Prediction of Air Purifier Effectiveness for Eliminating Exhaled Droplets in a Confined Room

Abstract

High-efficiency particulate air (HEPA) filter purifiers are a recommended method for eliminating respiratory airborne droplets. In this study, the movement of airborne droplets exhaled by occupants in an unventilated, two-bed dormitory room with an air purifier was simulated using computational fluid dynamics. The air was modeled using an Eulerian scheme while the droplets were modeled using a Lagrangian method. The airborne droplet number, the rate at which droplets are removed, and the rate at which droplets accumulate were calculated. A larger HEPA flow rate increased the droplet removal efficiency, with most of the droplets settling on boundary surfaces. Of particular note, the air purifier location within the room had a significant impact on reducing the droplet exchange between two occupants and improving the droplet elimination efficiency.

1. Introduction

The global outbreak of the coronavirus disease (COVID-19) has caused over 3 million deaths as of this writing. The disease is caused by SARS-CoV-2 RNA, which is mainly transmitted by direct/indirect contact with droplets or the inhalation of aerosols. An infected individual can release respiratory fluid containing SARS-CoV-2 by breathing, vocalizing, coughing, and sneezing [1,2,3,4,5,6,7,8,9]. Since respiratory fluids can range in aerosol size from 0.2 to 60 µm and to larger droplet sizes of up to 100 µm, the virus (0.1 µm) can stay in the air for many minutes until settling on the ground [10,11,12,13]. If a droplet/aerosol particle of 5 µm is left to fall freely in a closed environment, where typical airflow can range from 5 to 20 cm/s, the particle can travel from 100 to 400 m until reaching the ground [14]. Furthermore, a person merely walking into a room can resuspend aerosol particles up to 2 m from the ground [15]. Clearly, the trajectories and settling dynamics of respiratory particles in an enclosed room with occupants can be complex. Although a combination of personal hygiene, wearing facial masks, and social distancing are critical to reducing the risk of exposure to infectious particles, often individuals share confined spaces by necessity. Ventilation or air purifying systems can then be used as a control strategy where low-velocity exhaled airflows can be removed or dispersed quickly away from individuals. For example, [16] found that over 80% of droplets can be removed by the ventilation system in an aircraft cabin after three minutes. The ventilation system can minimize the concentration of the droplets in an enclosed environment; however, it can also disperse the droplets to nearby people [16,17].

Zhao et al. [18] found that HEPA filters can remove 83% of the aerosols that may carry SARS-CoV-2. Although the Centers for Disease Control and Prevention (CDC) has not provided recommendations for the use of portable air purifiers for SARS-CoV-2, the organization did suggest the use of portable HEPA purifiers as an adjunctive infection control strategy for SARS-CoV-1 in 2005 [19]. Although portable HEPA air purifiers can filter droplets from the air, the fan used to produce airflow through the purifier could suspend and disperse droplets in the air, potentially spreading the disease-carrying droplets [20,21]. Indeed, there have been concerns that portable purifiers could increase the spread of exhaled aerosols and worsen the situation [20,22,23].

Computational fluid dynamics (CFD) can be an effective tool for studying droplet movement in an enclosed environment with a portable air purifier since the model produces high-resolution space and time-variable wind fields in the confined space that automatically satisfy mass continuity constraints for the air. For example, the movement of exhaled droplets has been successfully predicted using a Lagrangian particle-tracking method by many researchers [24,25,26]. In this paper, we consider a case study of the exhaled droplet movements from two persons sleeping in an unventilated double occupancy college dormitory room containing a HEPA air purifier. The movement of the exhaled droplets from the two persons is investigated for different HEPA filter air flow rates and filter placement locations. The droplets exhaled from each person were tracked using the Lagrangian method in a CFD-DPM (discrete phase method) model. The rate at which droplets were filtered through the air purifier, surface deposition, and droplets entering the region surrounding each person’s head was predicted. From these data, recommendations are made for using a portable HEPA filter in enclosed spaces.

