Measurement of Breath Velocity and Volume on Speaking, and Performance of Local Exhaust System as Prevention Measure of Infection.

. This study proposes a novel approach combining a local exhaust ventilation system (LEV) and a whole underﬂ oor air distribution system (WUFAD) in the consulting room. This study assumes that two persons (doctor and patient) are sitting face to face and talking without a mask in a simple room regarded as a consulting room. To model people talking, exhaled air velocity and volume from talking were required, so we estimated velocity speed using an ultrasonic anemometer and volume using a PET bag and mask. Then CFD steady analyses were carried out, setting various parameters (hood height, hood ﬂ ow rate, horizontal hood position

Airborne infection caused by droplet nuclei and micro droplets moving as an aerosol is regarded as one of the possible ways to infect SARS-CoV-2 [1]. Though various measures have been taken to prevent it in many places, entire system-scale ventilation might not be enough to become the solution for preventing shortrange airborne transmission [2]. Still, many scenarios cannot avoid close-distance conversations, for example, in a consulting room, restaurant, or crowded train. The consulting room could be especially hazardous because it has significant potential for doctors to contact infected patients. Therefore, this study proposes a novel approach combining a local exhaust ventilation system (LEV) and a whole underfloor air distribution system (WUFAD) in the consulting room. In WUFAD, the conditioned air is supplied from the entire floor through many small halls or carpets, not through some diffusers. Because the capture performance of LEV is significantly affected by surrounding airflow [3], the WUFAD system, which can make a calm airfl ow fi eld [4], is expected to help LEV's performance. To model people talking, exhaled air velocity and volume from talking were estimated using an ultrasonic Measurement of Breath Velocity and Volume on Speaking, and Performance of Local Exhaust System as Prevention Measure of Infection.
Jun YOSHIHARA 1* , Toshio YAMANAKA 1 , Tomohiro KOBAYASHI 1 , Narae CHOI 1 , Noriaki KOBAYASHI 1 , and Ren ZHANG 1 Abstract. This study proposes a novel approach combining a local exhaust ventilation system (LEV) and a whole underfl oor air distribution system (WUFAD) in the consulting room. This study assumes that two persons (doctor and patient) are sitting face to face and talking without a mask in a simple room regarded as a consulting room. To model people talking, exhaled air velocity and volume from talking were required, so we estimated velocity speed using an ultrasonic anemometer and volume using a PET bag and mask.

Measurement of exhaled air velocity and volume from talking
To model people talking, exhaled air velocity and volume from talking were required, so we estimated velocity speed and volume. The ultrasonic anemometer (Sonic corporation; ultrasonic anemometer model DA-700 with TR-92T 30 mm probe) is installed in front of the mouth, as shown in Fig. 1 (a). Probe model TR-92T can measure the mean velocity of each X, Y, and Z axes between 30mm in 10Hz. Subjects are set to read the same Japanese script on the wall for 1 minute three times. When subjects talk, they put their jaw on E3S Web of Conferences 396, 02033 (2023) https://doi.org/10.1051/e3sconf/202339602033 IAQVEC2023 the mark to keep the same distance between mouth and probe as in Fig. 1. In this experiment, subjects were told to read the script as they tried to talk with a person standing at a wall distance of about 1.5m. All subjects were Osaka University students (3 males, and 3 females, aged 21 to 24, averaging 22.5). After getting the velocity of each axis, equation 1, and 2 was used to calculate the angle of the vector X-Y (α) and X-Z (β), shown in Fig. 1 (a). The volume of exhaled air when people talk was measured using a mask, check valve, and PET bag, as shown in Fig. 2. Subjects spoke with the mask whose edge was covered by rubber so as not to leak their breath. Their breath went to the PET bag through a silicon tube. In the middle of the tube, a check valve

Simulation cases
To evaluate the effect of the combination of the hood and underfloor air supply, four paramitas were made, hood flow rate (Qh), hood height (hood-head distance = Hh), horizontal hood position (Ph), and ventilation rate (ACH), shown in Table 4. If there is a hood or not should be evaluated. Hence the case without a hood is set (the hood fl ow rate is 0 m3/h in Table 4). All simulation cases are divided into four groups Case A, B, C, and D. Case A's purpose is to evaluate the effect of each parameter on the hood in high ventilation conditions (50 ACH). Therefore, SA from an underfl oor is fi xed as 1,000 m3/h (50 ACH), and Qh, Hh, and Ph are changed from the standard conditions, as shown in Table 5. Case B's purpose is to evaluate the eff ect of each parameter on the hood in normal ventilation conditions (6 ACH) as a consulting room; REHVA says 6-12 ACH for consulting room [2]. SA from an underfl oor is fi xed as 120 m3/h (6 ACH), and Qh, Hh, and Ph are changed from the standard conditions, as shown in Table 5. Case C's purpose is to evaluate the eff ect of the ventilation rate with the hood (Qh = 100 m3/h). Qh, Hh, and Ph are fi xed (Qh = 100 m3/h, Hh = 500 mm, Ph = 0 mm), and the ventilation rate is changed 6-50 ACH by changing the SA flow rate 120-1,000 m3/h, shown in Table 5. Case D's purpose is to compare the eff ect of the hood eff ect by each ventilation rate. To compare Case C, there is no hood in Case D, shown in Table 5. Case A is 17, Case B is 14, Case C is 6, Case D is 6, and the total number of cases is 43.

