contaminant concentrations were identified, providing information for the positioning of UV lights in the breakroom to efficiently eliminate/reduce the contaminants.

1. Introduction

Hospitals around the world have been on the frontlines of the coronavirus disease 2019 (COVID-19) pandemic. They have endured historic challenges across a wide spectrum of disciplines. Issues related to therapeutics, infection control, human resources, and finances have forced healthcare institutions to re-align their strategies and priorities and to explore innovative solutions to new challenges. Not least among these is the continued struggle to protect facilities, staff, and patients from a microbial threat in the face of constantly evolving evidence. Numerous regulatory bodies have also been employed to help in this effort. Organizations such as the Occupational Safety and Health Administration (OSHA) [1], the Center for Disease Control and Prevention (CDC) [2], the American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) [3], and the World Health Organization (WHO) [4] continuously propose updated guidelines on ventilation inside hospitals, working methods, and infection control.

Significant research has been performed on contaminant transmission sources, droplet sizes under different scenarios, ventilation systems, etc. [5,6,7], in order to gain a better understanding of virus transmission and infection. Some of the most common assumptions related to the airborne transmission of viruses, as well as the presence and transmissibility of SARS-CoV-2 through the inhalation of airborne particles, were examined [8]. SARS-CoV-2 RNA infectious particles have been found in aerosols ranging from 0.25 to >4 μm [9,10]. Aerosols are released during expiratory processes such as coughing, breathing, sneezing, and speaking from both healthy and infected individuals [11,12]. Unlike droplets, aerosols are smaller particles that linger in the air, accumulating in poorly ventilated spaces, and are inhaled by individuals over short- and long-distance ranges. Studies on saliva droplet sizes and respiratory secretions from sneezing, coughing, speaking, and breathing show their dependency on the properties of the generation processes such as exhaled air velocity, the viscosity of fluid, the flow path through the nose and mouth, as well as environmental conditions [13]. The relationship between droplet size, evaporation, and falling rate has been explored [14]. A cloud of pathogen-bearing droplets of different sizes is released when an individual infected with a respiratory illness coughs or sneezes. These droplets can travel up to 7–8 m from the point of origin [15,16]. Studies have shown [17] that larger droplets exhaled during a sneeze at an air velocity of 50 m/s can travel up to 6 m. Smaller particles, which are not heavily influenced by gravity, can move upwards from thermal plumes created by body heat and can be carried further by airflow from ventilation, door/window movements, and other effects [18]. Increased airflow along with wider spaces in outdoor environments contribute to a greater dispersion compared to indoor spaces, which are restricted by surrounding walls, doors, and local heating, ventilation, and air conditioning (HVAC) ventilation rates. For this reason, outdoor spaces are considered more protective against the transmission of airborne pathogens compared to indoor spaces [19,20].

Mitigating transmission is important in hospital settings where there is a high flow of patients mixing with uninfected visitors and health care workers at any given time. Prior research on airborne and fomite deposition in an intensive care unit (ICU) demonstrated potential risk for medical staff and patients. Surface and air samples from an ICU and a COVID-19 general ward (GW) were examined at Huoshehshan hospital in Wuhan, China [21]. The study indicated that fomite deposition was higher on the ground compared to elevated surfaces. The researchers noted that the soles of shoes of medical staff carry the virus when they move, and they can spread it to other areas in the hospital. Furthermore, it was observed that environmental contamination was found to be higher in ICU settings, and the transmission of virus can extend to 4 m and above. Reduced airborne contamination was observed in wards with negative pressure controls, and increased airborne viral RNA contamination was observed in areas with a greater congregation of people, regardless of dedication to COVID-19 patient care [22,23]. Previous guidelines by ASHRAE detailed the importance of adequate ventilation inside hospital settings [3] to decrease disease transmission. Particular emphasis has been placed on regular air changes, or replacing stale air with fresh air, in order to reduce contamination. However, few studies to date have attempted to simulate the impact of airflow among compact settings [24,25]. Many factors dictate the air velocity, flow pattern, and direction of air movement inside a room [19,26,27], such as the following: ventilation system type and maintenance, supply/return vent location, air exchange rate (ACH), air filter type (filters inside HVAC unit or installed separately), position of air filters [28], and amount of humidity in the air [29,30]. The above-mentioned factors are shown in Figure 1. These factors, in turn, play a major role in the airborne transmission of aerosols, resulting in less potential for contamination in some areas and more in others.

Depending on the type of filtration, particles pass through several filters to remove particulates from the air. High-efficiency particulate air (HEPA) filters with higher minimum efficiency reporting value (MERV) ratings are mainly used in hospital intensive care units due to their high efficiency in filtering the particles. In addition, ultraviolet (UV) lights can be placed across filters to remove contaminants. An experimental study was carried out on the effectiveness of UV-C lights placed directly across HEPA filters in hospitals [31]. The virucidal effects of UV irradiation on SARS-CoV-2 with different doses of irradiation were studied [32,33], which can predict the epidemiological trends of COVID-19.