2. Materials and Methods

2.1. Case Description

An older-style, double-occupancy dormitory room was modeled (Purdue university residences: https://www.housing.purdue.edu/Housing/Residences/HonorsCollegeandResidences/layout.html, accessed on 12 May 2023), as shown in Figure 1. In these rooms, which are still commonly in use (at the authors’ institution, for example), there is no building ventilation system to provide conditioned air in the room. There is no air conditioning and heat is provided via an in-room radiator. It is not uncommon in these rooms to have the windows closed and, thus, the only possible filtration of airborne droplets in these conditions is via a portable air filter. The room dimensions are 11 ft (3.35 m) wide by 18 ft (5.49 m) long by 8 ft (2.44 m) tall. Located on either side of the room are a lofted full bed and a non-lofted full bed. A desk was positioned at the end of each bed, a couch was located under the lofted bed, and a nightstand was located next to the non-lofted bed. Between the two beds was a narrow walkway. This setup is typical of many U.S. college dormitory rooms. In the current study, it was assumed that one person was sleeping in each bed with each person’s head located at the end of the bed nearest the wall. The outlines of the individuals were not included in the room geometry.

The room also included a typical portable HEPA filter. Two locations were considered for this filter. In one case the filter was located in the walkway between the beds, next to the non-lofted bed nightstand and the wall (Position 1). This position would be a reasonable one for an air filter since it would not present a tripping hazard. The second location considered was between the two beds, but near the center of the room (Position 2). In this position, the air filter was centrally located but could pose a hazard if a person was to leave their bed in the middle of the night.

A 14 in. (0.36 m) tall, 10 in. (0.25 m) diameter cylindrical tower air purifier was modeled after considering the most widely sold air purifiers for a 200 ft² (18.58 m²) space on a widely-used online shopping website. Most of these purifiers have volumetric flow rates greater than 135 cfm (0.0637 m³/s). Based on the dimensions and settings of these air purifiers, it was determined that the corresponding air speeds to be modeled would be 1.0 (3.3), 1.5 (4.9), and 2.0 (6.6) m/s (ft/s). Note that there is no HVAC-induced airflow in the model since these older-style rooms do not have conditioned air.

The typical size of the droplets exhaled during closed-mouth breathing is approximately 0.3–0.5 μm [27,28]. A monodisperse aerodynamic diameter of 0.4 μm was used in this study. The corresponding release rate is 525 droplets per breath, with each breath occurring every 4–6 s [16]. Thus, a droplet release rate of 105 droplets per second was used in this study. An exhalation speed of 3 m/s was assumed to correspond to the peak breathing speed during exhalation [16]. The density of each droplet was assumed to be the same as the density of water [29]. Furthermore, since the size of a person’s nostrils is small compared to the size of the room, it was assumed that the drops were exhaled from a point source, with a release angle directed 30° upward from the horizontal [28,30]. The droplets were released from each room occupant once the air velocity field in the room reached a steady-state (approximately 90 s after the start of the simulation). A summary of the simulation parameters is shown in

Table 1. Summary of simulation parameters (1 atm, 20 °C).

Particle Property/Parameter	Symbol	Value
Droplet density (kg/m3)	${\rho }_p$	998.2
Droplet diameter (µm) [31]	$d_P$	0.4
Exhaled droplet mass flow rate (kg/s)	$\dot{m}$	2.2−14
Exhaled droplet initial velocity (m/s) [31]	$u_P$	3
Gravitational acceleration (m/s2)	$g$	9.81
Air dynamic viscosity (Pa·s)	$\mu$	1.82 × 10−5
Air density (kg/m3)	${\rho }_f$	1.20
Turbulent intensity		5%
Turbulent viscosity ratio		10
Time step (s)	Δt	0.001
Number of iterations per time step		40

.

2.2. CFD-DPM Model Development

CFD-DPM modeling, adopting Eulerian and Lagrangian approaches for the fluid and particle (droplet) phases, respectively, was undertaken using ANSYS FLUENT (Ansys Inc., Canonsburg, PA, USA, latest v. R1). In other words, the air flow field was tracked at predefined positions during the simulation, whereas droplets were tracked by focusing on their absolute positions throughout the period of simulation.

2.2.1. Fluid Turbulence Model and Eulerian Mesh Characteristics

To include the effect of fluid turbulence on droplet dispersion, the airflow inside the chamber was simulated by solving the Navier-Stokes (RANS) equation with a standard k-ϵ turbulence model. The RANS model offers a good compromise between accuracy and computational efficiency and is widely used for most engineering applications [32]. The standard k-ϵ turbulence model has shown good performance in predicting the airflow in indoor environments [17,33].