Evaluation index
To estimate the hood's performance, η: capture efficiency has been used. Capture efficiency η is calculated by the equations as follows: [-]: Hood capture effi ciency [m 3 /h]:Hood fl ow rate [m 3 /h]:Ceiling exhaust fl ow rate was installed in order not to leak the breath that got into the PET bag. When subjects breathed, they could remove the mask, and after attaching it, they reread the script. After subjects spoke for one minute, the volume of breath in the PET bag was measured using a dry gas meter (SINAGAWA), shown in Fig. 2. Subjects were the same for the velocities experiment. All subjects carried out each experiment 3 times.

Analysis conditions
This study used computational fl uid dynamics (CFD) to simulate the cleanroom conditions. The general method of CFD is shown in Table 1. The analysis was a steady state, and the number of cycles was 3,000, confirming its convergence under 10-4. Fig. 5 shows the analysis room (2.4×3.8×2.2 = 20.064 m3), and the patient and doctor sit and talk facing each other. The distance from the patient mouth to the doctor's mouth is 1,200 mm to simulate close-distance conversation. It is assumed that SARS-CoV-2 infects the patient, and he is unconscious of it. Then they are possibly to talk without masks after SARS-CoV-2 has converged. Therefore, this research can be applied to another infection, though the value of quanta varies from each infection. Tracer gas is emitted from only the patient mouth, and its density is the same as general air to simulate droplet nuclei emitted from patients by talking. It is assumed that most of the droplet nuclei moved on the fl ow as a passive contaminant, then the movement can be simulated using tracer gas. Table 2 shows the boundary conditions of CFD. The velocity, volume, and angle of tracer gas were given as the experiment's result in this paper, which is discussed in later parts, to simulate average conversation in a steady state. Turbulence conditions are summarized in Table 3. The supply air (OA) comes from the whole fl oor, with two exhaust routes. One is a local exhaust, and the other is the ceiling exhaust to balance the room flow rate balance, shown in Fig 3(a). The local exhaust (hood) is installed above the patient in Fig 6. The hood shape is the flanged hood that performed high capture efficiency under crosswind conditions in Komori et al. [3]. The number of full-mesh is 2,444,148 (106(x)×183(y)×126 (z)), and the fi rst mesh from the wall is 100 mm.     Fig. 4(a) shows v z was the main element of the breath, and this measurement was found to be able to follow the conversation through 1 minute because this time changing shape was similar to the previous experiment; Gupta et al. [8]. The results of breath speed are summarized in Fig.  5(a). In Fig. 4(a), the negative value of the v z speed means the subject inhaled. The negative value should not be used for calculating the mean value because the purpose is to estimate the breath speed from talking. For this reason, only the elements vz >0 were used for calculating the mean value, as shown in Fig.5  (a). 0.30 m/s for mean value. Compared to previous research, an average initial air velocity from speaking, Kwon et al.,4.07 m/s for males [9], and Chao et al.  Fig. 6 shows the calculation results of the hood capture efficiency in Case A, B, and C. In Fig. 6. (c), the parameter was horizontal hood positions, η was high in Case B. However, η was signifi cantly changed in Case A. These results could be caused by the effect of the fl ow rate diff erence between the hood and the ceiling exhaust. In Fig. 6. (d), the parameter was air change rate by changing SA, η was close to 100% in any air change rate. Therefore, Fig. 6 shows that the horizontal hood position under 50 ACH conditions (SA 1,000m 3 / h) signifi cantly aff ects the hood capture performance. Fig. 7 is the calculation results of n d : the quanta concentration in front of the doctor's mouth (quanta/ m3) in Case A, B, C, D. In Fig. 7. (a) and (b), n d was low enough to fi nd it has almost no potential to infect (n d <10 -25 in Case A, n d <10 -3 in Case B), except hood 0 m 3 /h. In Fig. 7. (c), the horizontal hood position in Case B had a significant effect on the infection risk, contrary to the result of capture efficiency (Fig.  6. (c)). In Fig. 7. (d), n d was low in 6-50 ACH with hood 100 m 3 /h (n d =1.07×10 -4 in 6 ACH, n d =6.05×10 -43 in 50 ACH). n d was also low