There have been significant advances in recent years in numerical models of airborne transmission and ventilation in buildings such as offices or residential areas using computational fluid dynamics (CFD). This CFD study was then extended to hospitals, and it reached a peak during the COVID-19 pandemic period. The review paper [34] showed the ventilation impact on particle spread and identified efficient ventilation strategies in controlling aerosol distribution in clinical and non-clinical environments. The study also illustrated the importance of occupant-based ventilation systems to reduce cross infections [34]. Validation studies on air ventilation and microbe removal rates have been carried out with CFD in hospital isolation wards. The study by Yam et al. [35] showed microbe removal rates using different numbers of ventilation systems inside a hospital ward. Their results showed that rearranging the air return location from the common general ward to the inner cubicle ward restricted the contaminant spread to a wider area.

Whilst many CFD studies have been carried out on ventilation and airborne transmission in buildings and hospitals, there are significant limitations to the CFD model type and conditions used in these studies. Most of the above-mentioned research used the Reynolds-averaged Navier–Stokes (RANS) equation for modeling aerosol transport in buildings [36]. There are approximations on the mechanisms responsible for transport in the RANS model, resulting in uncertainties, especially in forced convection scenarios where there is active mixing of the flow [37]. Furthermore, most of the simulations were carried out under uniform and steady state conditions for buildings.

In unsteady and transient conditions, the complex flow structures inside the room were not resolved fully with the RANS model. The detailed flow structure is important to accurately capture the particle propagation inside the room. On the contrary, aerosol transport simulation with the lattice Boltzmann method (LBM) proposed in this study is intrinsically transient and can effectively capture the aerosol propagation under non-uniform conditions. Moreover, a very large eddy simulation (VLES) was employed in the LBM model to explicitly capture relevant scales of turbulence rather than modeling them, thereby ensuring accurate modeling of the particle propagation.

To this day, only a few studies have been carried out with LBM to simulate aerosol propagation in buildings and hospitals. A joint venture study with the German institute research Institut für Energie- und Umwelttechnik e. V (IUTA) showed particle/contaminant propagation and risk assessment in closed spaces, including the effect of filtering devices with simulations using the commercial CFD software PowerFLOW^TM 6-2021. LBM with the VLES method was experimentally validated to improve the effectiveness of filtration devices in realistic office environments [38]. Such studies highlighted the importance of having an efficient HVAC system with appropriate sizing and settings to generate an airflow pattern that will help mitigate the risks of cross-contamination. Airflow, as well as the movement of numerous airborne particles produced by patient coughs, was simulated with the LBM approach in an ICU room under negative pressure. The results indicated that bed orientation and additional air treatment units increased the number of particles extracted from the air and decreased particles deposited on surfaces [39]. Another CFD study with the LBM approach [40] was carried out to analyze the flow path inside the entire floor of a Paris hospital with a water vapor mass fraction tracker. Furthermore, contamination risk between different spaces was analyzed.

Breakrooms can be a particularly hazardous setting, as they witness frequently large numbers of health care workers at any time. These occupants may talk, sneeze, or cough while eating, an activity that usually precludes masks. Since it is a closed room, depending on the airflow distribution, droplets or aerosols can be transmitted to other occupants. Furthermore, there is a higher chance of fomite deposit on tables, chairs, microwaves, refrigerators, and other equipment. Though aerosol spread outside the breakroom is minimized if the doors are closed, fomites can still be transported to the outside by the flow of occupants coming in and out of the breakroom. The size of the breakroom is usually much smaller compared to the larger overall hospital space, and the airflow inside the room heavily dictates the contaminant transmission.

Though some of the available literature shows contamination spread in indoor environments with CFD modeling, they provide very little information on the contaminant propagation and fomite deposition of the particles released from the multiple talking occupants seated closely with each other and the effect of different HVAC vent layouts on these particles inside the indoor environment. Furthermore, as mentioned above, these studies used the RANS model where the flow and particle propagation may not be accurately captured with forced convection. Hence, the present study solely dealt with the airflow distribution and contaminant transmission between multiple occupants seated closely and talking inside the enclosed emergency hospital breakroom using the lattice Boltzmann method combined with VLES modeling. Following a baseline simulation, iterations were carried out with variations in breakroom HVAC supply and return vent locations, reducing the number of occupants and determining the appropriate placement of UV lights, in an attempt to model interventions that could reduce transmission and cross-contamination. The applicability of this study is not limited to COVID-19, and it can be applied for other airborne infectious diseases as well.

2. Materials and Methods

2.1. Breakroom Layout

Based on actual blueprints, a detailed three-dimensional digital model of a hospital breakroom with a ventilation system was constructed. Included in the model were tables, chairs, refrigerators, and heat sources, such as a microwave and toaster. The dimensions of the room were 5.9 m × 2.7 m × 5.4 m. There were two supply and returns vents, as shown in Figure 2. The details of the vent baffles, including orientation, were also included in the simulation.

Ten health care workers were modeled in the current configuration, wherein one occupant was represented standing and the rest of the 9 occupants were seated, as shown in Figure 3. This scenario was considered to account for full occupancy in the breakroom. The contaminant transmission path from a standing human to other sitting humans was also taken into account. Occupant positions were static for the purpose of the simulation.

The room dimensions and the distance between the occupants are illustrated in Figure 4. The height between the occupant head and the nearest vent is 1.4 m.

2.2. Boundary Conditions

The flow rates for the supply and return vents were 165 and 170 cubic feet per minute (CFM), respectively, with no losses being considered. HVAC flow was modeled with 100% recirculation, wherein the mass flow rate of 335 CFM was provided at the supply inlet and removed at exhaust or return vent outlets. The supply and return flow rates were obtained from blueprints of the ventilation system. Mass flow rates and the distance between vents are shown in Figure 5.