The air at room temperature with standard atmospheric pressure was used. The turbulence intensity was set at 5% and the turbulent viscosity ratio was 10 since these are the values recommended by ANSYS for general simulation conditions [34]. The air filter inlet was set as volumetric (for the room environment) in the CFD model (Figure 1b), with a volumetric flow rate of 50, 100, and 150 cfm, corresponding to air purifier speeds of 1.0, 1.5, and 2.0 m/s (

Table 2. Air purifier variable parameters.

Outlet Velocity (V, m/s)	Inlet Flow Rate (kg/s)
2.0	0.1
1.5	0.0847
1.0	0.0566
0	0

).

An Eulerian grid sensitivity analysis with grid sizes of 0.05 m and 0.1 m was performed. The total number of cells was 774,355 and 153,707, respectively, with the mesh average skewness being 0.057 and a maximum of 0.81. Sensitivity tests were performed on coarse and fine meshes. A total of 14 points were sampled along the x, y, and z axes from the center of the room to compare the air speeds, as shown in Figure 2. The air speed differences using the two different meshes are within 5% of each other. The DPM tracks the droplet trajectories and, thus, the effect of the mesh size on droplet tracking is limited [17]. As the differences were small, the coarser grid was used for subsequent analyses. The DPM prediction is more sensitive to the time step and, thus, a value of 0.001 s was used so that the maximum distance a particle would move in a time step would not exceed the mesh size.

The air speeds at the same 14 nodes shown in Figure 2 were compared at simulation times of 85 and 90 s. The speed change at each of the nodes was less than or equal to 0.03 m/s (≤3% of the smallest air purifier outlet speed); thus, the airflow field was considered to be at a steady state at 90 s. The droplets were released only after the airflow was at a steady state. For simplicity, the remainder of this article refers to the time after the droplets were released and the airflow was at a steady state.

2.2.2. Droplet Tracking Model

A discrete phase model (DPM) was used to model the droplet movements. A droplet’s acceleration was determined from the following force balance,

$$-{\displaystyle \frac{\mathrm{d}{u}_p}{\mathrm{d}t}}=f_D\left(u-u_p\right)+{\displaystyle \frac{g\left({\rho }_p-\rho \right)}{{\rho }_p}}+f_B$$

(1)

where $u_p$ is a droplet’s velocity, $t$ is time, $f_D\left(u-u_p\right)$ is the drag force per unit mass acting on the droplet due to the local air velocity $u$, $g$ is the acceleration due to gravity, which acts vertically downward, ${\rho }_p$ is the droplet density, $\rho$ is the air density, and $f_B$ is a randomly-oriented, Brownian acceleration since the droplets in this study are smaller than 1 µm. A Stokes drag force is calculated using a particle’s Reynolds number,

$$f_D={\displaystyle \frac{18\mu }{{\rho }_p{d}_p^2{C}_D}}$$

(2)

The parameter $C_D$ is the drag coefficient,

$$C_D=1+{\displaystyle \frac{2\mathsf{\lambda }}{d_p^2}}(1.257+{0.4e}^{-({\displaystyle \frac{{1.1d}_p}{2\mathsf{\lambda }}})})$$

(3)

where $a_1$, $a_2$, and $a_3$ are constants determined from the Reynolds number and are provided in Chen (1995) [17].

Since the droplets are sub-micron in size, the Brownian motion is included [35]. The Brownian motion is the random movement of the micron size particles in the air due to molecular impacts. The components of the Brownian force are modeled as a Gaussian white noise process with spectral intensity (ANSYS Fluent Theory Guide, Release 15.0),

$$S_0={\displaystyle \frac{216\mathsf{\upsilon }{k}_{B}T}{{\mathsf{\pi }}^2\rho d_p^5{\left(\frac{{\rho }_p}{\rho }\right)}^2{C}_d}}$$

(4)

where $T$ is the absolute temperature of the air, $\mathsf{\upsilon }$ is the kinematic viscosity, and $k_B$ is the Boltzmann constant with a value of 1.380649 × 10⁻²³ J/K. The Brownian force magnitude is,

$$f_B=\zeta \sqrt{{\displaystyle \frac{\pi S_0}{\mathsf{\Delta }t}}}$$

(5)

where $\zeta$ is a Gaussian random number from a standard normal distribution between −1 to 1, and Δt is the integration time step.