without a hood (Case D); however, under 6 ACH conditions, nd was high (n d =4.96×10 -2 ). Fig. 8(a)-(f) is the tracer gas concentration distributions in Case C, the parameter was air change rate, and the hood was fi xed to 100 m 3 / h, and Fig 8(g)-(l) is that of Case D, the parameter was air change rate, and there was no hood. Comparing Case C and D, the hood's eff ect is confi rmed. The air change rate became lower, and the better hood could work, especially under 6 ACH conditions, which is axis. However, to simulate average conversation in CFD, this measurement can be applied by giving initial air velocity from a 30 mm scale mouth. As a result, the mean value of all subjects was given as the CFD boundary conditions, as shown in Table 2. In addition, significant momentum elements should be considered to estimate the accurate angle. In Fig. 4 (a), the elements which v z 0.5 looks have enough momentum to estimate the precise angle. For these reasons, only the elements v z 0.5 were used for calculating the mean value β=11.9 °. This was given as the CFD boundary conditions, as shown in Table 2. The result of the volume measurements is summarized in Fig. 5 (b). As shown in Fig. 2 , this volume is the only exhaled breath exhaled in one minute by reading the script with a mask. Hence, the mean value of all subjects, 5,21 L/min, was given as the CFD boundary conditions, in Table 2. Though this result was small compared to the Gupta et al., 10.48±0.76 L/min for males (body surface area 1.8 m 2 ) [8], this might be because this result was given by measuring the total volume of the PET bag after speaking for one minute, so this value didn't consider speaking time. If time is considered, the results could be bigger than 5.21 L/min. The mouth size of the patient model was 17.04×17.04 mm (2.90 cm 2 ); calculated by the results, breath speed is 0.30 m/s, and breath volume is 5.21 L/min; this is consistent with the 30 mm scale mentioned above and previous research; Gupta et al. found mouth opening area 1.80±0.03 cm 2 [8].  [2]. Fig. 9 shows the contribution rate of the exhaust opening air to be exhausted at a point calculated using SVE 5. The red area (SVE 5 for the hood is 1) means the air at the point is exhausted through the hood 100%, and the blue area (SVE 5 for the hood is 0) represents the air at the point is exhausted through the ceiling exhaust 100%. In Fig.  9, the bigger an air change rate becomes by changing SA from the whole floor, the larger the flow rate difference between the hood and the ceiling exhaust. In Fig. 9 (b)-(f), the small red area means a tracer gas from the patient mouth couldn't expand to the doctor's breathing area, and Fig. 9 (a), the large red area means a tracer gas from the patient mouth could expand to the doctor's mouth. Therefore, Fig. 7. (d), Fig. 8, and Fig.  9 show that the underfl oor air distribution system can keep the doctor safe from COVID-19 over 12 ACH. However, only the underfl oor air distribution system is not enough under 6 ACH (SA=60 m 3 /h/persons), and then the hood can work to reduce infectious risk from n d =4.96×10 -2 (no hood) to n d =1.07×10 -4 (hood 100m 3 /h).

Conclusions
In this study, to evaluate the performance of local exhaust systems with whole underfl oor air distribution systems, measurement of exhaled air from talking and CFD analysis were carried out. The fi ndings obtained in this study are summarized as follows: by the result of exhaled air velocity 0.30 m/s was obtained, only the elements v z >0 were used for calculating the mean value. To consider enough momentum to estimate the precise angle, only the elements v z 0.5 m/s were used for calculating the mean vertical angle β=11.9°. As a result of the volume measurements, 5.21 L/min was obtained.
The horizontal hood position under 50 ACH conditions (SA 1,000m 3 /h) significantly affects the hood capture performance.
The underfl oor air distribution system can protect the doctor from COVID-19 over 12 ACH. However, only the underfloor air distribution system is not enough under 6 ACH (SA=60 m 3 /h/persons), and then the hood can work to reduce infectious risk from n d =4.96×10 -2 (no hood) to n d =1.07×10 -4 (hood 100m 3 /h).
The current results are limited to the CFD simulation so another experimental approach will be required. In addition, infectious risk should be evaluated consid ering droplets, wh ich might h ave more signifi cant potential to cause COVID-19 infection.