The ambient room temperature inside the breakroom was 24 °C, and inlet air temperature was set to 20 °C. The breakroom had some heat-generating sources, such as refrigerators, microwaves, and toasters. Heat inputs of 100 W and 500 W were applied to refrigerators and microwave/toasters. Similarly, all human occupants were set to a body temperature of 37 °C. All occupants were simulated as continuously talking as a worst-case scenario. The talking rate of each human was set to 0.45 l/s. This boundary condition was based on work conducted by Gupta et al. [41] and Morawska et al. [42]. The mean size of the droplet particle was 2.8 μm, and the area of the mouth opening during speech was set to 1.9 cm². A Gaussian distribution was selected for modeling the particle size, with a mean of 2.8 μm and a standard deviation σ of 0.5 μm. This distribution was selected based on the methodology developed and used from previous works [38,42]. The size of the droplets ranged from 0.46 μm to 4.9 μm, covering both aerosols and droplets in the study. In order to predict the flow from the mouth accurately, fine resolution was added in the mouth region. The selected computational domain was subdivided using a non-conformal mesh of cubic elements into multiple variable resolutions (VRs), and an appropriate resolution was selected for each different region to accurately resolve the flow structures following the methodology described in [38], which showed good agreement with the physical experiment. The finest cell size used in this study was 2 mm.

2.3. Assumptions Used in Simulation

• A few assumptions were made in the modeling of this scenario:
• Heat input and temperature for all heat generation sources were assumed to be constant throughout the study;
• Masks were not worn by any of the occupants in the breakroom to consider a worst-case scenario;
• The area of the mouth opening and the mean size of the particle were the same for all the occupants;
• The emitter that releases the particles and the mouth inlet were modeled at the lip region just outside the mouth to model the flow and the particles from the mouth effectively;
• Breath temperature was set to body temperature (37 °C);
• Contaminants were released from the mouth at regular intervals during the simulation period;
• Evaporation of the droplets was not modeled in this study;
• Near-wall deposition from electrostatic forces was ignored.

2.4. Simulation Methodology

In this study, an inherently transient and highly accurate lattice Boltzmann method-based PowerFLOW solver was used to predict the airflow and particle distribution inside the building. The pressure and velocity field were solved using the lattice Boltzmann equation. Extensive literature can be found on LBM [43,44,45,46]. The LBM equation given below describes the time evolution of the distribution function $f=f\left(x,c,t\right)$ in a phase space:

$$\frac{\partial f}{\partial t}+c·\frac{\partial f}{\partial x}=\mathsf{\Omega }\left(f\right)$$

(1)

where $x$ is the position vector, c is the microscopic velocity vector, and $\mathsf{\Omega }$ is the collision operator. In the LBM, the number of degrees of freedom for the velocity space is reduced by restricting the potential particle motion to a finite number of directions. The Boltzmann equation can be expressed as a set of algebraic equations for probability distribution at each state, f_i [43,47]:

$$f_i\left(t+∆t,\overrightarrow{x}+{\widehat{c}}_i∆x\right)=f_i\left(t,\overrightarrow{x}\right)+{\mathsf{\Omega }}_i(t,\overrightarrow{x}) (i=0,1\dots ,\mathrm{M})$$

(2)

where f_i is the particle velocity distribution function defined for the finite set of discrete particle velocity vectors {${\widehat{c}}_i$: I = 0, 1 … M}, and M is the number of directions of the particle velocities at each node. The temperature equation is solved using an LBM scheme and is fully coupled with the mass conservation and momentum equations, incorporating buoyancy effects. For high Reynolds numbers, a very large eddy simulation [48] turbulence modeling was incorporated into the lattice Boltzmann-based solver with a turbulent relaxation time that models the effect of chaotic motion on the statistics of fluid particle collisions.

The PowerFLOW solver has an integrated Lagrangian particle simulator that includes splash [49], breakup [50], and re-entrainment [51] models. The trajectory of particle motion through the air is predicted based on the local drag force acting on the particles, the pressure gradient, and the gravitational force. The trajectory of the Lagrangian particles is predicted by the following equation [52,53]:

$$m\frac{D{\overrightarrow{u}}_{Particle}}{Dt}=\frac{1}{2}\rho C_{D}A\left({\overrightarrow{u}}_{Particle}-\overrightarrow{u}\right)\left|{\overrightarrow{u}}_{Particle}-\overrightarrow{u}\right|+m\overrightarrow{g}$$

(3)

where m is particle mass, ${\overrightarrow{u}}_{Particle}$ is particle velocity, C_D is particle drag, $\overrightarrow{g}$ is gravity, $\rho$ is the density of the particle, and A is the particle cross-sectional area. The particles are assumed to be spherical, and the drag coefficient is based on the Reynolds number and the Schiller–Naumann correlation [54].

$$Re=\frac{\rho d_p\left|{\overrightarrow{u}}_{Particle}-\overrightarrow{u}\right|}{{\mu }_g}$$

(4)

The coupling between the Lagrangian particle simulator with the flow solver in PowerFLOW tracks the trajectory of millions of particles accurately [38,39]. Figure 6 shows an image of droplets transported by cough airflow simulated using the described methodology. This image demonstrates how turbulent structures impact the mixing of contaminants. The turbulent structures and contaminant mixing of the talking occupants in this breakroom study were captured similarly.