Environmental droplet removal mechanisms include aerosol coagulation, gravitational settling, evaporation, and precipitation scavenging [36,37]. The coagulation and scavenging of smaller droplets by larger droplets are more significant when the droplet concentration is large [38], which is not the case here. Droplet evaporation can significantly impact the time that droplets remain suspended in the air since smaller droplets have a smaller terminal speed. The present work does not consider this factor and, thus, represents a higher humidity scenario in which the droplets are less likely to be suspended in the air since their size remains unchanged after exhalation. Thus, the present study is a ‘limiting’ scenario in which droplets remain suspended for less time than might typically be expected.

The current model assumes that the droplets which have contacted room surfaces, e.g., the walls, beds, or desks, or have passed through the air purifier, are removed from the simulation. The dynamics of sub-micron droplet deposition onto surfaces is complex. In general, the droplet deposition rate increases as the droplet size increases [39], air speed decreases [40], and surface roughness increases [41]. As is shown in the following section, the predicted air speed adjacent to the boundaries is in the order of 1 cm/s or smaller and much of the surface area in the dormitory room consists of fabric and wood, which are considered to be rough surfaces. Thus, assuming that droplets are deposited at the boundary surfaces is reasonable. Yu et al. [42] found that the aerosol resuspension rate on clothing was only 0.04–0.33% and, thus, rebounding droplets are also neglected in this study. The rate at which droplets were removed from either a boundary or by the air purifier was recorded.

Each room occupant’s facial region was assumed to have a width of 0.2 m, a height of 0.2 m, and a length extending in front of the face of 0.4 m. Droplets were released from an occupant from the center of this volume at an angle of 30° to the horizontal. The droplet positions at each time step were recorded. As previously stated, the number of droplets in the room was recorded, as well as the number of droplets passing through the air filter, and those deposited on surfaces. To study the risk of infection, the number of droplets that were exhaled from one person and moved into the other person’s facial region was monitored. These analyses were performed in post-processing using user-created Python (latest v. 3.12.5) scripts.

3. Results and Discussion

3.1. Air Purifier at Position 1

The steady-state air flow generated by the purifier was directed upwards from the purifier, bifurcated at the ceiling, moved toward the walls, and circulated above each bed before moving downwards towards the floor (Figure 3). The air speeds in the regions above the beds were less than 0.4 m/s for an air purifier outlet speed of 1.5 m/s. Similar flow patterns were observed for all three purifier speeds.

The droplet trajectories were in the same direction as the airflow (Figure 4). Due to the proximity to the air purifier, the droplets exhaled by the person on the non-lofted bed moved directly toward the air purifier, and many, but not all, were eliminated from the simulation. The droplets that were not filtered were entrained into the upward air flow and dispersed throughout the room. The droplets released from the occupant of the lofted bed circulated above the lofted bed and moved away from the person’s head toward the desk side of the room. Although the droplets had an initial speed of 3 m/s when exhaled, drag with the surrounding air quickly decelerated the droplets so that they closely followed the surrounding air streamlines.

After approximately 40 s in the steady-state air flow field, the droplets exhaled by both occupants circulate throughout the entire room, as shown in Figure 3. Although most of the droplets remained near where they were produced, many moved to other regions within the room. The droplets from the lofted bed were further from the air purifier and, thus, were more likely to travel into other parts of the room before depositing on surfaces or entering the air purifier. With an increase in the air purifier speed, the droplets traveled further in the 40 s period. The number of droplets entering the other occupant’s facial region is shown in

Table 3. Percentage of droplets from 90 to 100 s (after reaching steady-state air flow) exhaled from one occupant entering the facial region of the other occupant, relative to the total number of droplets released over the 100 s period.

Purifier Air Speed (m/s)	From Lofted bed to Non-Lofted Bed	From Non-Lofted Bed To Lofted Bed
1.0	27.1%	18.0%
1.5	26.0%	18.7%
2.0	25.2%	16.7%

. The person in the non-lofted bed had a larger number of droplets generated by the other occupant entering their facial region. A higher air purifier speed increased these cross-occupant droplets. At high purifier speeds, the droplets traveled faster and were more likely to travel to the other occupant’s facial region. This behavior occurs because the sub-micron droplets closely follow the flow streamlines, even at higher flow speeds.

The total boundary deposition rate increased with higher air purifier speeds (Figure 5). The total deposition rate for 1.0, 1.5, and 2.0 m/s air speeds ranges from 0.7 to 0.8 drops/s at 150 s, which is similar to the deposition rate reported by Blocken et al. [43], which was 0.68–0.72 particles/s for particles smaller than 0.5 µm. The Blocken et al. [43] study also indicated that, with ventilation, the deposition rate is about five times higher than with no ventilation. The increase in the deposition rate with increasing air purifier speed also agrees qualitatively with the study by Eilts et al. [44].

