Blocking effect of desktop air curtain on aerosols in exhaled breath

The extant coronavirus (COVID-19) pandemic has obligated the wearing of protective face masks or face shields. Maintaining physical distance is also essential to control the spread of the virus.^1,2 Numerous studies have demonstrated that wearing face masks^3,–6 or face shields^7,–10 prevents the discharge of aerosols into the air, thereby diminishing the risk of airborne infections. However, during exhalation, face masks may leak the virus, which can be inhaled by another person when sufficient physical distance is not maintained.¹¹ Typically, the doctor and patient are in close physical proximity when blood samples are collected in medical facilities such as hospitals and clinics. Furthermore, medical procedures, such as specimen collection and endotracheal intubation, require close proximity to the patient and the removal of the face mask.¹² Thus, countermeasures to prevent infection are ineffective in such specific medical situations. Additionally, it is known that the use of personal protective equipment with extremely high virus-blocking capability, such as N95 and face shields,^13,–15 significantly impairs speech recognition, which may interfere with medical services. Therefore, it is essential to establish a system to protect staff during their medical duties.

Air curtains have been conventionally used in the operating rooms of hospitals in order to protect patients from contaminated air.^16,–19 The permanent-type air curtain system used in the operating room comprises an air curtain to block outside air and an excellent ventilation system based on precise equipment design. However, meticulous simulations of the airflow are essential before installing the air curtain system in an operating room. Such simulations are arduous to replicate in a general ward. Furthermore, it is impossible to relocate the permanent-type air curtain installed in the operating room. Therefore, it is not appropriate for general-purpose use in general wards.

Xu et al.²⁰ proposed a portable air curtain system encompassing personalized ventilation (PV)^21,–23 and a personalized air curtain (PAC).²⁰ This system successfully reduced the face-to-face virus exposure rate of persons by up to 98%. Sakharov and Zhukov²⁴ developed a face-shield-type air curtain and evaluated its performance. The results indicated that the air curtain prevented the dispersion of aerosol emitted from humans. In addition, the study indicated that it was possible to block the propagation of aerosol by utilizing a comparatively smaller system, such as a face-shield-type air curtain. However, the flow from the two-dimensional jet of the air curtain gradually loses its intensity on being dispensed from the discharge port. This causes a turbulent flow and a concomitant surge in the diffusion effect. This may enable the scattering of aerosol particles emitted from humans into the environment.

In order to protect healthcare workers from infections, we developed a desktop air curtain system (DACS). Figure 1(a) shows an overview of the DACS. The air curtain flow is produced by the generator at the top of the DACS. This flow is guided to the suction port at the bottom by protracting the length of the potential core. This prevents the dispersion of the air curtain, thus leading to the collection of all the aerosol particles emitted by humans at the suction port. Hence, the spread of the virus to the surroundings is prevented. Another noteworthy aspect is that the DACS is an integrated system consisting of both a discharge and a suction port. Therefore, the characteristics of this equipment do not depend on the installation location. As delineated in the ensuing section, since the DACS consists of circulation-type airflow, a high-efficiency particulate air (HEPA) filter can be installed inside the suction port to provide an air purification function as well. Figure 1(b) shows the use of DACS in a blood collection booth. The arm of the patient was placed above the suction port of the DACS. It can be observed that the arm blocks the air curtain flow. Hence, in this scenario, it is essential to investigate the effect of blocking on the dispersion of aerosols.

This paper delineates the experiments conducted by using the DACS in a blood-collection booth. Initially, the primary characteristic of the air curtain, namely the velocity field of the airflow, was evaluated. Next, air, simulating human exhalation (pseudo-expiration), containing aerosol particles was blown toward the air curtain generated by the DACS. The effect of the air curtain on blocking the spread of the aerosol present in the pseudo-expiration was investigated using particle image velocimetry (PIV) measurements. Finally, the scenario in which the arm of the patient obstructs the flow of the air curtain in the DACS was also examined.