For the purposes of this study, four simulations were performed.

• A baseline simulation was carried out showing airflow distribution and particle exposure from all the humans inside the breakroom with the original blueprint HVAC vent arrangement. Based on the results observed from the baseline simulation, three iterative simulations were then carried out to mitigate contaminant transmission inside the breakroom.
• The first iterative simulation had a modified vent arrangement, where the supply and return vents were placed diagonally to achieve uniform flow inside the breakroom.
• The second iteration was carried out with fewer, further-spaced occupants, with the intent of impacting exposure.
• Finally, the last iteration was carried out to identify the appropriate location for placing the UV lights in the breakroom. The aim of this simulation was to show that, in the current breakroom model, if the UV lights were positioned at the selected places predicted from the simulation, they should have higher exposure to all the particles (contaminated and non-contaminated), thereby efficiently killing the contaminated ones. Here, the process of UV light irradiation on the contaminants was not modeled, and only the higher contaminant regions for UV light placement were predicted. This was achieved by modeling screens in the simulation to capture the particles. In the simulation, initially, four different regions with higher particle concentrations were selected, and screens were modeled there. The velocity distribution was then analyzed in the area on the selected four screens. The screens with higher particle concentrations and lower velocities were then recommended for UV light placement. The selected screens were located at 0.6 m from the ceiling at four different locations.

2.5. Flow Setup

In order to replicate a real-life scenario, it is necessary to first simulate the airflow distribution inside the room. Usually, a flow simulation is carried out over a period of a few seconds to minutes depending on the size of the room and HVAC boundary conditions. It is necessary to have a converged flow inside the room before the particles are released from the occupants. This is because the airflow pattern, buoyancy effects from heat generating sources, and the local flow velocity from a human mouth influence the transmission of emitted particles. In the developing flow, all these effects evolve and change continuously, making it impossible to accurately capture the transmission of particles inside the room if released before the flow is stabilized. Hence, the flow simulation was first carried out to settle the flow inside the room.

2.6. Pathogen Transport Inside Breakroom

Once the flow was stabilized inside the room, the particles were released from all the talking humans at the start of the simulation. At around 60 s, the particle transmission inside the room was almost stabilized, and the particle modeling simulation was carried out for 60 s. Within the 60 s period, the particles mixed well and spread far into the breakroom. Some of the particles deposited on the ground, while others accumulated at the exhaust outlets. Contaminant particles from the talking humans were emitted at regular intervals during the 60 s simulation period. The flow inside the room for the particle modeling simulation was initialized from the previous converged flow simulation. This simulation allowed for the observation of particle transmission and contamination between breakroom occupants and fomite deposition during the 1-min transient physical time. Simulations were carried out for all occupants in the room, with variations in particle depositions attributed to proximity to air vents, proximity to other occupants, and room furniture. Though particle transmissions from all the occupants were simulated, the results are shown and discussed in detail only for the higher-transmitting occupants in the baseline and iterative simulations. These occupants had maximum exposure to the contaminant particles in the breakroom.

The percentage of contaminant exposure from one occupant to other occupants at a particular time was calculated based on the following equation:

$$\mathit{\%}{C}_{A\to B}=\left(\left(\frac{{NR}_{A\to B}}{{NE}_A-N{R}_{A\to A}}\right)\ast 100\right)$$

(5)

where NR_A→B is the number of particles received by Occupant B from A, NE_A is the total number of particles emitted by Occupant A, and NR_A→A is the number of self-emitting particles from Occupant A.

The maximum and minimum percentages of contaminant exposure are shown for occupants 2, 8, and 7 after 1 min in this study. These occupants, highlighted in Figure 3, are the major source of contaminant transmission as well as fomite deposition, and hence, they are described in detail. Similarly, the percentage of reduction in particles between the baseline and iterative simulation with a modified vent arrangement and reduced humans at a particular time was calculated as shown below:

$$\mathit{\%}{C}_I=\left(\left(\frac{NB-NI}{NB}\right)\ast 100\right)$$

(6)

where NB is the number of particles in the baseline simulation, and NI is the number of particles in the iterative simulation inside the breakroom. The percentage reduction in contaminants in the breakroom between the different cases is shown after 1 min in this study. Isotropic Cartesian mesh was used for the model. The grid resolution scheme and particle modeling method applied in this breakroom simulation were based on the methodology developed and validated with other hospital and office simulations [38].

3. Results

3.1. Flow Distribution Inside the Breakroom—Baseline Model

For the breakroom model, the flow was converged in a minute, and the flow simulation was carried out for 2 min. The average flow inside the breakroom was found to be approximately 0.15 m/s. Figure 7a shows the position of the velocity slide, Figure 7b shows the average flow distribution along the slice for the last 1 min, and Figure 7c shows the effect of buoyancy and thermal plumes rising from humans and other heat sources. The rear side of the room, away from the entrance where two supply inlets are located, received a higher flow than the front side of the room near the kitchen area. Less airflow distribution was seen in the center of the room and the area near the microwave.