Figure 6 demonstrates that most of the droplets were removed from the domain due to contact with the boundary surfaces. After 140 s, the droplet removal rates at purifier speeds of 1.5 and 2.0 m/s reached steady-state at a value between 185 and 200 droplets/s. In contrast, the purifier droplet removal rate fluctuated between 7–10 droplets/s. The 1.0 m/s purifier air speed remained in a transient stage at 140 s since the total droplet removal rate did not yet equal the droplet generation rate. Regardless, the trend for the 1.0 m/s speed is the same as for the other two speeds: the vast majority of droplets are removed by the boundaries and not by the air purifier. This situation may be specific for the small, unventilated dormitory room studied here. The aerosol deposition rate within 2 m of the emission source will be large, as reported by Eilts et al. [44]. More specifically, for rooms with a high surface area-to-volume ratio, the surrounding surfaces can allow particle deposition by impaction and interception.

Simulations were also performed without the Brownian motion, with a smaller droplet density (500 kg/m³), and using a different turbulence model (Menter Shear Stress Transport). In all cases the droplet removal rate at the boundaries remains similar and far exceeds the removal rate by the air purifier. These results suggest that bulk flow features rather than small scale features dominate droplet removal.

Figure 7 and Figure 8, respectively, plot the number rate of droplets settled on the surfaces and removed by the purifier for different air purifier speeds. An increase in air purifier speed increased the number rate of droplets collecting on surfaces but decreased the number rate of droplets filtered by the purifier. Figure 7 shows a rapid increase in the rate of droplet deposition on surfaces within the first 10 s after the start of droplet generation. The deposition rate increases more slowly between 10 and 100 s as the droplets spread around the room, reaching a maximum of 190 droplets/s at 100 s. Increasing the air purifier speed increases the removal rate, but not significantly.

The rate at which droplets are removed by the purifier remained at less than 10 droplets/s in all cases, which is insignificant compared to the droplets deposited on surfaces. The 1.0 m/s velocity results exhibited a peak value at 56 s (Figure 8). These results suggest that the system is still in a transient stage and the droplet removal rate by the purifier is not at a steady state. There is no consistent trend in the droplet removal rate against the air purifier speed, but after 80 s the removal rate increases with air speed. Thus, increasing the air purifier speed increases the removal rate due to both the contact with the boundary surfaces and the airflow through the air filter.

The current results are consistent with previous studies. In their investigation of droplet dispersal in a model music classroom, Narayanan and Yang [22] showed that 92% of the droplets were deposited on boundary surfaces while only 3.2% of the droplets were removed by the building ventilation system and 3.8% by an air purifier. Chen et al. [23] reported similar results for a model dental clinic, with less than 8% of the droplets removed by an air purifier and over 80% of the droplets settling on the boundary surfaces. Blocken et al. [43] found experimentally that an air cleaning system can increase the PM 0.5 concentration in a room within a 30 min period. The current and previous results indicate that air purifiers filter only a small fraction of the droplets suspended in a room, with most of the droplets accumulating on boundary surfaces.

3.2. Air Purifier at Position 2

The results presented in the previous section demonstrate that the flow field generated by the air purifier significantly affects the droplet trajectories and the droplet removal rate. To investigate the effect of the air purifier’s position on the droplet trajectories, an additional simulation was performed with the air purifier placed in the middle of the room between the two beds (Figure 9) at an air purifier speed of 1.5 m/s.

The droplet trajectories changed after changing the air purifier’s location, although the general air flow patterns were similar (Figure 9a). The droplets from the lofted and non-lofted bed travel almost directly into the air purifier air stream (Figure 9b) and the droplets from both occupants were dispersed throughout the room by the upward-directed flow from the purifier. As a result, more droplets moved from the lofted bed occupant to the non-lofted bed region when compared to when the air purifier was in position 1. The total number of droplets removed by the purifier continued to increase after 20 s of simulation but was still less than 15 droplets/s. After approximately 60 s, the number of droplets in the room reached a steady state, where the rate at which droplets were removed by the surfaces and the purifier equaled the rate at which droplets were exhaled by both occupants. Even though the purifier in location 2 was able to remove more droplets than when it was at location 1, the total number rate removed by the surface boundaries was still much larger than the number rate of droplets removed by the purifier (Figure 10).