Figure 2 depicts a schematic of the equipment. The DACS is placed on the desk, and the air curtain flows downward from the discharge port toward the suction port. The inlet and outlet for air, located at the bottom of the DACS, are connected to 65 mm diameter aluminum ducts. The flow rate is controlled by the speed of rotation of the DC fan connected between the ducts. A particle generator (Seika Digital Image, CTS-1000), used for the visualization of airflow, was coupled between the DC fan and the inlet of the DACS. Dioctyl sebacate particles with a size of 2–3 μm were produced from the particle generator. When the particle generator was operated, the particle concentration gradually increased, owing to the circulation of the air. In order to avoid this, ∼70% of the dioctyl sebacate particles in the air curtain were removed using a nonwoven filter attached upstream of the DC fan. Thus, the concentration of particles was maintained constant. In a preliminary experiment, when the flow rate of the DACS was too small, the spatial uniformity of the air curtain deteriorated. Hence, the flow rate Q_air of the airflow passing through the discharge port and suction port was set as Q_air = 0.039 m³/s, where the spatial uniformity becomes less than or equal to 3% at the discharge port.

The internal structure of the DACS is shown in Fig. 3. The spanwise and vertical directions of the DACS are defined as the Y and Z directions, respectively. Figure 3 also depicts cut sections exhibiting the salient features of the DACS. The terms inflow and outflow specified in Fig. 3 refer to the airflow from the air inlet to the discharge port and from the suction port to the air outlet, respectively. Figure 3(a) illustrates the internal structure at the bottom of the DACS using a surface sectioned at Z = 0.08 m. The air inlet and outlet are placed at the base portion of the DACS. After passing through the air inlet, the inflow branches into two paths. A flow control plate is installed at this branch point. The angle of this plate is adjusted such that the flow rates in both directions of the branch are the same. The inflow then moves to both the sides of the DACS and rises inside the square pole extending on the Z-axis as shown in Fig. 3(b). Subsequently, as illustrated in Fig. 3(c), the inflow advances to the air curtain generator located at the top of the DACS. The two branched airflows rejoin in this zone and pass through the slit part located above the air curtain generator as shown in Fig. 3(d), before flowing into the contraction part. The airflow is then rectified at the contraction part of the duct as indicated in Fig. 3(e). Finally, the airflow is released from the air curtain generator in the form of an air curtain flow. Eventually, the air curtain flows into the suction port of the DACS as an outflow and ultimately exits through the air outlet.

Figure 4 shows the dimensions and coordinate system of the DACS. The center of the cross section of the discharge port is set as the coordinate origin (x, y, z) = (0, 0, 0), where the coordinates x, y, and z represent the crosswise, spanwise, and vertical directions, respectively. The discharge and the suction port have thickness D = 50 mm and width W = 500 mm, with cross-sectional areas of 0.02  m² and 0.025  m², respectively. The two ports are located 500 mm away from each other. A Rouse–Hassan curve²⁵ conducive to flow rectification was deployed as the shape of the contracting section of the duct inside the air curtain generator. At the center of the contracting section, a two-dimensional cut NACA0036-shape airfoil²⁶ was placed, in order to attain the boost effect of the flow rate²⁷ by merging the airflow in the wake of the cut airfoil (see the  Appendix). The chord length and maximum thickness of the NACA0036 airfoil were 80 and 28.8 mm, respectively. The airfoil was sectioned at a position of 85% from its leading edge, resulting in a cut surface of the airfoil with a width of d = 9 mm. The cut surface of the NACA0036 airfoil was arranged to coincide with the outlet, namely the discharge port.

Particle image velocimetry (PIV) was used to visualize the air curtain flow and the aerosol particles emitted from exhaled breaths of humans. Figure 5 shows the layout of the equipment utilized for PIV measurement. In this study, PIV measurements of the air curtain flow and aerosol particles were assessed in several orientations of the equipment. The PIV measurement of aerosol particles is detailed in the forthcoming Sec. III B (Instantaneous velocity field of aerosol particles emitted from a pseudo-breath generator (mannequin)). Figure 5(a) depicts a 3D view of the DACS and the measuring instruments, namely the camera and the laser. The high-speed camera (nac, MX-5) and the laser with an output of 5 W (Seika Digital Image) were oriented at an angle of 90°. Figure 5(b) illustrates the top view of the layout of DACS and measuring instruments. It was observed that when the high-speed camera was placed parallel to the y axis, the air curtain flow was obstructed by the square pole on the front. The entire area of the airflow was photographed by maintaining the angle between the DACS and high-speed camera at 80°. The distance between the central axis of the DACS (x = y = 0) and the high-speed camera was 3000 mm and that of the laser was 3500 mm. Figure 5(c) illustrates the DACS photographed by the high-speed camera. The boundary of the rear part of the DACS is dark and almost invisible. Therefore, it is indicated by a yellow dashed line.