3.2. Contaminant Transmission Inside Breakroom—Baseline Model

The simulation methodology noted above allowed for the quantification of contaminant transmission and exposure experienced by each occupant. This is demonstrated in Figure 8. In Figure 8a, Occupants R1–R10 represent the amount of contaminants received from the other occupants, and Occupants E1–E10 represent the amount of contaminants emitted by each occupant. Occupants 9 and 4 were the most highly exposed; 34% of all the contaminants transmitted were received by Occupant 9, and 12% were received by Occupant 4. Of the 34%, Occupant 9 sitting in the center between Occupant 2 and 1 received the highest transmission of contaminants (66%) from neighboring Occupant 2. Similarly the second-highest receiver, Occupant 4, sitting in the corner near the fridge, received the maximum contaminants (71%) from Occupant 8, who was seated across the table. Conversely, Occupants 10 and 5 received the least amount of particles; Occupant 5 received 5% and Occupant 10 received just 3% of the total contaminants.

Overall exposure for all occupants in the baseline simulation was demonstrated in Figure 9. The different colors in the legend represent the color of the particles emitted by each occupant. For example, green-colored particles are shown for Occupant 3. We noticed three main different areas within the breakroom:

-

A cluster of particle where cross-contamination risks from airborne particles were important;

-

A region where the particles were directly carried toward the main table with high risk of infecting the neighboring surfaces;

-

A low-contamination zone populated by very few particles.

3.3. Particle Transmission from Select Occupants

Particle transmissions from sample occupants 2, 7, and 8 are shown in Figure 10, Figure 11 and Figure 12, respectively. Contaminant exposure was noted to be asymmetric for adjacent occupants. As illustrated in Figure 10, Occupant 9 received ~66% of the emitted particles from Occupant 2, followed by Occupant 1 who received ~16%. Occupant 9 also received a disproportionate amount of particles from Occupant 2. Exposure patterns of varying similarity were noted for Occupants 7 and 8. Occupant 10, standing on the front side, received the least amount of contaminants.

Table 1. The percentage of contaminants received by other occupants from Occupant 2.

Occupants	%Contaminant Exposure from Occupant 2 to Other Occupants
Occupant 1	16
Occupant 3	4.7
Occupant 4	2
Occupant 5	0.5
Occupant 6	2.0
Occupant 7	3.4
Occupant 8	5.1
Occupant 9	66
Occupant 10	0.1

Red color represents maximum contaminant received and blue color represents least contaminant received.

shows the percentage of contaminants received by all the occupants from Occupant 2. Since most of the contaminants emitted by Occupant 2 were transmitted to the neighboring occupants, a negligible amount of contaminants reached Occupants 10 and 5 seated at the front end of the breakroom. All the other occupants received lesser particles compared to Occupants 9 and 1.

Occupants 4 and 8 were seated in the table directly below the inlet supply vent, as shown in Figure 11. Due to the higher air circulation in this area, most of the contaminants emitted from Occupant 8 reached Occupant 4 (~71.3%), who was facing the occupant.

Occupants 1, 5, and 10 across the breakroom received the least amount of contaminants, as shown in

Table 2. The percentage of contaminants received by other occupants from Occupant 8.

Occupants	%Contaminant Exposure from Occupant 8 to Other Occupants
Occupant 1	0.7
Occupant 2	5.2
Occupant 3	10.7
Occupant 4	71.3
Occupant 5	0.5
Occupant 6	1.3
Occupant 7	6.4
Occupant 9	3.7
Occupant 10	0.2

Red color represents maximum contaminant received and blue color represents least contaminant received.

. Some small amount of cross-contamination was seen between the right and left side of the breakroom with the contaminants released from Occupant 8.

Lower particle transmission to other occupants and more fomite deposition on the neighboring surfaces were seen for the particles emitted from Occupant 7, as shown in Figure 12. It was observed that ~50% of the particles emitted from Occupant 7 reached Occupant 5, sitting at the front, followed by occupant 6, who received ~23%. The rest of the occupants received less than 5% of the particles emitted from Occupant 7.

No cross-contamination was observed between the right and left side of the room. Occupants 4 and 8 seated on the left side did not receive any contaminants from Occupant 7, as shown in

Table 3. The percentage of contaminants received by other occupants from Occupant 7.

Occupants	%Contaminant Exposure from Occupant 7 to Other Occupants
Occupant 1	4.6
Occupant 2	1.2
Occupant 3	3.7
Occupant 4	0.0
Occupant 5	50
Occupant 6	23.6
Occupant 8	0.0
Occupant 9	3.7
Occupant 10	0.2

Red color represents maximum contaminant received and blue color represents least contaminant received.

. This is due to the impact of airflow around Occupant 7 that moved the contaminants from the rear side to the front side without much transmission to the sides.

3.4. Fomite Deposition in Baseline Model

The fomite deposition from droplets emitted by all the occupants is shown in Figure 13. The color of the hit points represent the fomites deposited by different occupants and is shown in the legend. The highest deposit was seen on the breakroom walls, followed by the tables. The side table below the TV received the maximum fomite deposit (shown in purple color points in Figure 13) from Occupant 7, who was seated closely to the side table. The depositions from other occupants sitting on the front area of the breakroom were localized and negligible.

3.5. Flow and Contaminant Transmission in Iterative Model with Modified Vent Arrangement

In the first iterative simulation, supply and return vent locations were re-arranged in such a way that the inlets and outlets were placed diagonally opposite to each other (Figure 14a). A more uniform flow was seen inside the room with the modified vent arrangement (Figure 14b). Unlike the baseline simulation, wherein a higher flow was seen on the rear side where both supply vents were placed, in the modified arrangement, the airflow from the rear right and front left supply vents mixed well in the room and exited through the rear left and front right exhaust vents. This resulted in a more uniform flow distribution, thereby reducing the higher local contamination and flushing out the contaminants through exhaust vents.

