key: cord-0835183-5ronkll3 authors: Liu, Zhijian; Wang, Tianci; Wang, Yongxin; Liu, Haiyang; Cao, Guoqing; Tang, Song title: The influence of air supply inlet location on the spatial-temporal distribution of bioaerosol in isolation ward under three mixed ventilation modes date: 2022-03-06 journal: Energy and Built Environment DOI: 10.1016/j.enbenv.2022.03.002 sha: 805de33d07e3236522af44685e2ef9945d102efc doc_id: 835183 cord_uid: 5ronkll3 The outbreak of COVID-19 and the spread of infectious pathogens through bioaerosols have once again aroused widespread concern worldwide. Isolation ward is an important place to prevent the spread of infectious bioaerosols. However, infection of health care workers (HCWs) in the isolation ward often occurs, so it is urgent to carry out relevant research to reduce the cross-infection between HCWs and patients. In this paper, the temporal and spatial distribution characteristics of bioaerosols under three mixed ventilation modes in a single ward were studied, namely, upper supply side return air of Case 1 and side supply and side return ventilation are Case 2 and Case 3 respectively. The results show that the removal efficiency of bioaerosol in the ventilation mode of Case 3, in which directional airflow is formed from the air supply inlet to the release source and then to the exhaust outlet, is 46.6% and 67.7% higher than that of Case 1 and Case 2, respectively. In addition, ventilation methods based on mixed theory do not guarantee good air quality in the breathing zone (1.3m to 1.7m) of HCWs, which may increase the inhalation risk for HCWs. It is hoped that our results can provide some useful suggestions for optimizing the airflow layout of the isolation ward, reducing the risk of cross-infection, and virus elimination. Since the 21th century, humans have experienced a variety of infectious diseases caused by airborne pathogens, such as severe acute respiratory syndrome (SARS), Middle East Respiratory Syndrome (MERS), tuberculosis, and airborne diseases have been of great concern around the world [1] [2] [3] [4] [5] [6] . Although droplets and contact transmission are considered to be the main routes of transmission of COVID-19, there was evidence that aerosol transmission is also an important route [7, 8] . Isolation wards is one of the most effective devices to prevent the spread of infectious diseases outside. But it also carries a higher risk of cross-infection between HCWs and patients. During this outbreak, a large number of health care workers (HCWs) are infected. Effective ventilation strategies were extremely important to reduce cross-infection [9] . Guidelines for the design of isolation ward have been published by different national health boards [10] [11] [12] . However, there is still no consensus on standards such as the placement of air inlets and outlet, as well as the number of air changes to minimize the risk of transmitting infection. Indoor air distribution mainly depends on the location of the inlet and outlet. How to improve the efficiency of ventilation, ensure indoor air quality and reduce the inhalation and contact risk of HCWs has always been the focus of research in the field of health care. Liu et al. studied aerosol diffusion in ventilation rooms with downward air supply, and the results showed that the lower air supply velocity was beneficial to aerosol diffusion control [13] . Qian et al. investigated the downward ventilation system of hospital wards. It was suggested to further study the location of air supply diffuser and exhaust outlet in the downward ventilation system. Airflow structures that efficiently remove pollutants should be designed to reduce the risk of cross-infection in general hospital wards [14] . Lu et al. believed that the air velocity in the occupied area of mixed ventilation was low and had no direction. The mixing of fresh and old air leaded to the widespread spread of pollutants in the breathing zone. This can lead to a higher risk of infection. It was also pointed out that the pollutant removal performance of displacement ventilation is more sensitive to the location and direction of pollution sources [15] . Research by Villafruela et al. showed that when other parameters were equal. The position of the air inlets and outlets and the type of air diffuser had a great influence on the ventilation quality and air flow pattern. It also pointed out that the conclusions obtained with one type of diffuser cannot be deduced to other diffusers [16] . Berlanga et al. evaluated four different hospitals mixed ventilation modes and compared them with replacement ventilation modes, and found that replacement ventilation had better performance in hospital room environment [17] . Zhang et al. conducted a comparative study on the mixed ventilation of air supply from the ceiling or the upper side wall and the attached ventilation based on adaptive wall in the negative pressure isolation ward under the same ventilation times, and found that the attached ventilation could reduce the average concentration of pollutants by 15%-47% [18] . Wang et al. studied the fate and patient intake ratio of respiratory particles in four-bed wards under air flow modes of impingement jet ventilation and two types of mixed ventilation. The results showed that mixed ventilation was less effective in removing particles and reducing intake in susceptible patients than impingement jet ventilation. It was also pointed out that displacement ventilation has better air quality than mixed ventilation in occupied areas [19] . Bin Zhao et al. analyzed the relationship between infection probability and ventilation efficiency of HCWs in a confined space by combining the Wells-Riley equation [20] . However, the above research focused more on improving the ventilation