The measurement system used by Takamure and Ozono²⁸ was adopted. A constant-temperature hot-wire anemometer (KANOMAX, System-7000) was used to measure time-series velocity. The velocity w in the z direction was measured using I-type hot-wire probes (KANOMAX, 0247R-T5). In the vicinity of the discharge port (z/D = 0.4), a probe with a prong bent at 90° (KANOMAX, 0248R-T5) was arranged; hence, it was orthogonal to the z direction. At this time, the wire is parallel to the y axis. The I-type hot wire probe had a wire diameter of 5.0 μm and a sensing length of 1.0 mm. The analog signal obtained from the hot-wire anemometer was AD-converted using a logger (KEYENCE, NR-600) and recorded on a personal computer. The data were sampled at a frequency of 5 kHz using an analog cutoff filter at 2 kHz.

The PIV measurement helps acquire the velocity field of the laser sheet plane. However, the measurement accuracy may decrease due to the influence of the reflection and scattering of the laser sheet light near the equipment. In contrast, despite the fact that the hot-wire anemometer is a single-point measurement device, it has good measurement accuracy and very high response. Therefore, the velocity of the air curtain flow was measured using PIV and a hot-wire anemometer.

The vector field of the average velocity of the air curtain flow obtained from PIV measurements is shown in Fig. 6(a). Here, the color bar represents the magnitude of the resultant vector of the three velocity components. The direction of the resultant vector of the average velocity was pointed downward from the discharge port toward the suction port. Despite the fact that airflow is being drawn, the airflow velocity near the suction port is infinitesimal. Figure 6(b) shows the measurement results of PIV and the hot wire anemometer in the center (x = y = 0). The right axis indicates the average velocity w_ave in the z direction of the suction port. The left axis depicts the values of w_ave normalized by the cross-sectional area WD and flow rate Q_air at the suction port. The dashed lines shown in Fig. 6(b) represent the average flow rate at the cross section of the suction port [estimated by $wave/{(Qair)−1WD}=1$]. Additionally, the PIV measurements shown in Fig. 6(b) were utilized to generate the dashed-dotted line shown in Fig. 6(a). The results indicate that the distributions of the PIV measurement and the hot-wire anemometer are similar at z/D < 7. At 2 ≤ z/D ≤ 8, a boost effect of the flow rate at the region where $wave/{(Qair)−1WD}>1$ was also observed. This is because the flow rate increases when the airflow alongside the sectioned airfoil of NACA0036 in the contraction part merges with the central axis²⁷ (the details are explained in the  Appendix). At the zone of z/D > 7, the results of the PIV measurements indicate that the velocity is much smaller than $wave/{(Qair)−1WD}=1$⁠. On the other hand, the results of the hot-wire anemometers are asymptotic to $wave/{(Qair)−1WD}=1$⁠. The measurements performed using the hot-wire anemometer revealed that the flow velocity w_air and flow rate Q_air were in good agreement near the suction zone. As shown in Fig. 6(a), the measurement accuracy appears to be diminished because of the diffusion of the laser sheet light, owing to the reflection from the suction port surface. However, since the hot-wire anemometer is unaffected by the reflection of the laser light, the measurement accuracy is not compromised even in the vicinity of the suction port. Based on the measurements from the hot-wire anemometer, it was confirmed that the flow rate of the air curtain generated by the DACS is maintained up to the suction port.