Compared to the baseline vent arrangement, overall, less contaminant transmission was seen with modified vent locations (Figure 15). The most exposed occupant appeared to be Occupant 3, with the lowest exposure noted for Occupant 5. Occupant 3, sitting across from Occupant 7, received 28% followed by Occupant 4, sitting across from Occupant 8, who received 16% of the total contaminants. In this modified arrangement, the amount of contaminants received by each occupant was reduced compared to the baseline simulation. Occupants sitting across from each other, such as 3, 7 and 4, 8, were more susceptible to contamination than adjacent occupants. Another reason for reduced contamination is the particles being removed well by the HVAC exhaust vents in this modified arrangement.

The recipient of the majority of Occupant 7′s contaminants was Occupant 3 placed across Occupant 7, at 87.7% (Figure 16).

The particles emitted from Occupant 7 were contained locally around Occupants 3 and 6, thereby reducing cross-contamination across the room. Further, the rear left exhaust vents placed above 8 and 4 flushed out many particles, reducing contaminant transmission.

Table 4. The percentage of contaminants received by other occupants from Occupant 7 in the revised vent arrangement simulation.

Occupants	%Contaminant Exposure from Occupant 7 to Other Occupants
Occupant 1	0.1
Occupant 2	2.5
Occupant 3	87.7
Occupant 4	0.9
Occupant 5	0.2
Occupant 6	5.2
Occupant 8	0.9
Occupant 9	2.1
Occupant 10	0.3

Red color represents maximum contaminant received and blue color represents least contaminant received.

shows a negligible amount of contaminants received by Occupants 1, 4, 5, 8, and 10 seated away from Occupant 7.

3.6. Reduced Occupants in Breakroom

Reducing the number of occupants and seating them separated with sufficient distance significantly reduced the amount of particles received, as shown in Figure 17. Occupants 5, 8, and 9 were removed from the baseline case to provide greater distance between the adjacent occupants. Figure 18 demonstrates the areas of particle exposure assessment for this iterative simulation. Removing Occupants 8 and 9 decreased the contaminant transmission to Occupant 4 significantly, and they received the least amount of particles. Occupants 3 and 7 were cross-contaminating each other. Contaminant transmission to Occupant 10 was reduced heavily by removing the adjacent-seated Occupant 5.

The total number of contaminants received for the baseline and iterative simulations with modified vent arrangements and reduced humans is shown in Figure 19. The baseline simulation with the original HVAC vent layout of two supply vents in the rear and two exhaust vents in the front had a higher contaminant concentration inside the breakroom. The contaminants received by the occupants were reduced by 26% in the modified HVAC vent layout, wherein one supply vent was moved to the front and one exhaust vent was moved to the rear area of the breakroom, where there were many occupants. Due to the mixed HVAC layout in the modified vent arrangement, there was a more uniform flow distribution inside the breakroom, and more particles were removed from the breakroom through the exhaust outlet. In the baseline simulation, only 4357 particles were removed, whereas in the modified vent arrangement, around 9974 particles were removed from the breakroom by the exhaust outlets. The exhaust vent that was moved to the rear area in the modified vent arrangement alone removed 7081 particles out of the 9974 particles from the breakroom. Additionally, reducing the occupants and increasing the distance between the occupants reduced the contaminants received by other occupants by 70%. It should be also noted that the particles emitted inside the breakroom was also less due to the reduced number of occupants.

3.7. Contamination and Potential Mitigation by UV Lights

The amount of particles deposited on the screens during the 1 min period was measured. The screens recorded the number of particles deposited at each time. The screen opening size was set to 1 m, and the pass through fraction was set to 1. These settings allowed all the particles reaching the screen to pass through and deposit on the screen. The number of particles or contaminants deposited at each screen was recorded and is shown in Figure 20. Particle concentration was the highest in S4 and S1, followed by S2 (Figure 21). Mean particle velocity for the four screen areas are noted in Figure 21 as well.

4. Discussion

4.1. Analysis in Baseline Model

4.1.1. Airflow Distribution

The blue print of the breakroom model showed that the diffuser blades are curved and aligned with the ceiling, thereby causing the air flow to spread towards the sides of the ceiling before falling down. This phenomenon can be seen in Figure 7, where the air from the supply inlets traveled along the sides of the vents, fell on the tables/ground on either sides, and then rose up. This type of flow distribution can lead to a Coanda effect [55], wherein the airflow leaving the diffuser clings to the ceiling near the HVAC vents due to the low pressure created between the diffuser and ceiling. With time, the accumulation of dirt particles in the HVAC system and ceiling can lead to this type of flow pattern. Hence, it is necessary to clean the HVAC system and room at regular intervals. The current layout has more people seated on the rear side, where there is a higher flow circulation.

4.1.2. Cross-Contamination between Occupants

The importance of having enough space between occupants is shown in Figure 8. Occupant 9, sitting in the center, received the highest amount of contaminants from adjacent occupants as well as from behind. The second highest receiver was Occupant 4, who received mainly from Occupant 8. Occupants 10 and 5, who were sitting in the low flow regions on the front side of the room, received less contaminants than other occupants. The right and left sides of the room seemed well isolated from each other. On a regular day with full capacity, there is a higher chance of cross-contamination between occupants who are seated in close proximity. Occupants 9 and 4 were seated in the least favorable positions since they received the highest exposure from all the adjacent occupants.