efficiency of ward. In addition, compared with airborne transmission, the risk of contact transmission caused by bioaerosols deposited on all surfaces of ward cannot be ignored, especially in the hospital bed frequently touched by the HCWs and physical contact with the patient. Deposition of infectious bioaerosols exhaled by patients in wards is affected by many factors. The survival potential of bioaerosol on a surface depended on the properties of the surface, especially its water content, and also on the type of virus and its tolerance to dry conditions [4] . The New England Journal reported that SARS-COV-2 remained alive in aerosols for three hours during the experiment [21] . Zhao et al. used numerical simulation to predict the concentration and deposition of aerosol particles in displacement and mixed ventilation room, and the results showed that the deposition and concentration of particles with different particle sizes were greatly affected by ventilation method, and the particle sedimentation rate of displacement ventilation room was lower than that of mixed ventilation room with the same particle properties [22] . Therefore, it is necessary to analyze the deposition of bioaerosols under different ventilation modes in the ward to guide the disinfection of ward and reduce the risk of contact transmission. Since the 1970s, Computational fluid dynamics(CFD) method has been introduced into the ventilation system research, and it has been increasingly used in the ventilation system of operating room [23] [24] [25] , general hospital ward [26] [27] [28] , isolation ward [29] [30] [31] and even subway vehicles [32] and industrial buildings [33] . This paper studied the distribution of bioaerosols in a typical single-person isolation ward by CFD method. The accuracy of the mathematical model was verified by comparing the experimental data with the numerical simulation results. The deposition, suspension and removal characteristics of bioaerosols in empty rooms and actual rooms under three air supply modes were quantitatively analyzed. The relative positions of the air supply inlet and hospital bed, as well as the relative positions of the air supply inlet, release sources and exhaust outlet on the migration characteristics of bioaerosols were considered, and the temporal and spatial distribution of bioaerosols were predicted. Meanwhile, the concentration change in the breathing zone of the entire isolation ward was quantitatively analyzed. It is hoped that our results can provide some useful suggestions for optimizing the airflow layout of the isolation ward, reducing the risk of cross-infection, and virus elimination. At the same time, it must be acknowledged that the three ventilation modes studied in this paper cannot represent all situations, which is also the limitation of this study, but the situation studied in this paper is well representative [34] . In our follow-up work, we will continue to carry out relevant studies, hoping to obtain more valuable research results. As illustrated in Fig. 1 , the physical model used in this study is a typical single Based on these three ventilation strategies, the effects of different air supply inlets in isolation ward on bioaerosol deposition, spatial distribution and removal efficiency were studied. Our previous research found that facility layout may influence aerosol distribution [35] . Therefore, two scenarios were considered: empty room and actual room. In all scenarios studied in this paper, the air supply flow is 430m³/h, equivalent to 14 air changes per hour (ACH). According to the research focus of this paper, the internal layout of the isolation ward is simplified accordingly. In the actual room, it contains only a single bed and a patient. Such simplification is considered acceptable. Grid density and quality affect the accuracy of numerical simulation results [36] . The isolation ward space was meshed by ICEM CFD ANSYS software. The mesh used in the simulation is unstructured tetrahedral mesh, the minimum mesh size is 0.02m, the mesh quality is greater than 0.3, meet the requirements of calculation. Mesh independence tests were carried out with three different mesh numbers. In order to ensure sufficient refinement of the mesh, regions with large velocity gradient and human body surface were encrypted in three meshes. In these 6 scenarios, the velocity of the vertical center line of the air inlet was used to verify the independence of the grid, as showed in Figures 2 and 3 . The comparison results show that the velocity difference of three kinds of grids is small. Considering the accuracy and computation amount, the grid with medium mesh density is selected for subsequent simulation. In order to simplify the problem of this study, we introduce the following hypothesis: (1) Indoor air is an incompressible ideal gas [37] ; (2) Particles hitting all surfaces are caught immediately and will not bounce, that is, all surfaces are set as trap boundary conditions [38] ; (3) When the particle reaches the exhaust outlet, it is considered that the particle is removed, and the exhaust outlet is set as the boundary condition of escape [38] ; (4) Each surface is adiabatic and the temperature is uniform throughout the room [39] ; (5) The evaporation, condensation and breakage of particles are not considered, that is, the size and shape of particles remain unchanged [39] . Methods for simulating indoor turbulence include direct numerical simulation (DNS), large eddy simulation (LES) and Reynolds average Navier-Stokes (RANS). The application of DNS and LES in turbulence simulation requires high computer memory and computational speed [40, 41] . In recent years, RNG k-ε turbulence model of RANS method has been extensively used in indoor airflow simulation, and has