Figure 7 shows the spatial profile of the air curtain flow. Figure 7(a) depicts the y direction profile of w_ave. The right axis indicates the values of w_ave. The left axis shows the values of w_ave normalized by the cross-sectional area WD and flow rate Q_air at the suction port of the DACS. The error bar represents the magnitude of the rms value of the velocity fluctuation w′. It can be inferred that at the zone where z/D = 3.2, the distribution of w_ave in the y direction varies, but at z/D = 6.4 and 9.6, the uniformity is confirmed. The inhomogeneity of averaged velocity was ascertained to be 2.8% at z/D = 6.4 and 2.2% at z/D = 9.6. Additionally, at z/D = 3.2 and 6.4, the average velocity manifests a value of $wave/{(Qair)−1WD}>1$ over the whole measurement region. This confirms the occurrence of the boost effect of the flow rate. Furthermore, it was observed that the magnitude of the rms value of the velocity fluctuation decreased when moving closer to the suction port. It was inferred that the velocity fluctuation increased when moving toward the upstream (z/D = 3.2). This occurred due to instability in the wake region caused by the merger of two streams flowing along the surface of the cut airfoil of the NACA0036 shape.²⁹ At the zone where z/D = 6.4 and 9.6, the rms value of the fluctuating velocity is small. This signifies that the potential core of this air curtain continues without turbulence until the suction port.^30,31 Figure 7(b) represents the x direction profile of w_ave. The green zone shown in Fig. 7(b) corresponds to the position of the discharge port. At z/D = 0.4, a velocity defect exists in w_ave at x/D = 0. This is because a wake stagnation region occurs on the cut surface of the NACA0036 airfoil installed in the center of the discharge port. The value of w_ave attains a maximum at |x/D| ∼ 0.3 and decreases rapidly at |x/D| > 0.5. At the zone of z/D = 3.2, $wave/{(Qair)−1WD}>1$ exhibits a value close to x/D = 0. This confirms the occurrence of the boost effect of the flow rate. Such an effect is also observed on the central axis, even at z/D = 6.4. The corresponding velocity distribution formed a Gaussian profile. At z/D = 9.6, $wave/{(Qair)−1WD}$ exhibits a value of 1 near x/D = 0. This value is the same as the flow rate in the cross section of the suction port. The velocity measurements revealed that the flow rate of the air curtain at the center (x = y = 0) is maintained at the same level as that of the discharge and suction ports.

A pseudo-breath generator was created to generate airflow that assumed exhalation by humans. Figure 8 shows the schematic of the pseudo-breath generator. Air is blown from an air compressor (AIRREX, HX4004) to a particle generator (Seika Digital Image, CTS-1000). This air, which contains dioctyl sebacate particles with a particle diameter of 2–3 μm, flows into an acrylic air buffer at a flow rate of 52 l/min. The air stored in the buffer passes through a pressure-resistant rubber hose and is connected to a stainless steel tube of 18 mm inner diameter. The stainless steel tube is released at the mouth of the mannequin, and a steady flow of 52 l/min is blown out. The flow rate of the pseudo-breath generator is approximately the same as the instantaneous maximum flow rate (60 l/min) in the respiratory model set by Akagi et al.⁷ The distance between the outlet of the exhaled breath of the pseudo-breath generator and the center of the DACS was 250 mm. This setup was located 200 mm downstream of the discharge port of the DACS (z = 200 mm).

Figure 9 shows the top view of the DACS, measuring equipment, and pseudo-breath generator. The distances between the center of the DACS (x = y = 0) and the high-speed camera and laser are 3000 and 3500 mm, respectively. The angle between the DACS and the high-speed camera is set to 105°. The pseudo-breath generator is positioned perpendicular to the air curtain flow. In this study, three cases were investigated. Case I is the scenario where the DACS is stopped (not operated) and aerosol particles are generated from the pseudo-breath generator. Case II represents the scenario in which the DACS is operated and aerosol particles are generated from the pseudo-breath generator. Finally, case III assumes the use of this setup in a blood collection booth, where the DACS is operated and aerosol particles are generated from the pseudo-breath generator. In this scenario, the arm of the mannequin having a diameter of 82 mm was positioned at a distance translated by 100 mm in the y direction from the center of the gate of the DACS.