Contaminant transmission appeared to be mainly influenced by the flow from the HVAC system, the number of occupants present, and the spacing between the occupants. The airflow direction was from the rear side of the room to the front side, as shown by the arrow direction in Figure 10. Therefore, the contaminants emitted by the occupants in the rear side of the breakroom were carried by the flow towards the front side of the room infecting the neighboring occupants seated along the path. This effect could be seen for the particles emitted by Occupant 2. Most of the particles emitted from Occupant 2 were first exposed to Occupant 9, who was in close proximity. The second-highest exposure was for Occupant 1, seated next to Occupant 9. Due to higher flow mixing around the center table region, most of the particles were concentrated in this region around Occupants 9, 1, and 6. Occupant 10, standing in the front side of the breakroom, received the least amount of particles emitted from Occupant 2. Other occupants sitting in both the corners received some amount of contaminants from Occupant 2.

This phenomenon was again noted for Occupants 4 and 8. Higher airflow mixing in the corner region where they were seated led to higher cross-contamination between the two, as shown in Figure 11. Furthermore, some of the contaminants were blocked by the refrigerators placed behind Occupant 4, resulting in higher transmission.

4.1.3. Overall Exposure from All Occupants

The high and low areas of exposure for the baseline model with all occupants are shown in Figure 9. Particle exposure was higher in the region under the red box because there was a higher number of occupants sitting in close proximity to each other, and all the occupants were talking in this study. Further, there was high airflow circulation in that area, resulting in a higher chance of cross-contamination between the neighboring occupants. The next zone of contamination is shown in the green box, which has increased fomite deposition. For the baseline vent model, it was not recommended to have anyone sitting in the Occupant 7 position because the particles emitted from the Occupant 7 position contaminated the neighboring objects, as well as other people coming to that area. This can be observed by the white dots in the side table area in Figure 9. Lower exposure was seen under the box area in the front low flow region near the kitchen.

4.1.4. Fomite Deposition

The lighter aerosols were carried by the airflow and traveled inside the room for some period of time. The heavier particles fell and deposited on the ground or other objects. These objects can become fomites when the deposited particle is infectious. Contamination occurs by infected aerosols when humans breathe or inhale particles through the nose and mouth, whereas contamination by fomite occurs by touching the fomites and placing the contaminated hand on the eyes, nose, and mouth. The highest fomite deposition was seen on the walls followed by the tables, as seen in Figure 13. The walls in the simulation were comprised of the sidewalls, floor, and ceiling of the breakroom. More fomite deposition is seen on the ceiling. This is due to the contaminants from Occupants 2, 3, 4, and 8. In the baseline simulation, the two supply vents were placed in the rear area above where Occupants 2, 3, 4, and 8 were placed. The air from the supply vents hit the table, rose above, and spread to the sides, as shown in Figure 7. Due to this, some of the heavier particles settled on the ceiling, as shown in Figure 9. The highest fomite deposition on the ceiling was from Occupant 4. The next-highest fomite deposition was on the tables, mainly the side tables near Occupant 7. Compared to particle exposure on the seated occupants, more fomite deposition on the neighboring surfaces was seen from the particles emitted from Occupant 7. The main reasons for this are the air flow path and no occupants along the contaminant travel path. The airflow from the supply vent located above Occupants 7 and 3 carried the contaminants toward the front side of the room mostly in a straight direction, as shown in Figure 13. Since there were no occupants sitting or standing in the flow path, fewer contaminants were transmitted to them. However, the particles were deposited on the nearby table, television, and other objects along the side.

4.2. Airflow and Contaminant Propagation Analysis in Iterative Model with Modified Vent Position

In the iterative simulation with the updated vent arrangement, contaminant spread between the occupants was uniform and not highly concentrated from one single occupant, unlike the baseline. The flow pattern inside the room had significant influence on particle distribution as observed between different occupants. In the baseline case, the flow from the vents directly spread from the rear side of the room to the kitchen area, whereas more mixing of the flow was seen inside the room in this new iteration. In the baseline simulation, Occupant 5 transmitted 80% of all their emitted particles to the neighboring Occupant 10, standing slightly behind, whereas in the iterative run, Occupant 10 received only 10% from Occupant 5. The mixed flow inside the room spread the contaminants uniformly, thus contaminating neighboring occupants such as 1, 6, and 7 with almost equal concentration. Occupants 5 and 7 received the least contamination from neighboring occupants. Very little cross-contamination was seen between the left and right side of the room even in the revised vent arrangement. The contaminant concentration in the breakroom was reduced by ~26% with the revised vent arrangement. In this arrangement, a higher concentration was seen between the occupants seated across from each other and directly under the flow. Wearing a mask or removing the occupant from the opposite locations would further reduce contamination. The contaminants received by all the occupants, as well as the most and least infected occupants, can be seen in Figure 15.