The percentages of droplets exhaled from one occupant and entering the other occupant’s facial region also slightly reduced after changing the purifier to position 2. In this new location, 24.0% of the droplets exhaled from the lofted bed occupant entered the non-lofted bed occupant’s facial region. In position 1, this value was 26.0%. Approximately 12.6% of the droplets released from the non-lofted bed occupant entered the lofted bed occupant’s facial region when the purifier was in position 2 as compared to 18.7% when the purifier was in position 1. These results demonstrate that the location of the purifier affects the droplet contamination. Chen et al. [23] also found that changing the air purifier location can be significant in a model dental clinic. They found that the percentage of droplets produced by one patient entering the breathing zone of another person can change from over 20% to less than 5% depending on the purifier location.

The suspended droplet concentration at a steady state is significantly affected by the purifier’s location, while the air speed has only a minor effect when increased from 1.0 to 2.0 m/s (

Table 4. Suspended droplet concentrations in the air after reaching a steady state.

Purifier Location and Air Speed	Suspended Droplet Concentration (×10−14 kg/m3)
Location 1, 1.0 m/s	5.94
Location 1, 1.5 m/s	5.05
Location 1, 2.0 m/s	4.11
Location 2, 1.5 m/s	2.70

). These results agree with the findings of Narayanan and Yang [22] who show that the purifier location greatly affects the number of droplets removed by both the boundaries and the purifier. In the second location used in the present study (middle of the room), the surface deposition rate and removal rate were higher than when the purifier was in the first location (near the side table). The position of the portable air purifier is important for air purification performance. Some positions will cause a high suspended aerosol concentration due to the purifier airflow. These conclusions are consistent with those of Li et al. [35] and Burgmann and Janoske [45].

3.3. Without an Air Purifier

As a point of comparison, simulations were also performed for a scenario in which there was no air purifier in the room. In this case, the air in the room was stagnant and any small movement was the result of the exhalation of droplets. Recall that the computational model did not include room ventilation air. The droplets released from the two occupants for this scenario are shown in Figure 11. In this case, the droplets remained close to where they were emitted and ultimately collected on boundary surfaces. There was no exchange of droplets between the two occupants.

4. Additional Remarks

The presented results draw a baseline to understand the effectiveness of dispersing droplets away by using an air purifier in the absence of ventilation in a confined space. Regardless of the scenarios simulated in this study, a substantial number of droplets remained suspended and dispersed, indicating the boundary limit to which the risk of viral transmission via an air purifier can be mitigated.

Future investigations should consider improvements to the present model, including: (1) incorporating droplet size dispersity to improve the model predictions since transport and deposition dynamics may be different for different droplet sizes. (2) Incorporating ventilation air in order to quantify the removal rate of droplets while examining the risk of creating droplet-rich zones in the room due to modified flow patterns. (3) Modeling droplet evaporation to quantify how long particles remain suspended in the air as droplet size decreases with time. (4) Including diverse mechanisms of aerosol removal, considering that smaller aerosols may be removed by coagulation and scavenging by larger aerosols while larger-sized scavenging aerosols have larger sedimentation speeds.

Overall, as with many mathematical simulations, even with representative CFD models, there are limitations, including the scenario selected to represent “reality”. While this model predicts the flow dynamics and dispersion rates on a highly intricate scale depending on the particle emission rate, e.g., by coughing, there may be a need to parameterize sub-grid-scale turbulence. Such detailed parameterizations are essential for accurately simulating confined ultrafine particle flows in quasi-static states [16,46]. Furthermore, the movement of individuals within the domain and the positions of different furniture, etc., need to be reasonably ascertained to understand the positions of thermal plumes and eddy formations in dynamic conditions.

5. Conclusions

This work was carried out to determine the effectiveness of an air purifier for removing exhaled droplets in a typical college dormitory room with no ventilation. The trajectories of the droplets exhaled by two occupants were modeled using a commercial computational fluid dynamics software package and Lagrangian particle tracking. The droplets were sufficiently small so that they essentially tracked the flow streamlines, although Brownian motion did act on the droplets due to their size. The simulations indicate that an air purifier is not effective at directly removing the droplets; however, the airflow suspends the droplets and disperses them throughout the room facilitating their removal by contact with the room surfaces. Therefore, in a real-world application, care should be taken to determine the position and operating speed of the air purifier since moving the air purifier’s location and changing the air speed of the purifier modifies the removal rate and flow field.