been verified in RANS turbulence model [42] . Compared with other RANS methods, the RNG k-ε model has higher accuracy, calculation speed and stability in indoor environment modeling [43] . Therefore, this study chooses the RNG k-ε turbulence model to calculate turbulence. Equation 1 is the general form of the governing equation: where is the air density, ⃗ is the air velocity vector, represents each of the three velocity components, is the effective diffusion coefficient of , and is the source term. In the simulation, the velocity quantity on the wall surface was obtained under the condition of no slip. The air supply inlet was set as velocity inlet with turbulence of 5%. As mentioned above, the air supply rate of the isolation ward is 430m³/h, which is equivalent to 14 air changes per hour (ACH). The magnitude of the velocity was calculated by the air exchange rate, which was 1.843m/s. To maintain a negative pressure environment in the isolation ward. The exhaust outlet was set as pressure outlet boundary with a pressure of -15 Pa and turbulence intensity of 5%. Conservative control equations were discretized by the finite volume method. In order to improve the accuracy of numerical calculation, constraint terms and diffusion constraint terms in the second-order upwind discrete control equation were adopted, and the SIMPLE algorithm was used to calculate the flow field. In order to track the particle trajectories, the Lagrange particle tracking method was used to calculate the particle phase motion, and the discrete random walk model was used to simulate the particle dispersion in turbulence. In the process of simulation, the steady-state calculation of airflow field was carried out first, and then the transient calculation was carried out by adding particles. Its governing equation is: where u i and u pi are the instantaneous velocities of the fluid and particles, respectively;  is the molecular viscosity of the fluid; and are the densities of the fluid and particles, respectively; d p is the diameter of the particles; Re is the particle Reynolds number; C D is the drag coefficient; g i is the gravitational acceleration in the i direction; and F ai is the additional force exerted on the particles. According to the simulation and experimental results of Chang and Hu, Saffman lift force has a greater impact on particles with a particle size of 2.5-5 μm, while Brownian force has a greater impact on particles with a particle size of less than 0.5 μm [44] . In addition, the ratio of air density to particle density is very small, so compared with other external forces on particles, pressure gradient force, virtual mass force and the Basset force can be ignored in this study [38] . It is assumed that the temperature of the room is evenly distributed and there is no temperature difference, so thermophoretic force is ignored [39] . In the simulation, the size of the patient's mouth was 0.025 m × 0.025 m, and the respiratory rate was about 10 L/min. The particle size used for simulation in this paper was 2.5μm, and 900s were released to stabilize the particle concentration. The simulation was conducted in 1400s, because bioaerosol particles had been basically deposited on the surfaces of the isolation ward or removed at this time. Therefore, this study ignores the influence of Brownian force and only the Saffman lift force was considered. To verify the correctness of the mathematical model used in this paper, Yang et al.'s experimental data of full-size airflow velocity field in an ISO-5 clean room was used [45] . Fig. 4 shows the layout of the clean room and the location of the measuring points. In this paper, the Lagrangian particle tracking method was used to calculate the motion of the discrete phase. In order to effectively verify the correctness of the model, the experimental data in the literature [46] were compared with the simulated value. Fig. 6(a) ventilation rate of the room was 9.216 ACH. In the experiment, and the particle size of smoke was in the range of 0.5-5μm. In the simulation, particles smaller than 1μm were included in the range of 1μm, and the density was 865.0kg/m 3 . The specific process of the experiment is as follows: particles were initially injected into zone 1, and the opening between the two zones was closed before the measurement began. When particles were evenly distributed in zone 1, ventilation and opening were opened, and the particle mass concentration in the two zones was measured. The particle mass concentration ( ) is determined by Equation (3). In the simulation, the total particle mass of injection zone 1 was determined by measured data. The particle mass was evenly distributed in each size group. It was assumed that the initial concentration is uniformly distributed in 5 particle sizes ranging from 1 to 5μm. The experimental and simulated values of particle mass concentration in zone 1 over time are shown in Fig. 6(b) . Since the rebound of particles was not considered in the simulation, which may occur in the experiment, the error between the simulated values and the experimental values was acceptable. Therefore, accuracy of Lagrangian discrete tracking model was verified. Where is the total mass of suspended particles and is the volume of each zone. The distribution of airflow in the isolation ward directly affects the aerosol migration path. Fig. 7 (a) At T=200 s, the bioaerosol has filled the whole room. Influenced by the direction of the initial airflow, the air was sent into the room, hit the ground vertically and then moved horizontally to both sides. Due to the interaction of the two airflow, the vortex was formed near the hospital bed eventually, which made a large amount of biological aerosols exhaled by the patient confined to the vicinity of the patient (400 s