4.3. Overall Exposure from All Occupants in Iterative Model with Reduced Humans

As mentioned in the results section, with reduced occupants and more spacing, the overall transmission between the occupants decreased considerably. Two occupants, namely Occupants 4 and 9, were removed from the area under red box shown in Figure 18. The increased distance between the occupants in the red zone reduced the particle concentration and exposure in this iterative simulation. Furthermore, removing the two occupants in the area under the red box also reduced the particle emission, thereby reducing the contaminant transmission between Occupants 1, 2, and 4. The green zone showed slightly reduced contaminants compared to the baseline after removing Occupant 5. However, since there was not much cross-contamination between the right and left side of the room, there was not a significant change in the contaminant transmission in the green zone compared to the baseline case with all occupants. The front portion near the kitchen region with reduced airflow and one occupant removed received the lowest amount of contaminants. Figure 17 shows the risk-dominant areas of the breakroom.

4.4. Contaminant Load for Different Cases

The baseline model and the iterative model with modified vent arrangements had the same number of occupants. However, the contaminant load in the breakroom was less for the modified vent case compared to the baseline (Figure 19). In the baseline model, both the return vents were in the front part of the room (Figure 7), whereas in the modified vent arrangement, one return vent was in the front part, and another one was in the rear part of the breakroom where the occupants were closely spaced (Figure 14). Although the same amount of particles was released from all the occupants in both the simulations, more particles reached the outlet or return vents in the modified vent arrangement due to its closer proximity and location in the high concentration zone. In the baseline simulation, there were two supply inlets in the high concentration zone that pushed the particles towards the return vents in the front part of the breakroom, but not many particles reached the outlet. Due to this, the particles stayed in the breakroom for a longer time and were exposed to all the occupants. This is because the baseline model, with all humans present, had more particles compared to the iteration model, with modified vent arrangements and all humans. Hence, it is important to have an efficient HVAC design to flush out the contaminants from the room as quickly as possible. For the iterative case, with reduced humans, the contaminant load was less due to the reduced number of people in the breakroom.

4.5. Placement of UV Lights

In buildings, it is necessary to pre-determine where the UV lights need to be installed in order to efficiently remove the contaminated particles. For a new building or an existing building, the area where the particles are highly concentrated can be shown quickly though the simulation. Here, in this study, the order of the different contaminant concentration zones was simulated by modeling multiple screens in the high concentration areas. As mentioned in Section 4.1.4, it was observed that in the baseline simulation, the lighter aerosol particles emitted by Occupants 3, 4, and 8 rose above and hovered near the ceiling and needed to be cleared. Hence, the screens were placed in these high concentration areas. In Figure 21, it can be seen that particle concentration is maximum in S4, followed by S1 and S2. The velocity in the area around S4 and S2 was low, thereby allowing the particles to stay and accumulate there for a longer time and increasing the exposure to the occupants. Particle concentration was lower around S3, due to higher flow, which enabled the particles to travel and move away from that region. The installation of UV lights in the high concentration zones (S4 and S2) could be recommended because more particles were concentrated and velocity was low in that area, thus making the particles stay for a longer time; thereby, they can be removed or mitigated with UV lights. The selected filter positions are depicted as white boxes in Figure 21c.

4.6. Observations from All Simulations and Future Work

Based on the above results, some important considerations must be taken in to account. Specific seating arrangements can be avoided or optimized. Care must be taken to avoid high-exposure areas for food placement. Thoughtful and efficient ventilation system layouts can be mapped for more efficient room design, and certain high-exposure areas can be targeted for UV light placement to inactivate viral particles and improve source control.

This study design, as well as the general assumptions used in the simulation (see Section 2.3), offers a few limitations that impede the generalizability of results but are inherent to simulation-based methodology. Fomite deposition and airflow patterns can only act as surrogate endpoints for transmissibility, and they may not correlate completely with the transmission of infection between individuals, given the many more variables involved in infection transmission. Finally, there might be some limitations regarding how the results of a simulation on a closed system with a limited number of occupants can be extrapolated to larger area with a more dynamic movement of individuals and other objects.

The evaporation of droplets affects the size and the length of stay of the droplets, depending on the humidity in the given space. Future work will include the evaporation of droplets and effects of solar radiation in different weather conditions.

5. Conclusions

In this paper, we showed that the airborne contaminant spread was more in the rear breakroom area with high airflow and multiple occupants seated closely. Based on the current study, it is recommended to have exhaust vent outlets in the area where multiple occupants are seated in order to flush out the contaminants quickly from the room. The baseline simulation had no exhaust outlets in the rear breakroom area where more occupants were seated. This resulted in a higher contaminant concentration and spread compared to the modified HVAC arrangement, wherein one exhaust vent was moved to the rear breakroom area from the front. A 26% reduction in the contaminants received by other occupants was observed in the modified configuration. In addition to this, the fomite deposition was also less and seen mainly near occupants due to more uniform flow inside the breakroom in the modified HVAC configuration. Reducing occupants and maintaining enough distance reduced contaminant transmission. A 70% reduction in contaminants was observed with reduced humans, though it should be noted that emission was also reduced compared to the baseline due to less occupants. Furthermore, the simulation results can be used to identify the optimum placement of UV lights. Hence, from the current study, we have shown that having enough distance between the occupants and appropriate positioning of the exhaust outlet can significantly reduce contaminant transmission. To further clear the remaining areas with a high concentration of contaminants, we predicted the appropriate location for UV light placement. This observation in hospitals can be extended to other buildings, such as offices, factories, restaurants, and shops, wherein multiple occupants will be present at a given time and location.