Initially, the behavior of the aerosol particles generated from the pseudo-breath generator, corresponding to case I, was observed when the DACS was not operating. Figure 10 shows the instantaneous field of aerosol particles (left figure) and velocity vectors of aerosol particles (right figure; Multimedia view). The color bar of the vector represents the magnitude of the resultant velocity vectors of the three components. As shown in the left figure of Fig. 10, the aerosol particles diffuse while moving forward. The PIV measurement indicates that (right figure of Fig. 10) the velocity of the aerosol particle attains the maximum value instantly upon the ejection of the particles from the mannequin’s mouth. Subsequently, the velocity gradually attenuates.

Figure 11 shows the instantaneous field (left figure) and velocity vector (right figure) of aerosol particles in case II when the DACS is operating (Multimedia view). Similar to case I, the velocity of aerosol particles achieves its maximum immediately after ejection from the mannequin’s mouth. The particles diffuse when moving forward. However, when the aerosol particles approach the gate of the DACS, it can be observed that the orbit of particles changes downward and is eventually sucked into the suction port.

Figure 12 shows the instantaneous field in case III when the DACS is operating and the arm is placed on the gate (Multimedia view). The visualization of aerosol particles in the left figure of Fig. 12 indicates that when the particles approach the gate of the DACS, they orbit downward in accordance with the air curtain flow, similar to case II. However, as can be inferred from the result of the PIV measurement on the right of Fig. 12, the instantaneous velocity vector may be disrupted on the upper side of the arm. In other words, the arm placed on the gate may obstruct the air curtain flow.

In Sec. III B, it was indicated that the air curtain flow may be disrupted when the arm is placed on the gate of the DACS. In order to quantitatively evaluate the effect of airflow due to the placement of the arm on the gate, this section evaluates statistically the effect of the presence or absence of the arm on the behavior of aerosol particles. Figure 13 depicts the spatial distribution of the rms value of the velocity fluctuation of aerosol particles in case II and case III. The color map shows the magnitude of the rms value of the resultant velocity. A minor difference in rms values was observed near the mouth of the mannequin. This is because the particle tracking becomes unstable due to the high particle concentration near the mouth. Therefore, the measurement accuracy decreases. When the arm is not placed on the gate (case II), minute fluctuations are observed at the confluence of aerosol particles and air curtain flow emitted from DACS. However, when the arm is placed on the gate (case III), a large fluctuation is observed in the region above the arm. This implies that the air curtain flow is constantly turbulent when the arm is placed on the gate.

The analysis of the average velocity field confirmed the blocking effect on aerosol particles, owing to the turbulence generated above the arm. Figure 14 shows the average velocity field of aerosol particles in each case. In case I where the DACS is not operating, the aerosol particles continued to diffuse even after passing through the gate of the DACS. On the other hand, in case II (DACS is operating), the aerosol particles travel downward on the air curtain and are not observed on the opposite side of the air curtain. In case III (DACS was operating and the arm was placed on the gate), aerosol particles were not observed on the other side of the DACS. This indicates that they were blocked by the air curtain.

The aforementioned studies reveal that aerosol was blocked by an air curtain with or without an arm placed on the gate of the DACS. The DACS also helps to prevent secondary infections because, in its absence, the contaminated air is filtered and reused due to the circulation of airflow. This desktop-type air curtain system is effective as an indirect barrier not only in the medical field but also in situations where sufficient physical distance cannot be assured, such as at the reception counter, or in situations where people face each other and are obligated to remove their masks, such as in restaurants or food courts.

A desktop-type air curtain generator capable of being installed on a desk was developed as an air curtain for medical use. Pseudo-exhaled air containing aerosol particles emitted from a mannequin was blown toward the air curtain generated by the DACS. The aerosol-blocking effect of the DACS was investigated using PIV measurements. Several scenarios were assessed in which the arm of the patient was placed on the gate of the DACS, assuming its use in the blood collection room. The principal conclusions are as follows:

1. By measuring the airflow velocity using a hot-wire anemometer, it was confirmed that the flow rate of the air curtain flow generated by the DACS sustains until the suction port.

2. When the air curtain is not operating (case I), the velocity of aerosol particles emitted from exhaled breath takes the maximum value immediately after being ejected from the mannequin’s mouth. The particles continue to diffuse even after passing through the gate of the DACS.

3. When the air curtain is operating and the arm is not placed on the gate (case II), on approaching the gate, the aerosol particles travel downward in the air curtain. Hence, the aerosol particles are not observed on the opposite side of the DACS.

4. When the air curtain is operating and the arm is placed on the gate (case III), aerosol particles follow the air curtain flow and orbit downward when approaching the gate of DACS. The aerosol particles are blocked by the air curtain flow both in the presence and in the absence of the arm on the gate of the DACS.

This work was supported by the Adaptable and Seamless Technology Transfer Program through Target-driven R & D (A-STEP) from the Japan Science and Technology Agency (JST), the Grant-in-Aid for Early Career Scientists (Grant No. 21K14074), the Grant-in-Aid for the Project of Design and Engineering by the Joint Inverse Innovation for Materials Architecture (Grant No. DEJI2MA) of the Ministry of Education, Culture, Sports, Science and Technology, Japan, the Foundation of Public Interest of Tatematsu, and the Fuji Science and Technology Foundation. The authors would like to thank the staff at Nagoya University Hospital for providing useful knowledge for conducting this research.

The authors have no conflicts to disclose.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

In order to design the DACS, initially, the NACA0036 airfoil with the rear end sectioned was inserted inside the contraction part of the duct. The boost effect of the airflow was investigated on this airfoil in advance, using a numerical simulation. The simulation was performed using SCRYU/Tetra, a general-purpose thermofluid simulation software that uses a hybrid mesh to represent the surface shape of Software Cradle Co., Ltd. The fluid simulation was performed using a large eddy simulation (LES), and the standard Smagorinsky model was used as the turbulence model.

The computational domain is illustrated in Fig. 15. In this simulation, the actual size of the DACS was set. The computational domain was 610 mm in the x direction, 500 mm in the y direction, and 775 mm in the z direction. The grid number was ∼13.5 × 10⁶ in the computational domain, and the time step was set such that the maximum Courant number was less than 0.5. The boundary conditions are the inflow and outflow to the inlet and outlet areas, respectively, with a uniform flow rate of Q_air (=0.039 m³/s). A static pressure condition was imposed on the boundary with a width of 130 mm, located at the bottom of the y–z plane. The boundary region of the x–z plane was subjected to the free-slip condition, and the other boundary regions were subjected to the non-slip condition.

Figure 16 shows the average velocity field of the air curtain flow when the NACA0036 airfoil with the rear end cut is not placed (notated as Normal duct) and when it is placed (notated as duct with cut airfoil) in the contraction part of the duct. In the case of the normal duct, the airflow is uniformly ejected in the z direction near the discharge port of the contraction part, and the flow velocity is slightly attenuated as it flows down. Since the cross-sectional area of the outlet of the duct (discharge port) with cut airfoil is smaller than that of the normal duct, the airflow accelerates at the blowout part. The airflow ejected from the contraction part flows toward the center axis (indicated by the one dot chain line) and then merges at approximately z = 100 mm (z/D = 2). The flow velocity of the merged airflow is faster than that of the normal duct. This is termed the boost effect of the flow rate. The flow velocity at the suction port of the airflow is almost the same as in the case of the normal duct and that of the duct with a cut airfoil.

Figure 17 shows the average velocity of the airflow in the central cross section (x = y = 0) obtained by numerical simulation. For comparison, the results of measurement using a hot-wire anemometer when using a duct with a cut airfoil are also shown. When the normal duct is used, the potential core region (region that takes $wave/{(Qair)−1WD}=1$⁠) continues over the range of 0 ≤ z/D ≤ 5. At z/D > 5, the airflow decreases slightly, but it accelerates again near the suction port due to the effect of the suction flow. When using the duct with the cut airfoil, the airflow takes a negative value near z/D = 0. In the range of 0.3 ≤ z/D ≤ 2, the airflow accelerates rapidly, and at z/D ≥ 2, $wave/{(Qair)−1WD}$ is greater than 1. The flow velocity becomes maximum at z/D = 4. At z/D = 4, the maximum velocity of w_ave is ∼1.4 times that of the normal duct, which means that the blocking effect of the air curtain is better. The airflow decelerates at z/D > 4, but the airflow velocity is always higher than that of the normal duct. Thus, a duct with a cut airfoil generates a stronger air curtain than a normal duct.