key: cord-0934643-3dlvzew6 authors: Yan, Yihuan; Li, Xueren; Yang, Lin; Yan, Ping; Tu, Jiyuan title: Evaluation of cough-jet effects on the transport characteristics of respiratory-induced contaminants in airline passengers’ local environments date: 2020-08-16 journal: Build Environ DOI: 10.1016/j.buildenv.2020.107206 sha: 7fb7aee0aac77c493e2b760fd943c1e733cb2dd5 doc_id: 934643 cord_uid: 3dlvzew6 Urgent demands of assessing respiratory disease transmission in airliner cabins had awakened from the COVID-19 pandemics. This study numerically investigated the cough flow and its time-dependent jet-effects on the transport characteristics of respiratory-induced contaminants in passengers' local environments. Transient simulations were conducted in a three-row Boeing 737 cabin section, while respiratory contaminants (2 μm–1000 μm) were released by different passengers with and without coughing and were tracked by the Lagrangian approach. Outcomes revealed significant influences of cough-jets on passengers' local airflow field by breaking up the ascending passenger thermal plumes and inducing several local airflow recirculation in the front of passengers. Cough flow could be locked in the local environments (i.e. near and intermediate fields) of passengers. Results from comparative studies also revealed significant increases of residence times (up to 50%) and extended travel distances of contaminants up to 200 μm after considering cough flow, whereas contaminants travel displacements still remained similar. This was indicating more severe contaminate suspensions in passengers’ local environments. The cough-jets was found having long and effective impacts on contaminants transport up to 4 s, which was 8 times longer than the duration of cough and contaminants release process (0.5 s). Also, comparing to the ventilated flow, cough flow had considerable impacts to a much wider size range of contaminants (up to 200 μm) due to its strong jet-effects. 1 Global pandemics seem to become normalities in 2020. From the earlier 1918 Spanish flu to the latest 2 Coronavirus diseases 2019 , the types and frequencies of the infectious diseases 3 outbreaks have grown quickly and continuously been in the spotlight with equally familiar names as 4 SARS, Swine Flu (H1N1), Ebola, Middle East Respiratory Syndrome (MERS), Zika, etc. [1, 2] . The 5 most recent outbreaks of COVID-19, caused by severe acute respiratory syndrome coronavirus 2 6 (SARS-CoV-2), has resulted in near 8 million people being infected with over 430,000 death reported 7 worldwide [3] . Due to vastly increased imported cases of COVID-19 worldwide, many countries have 8 locked down their cities with compulsory restriction measures (e.g. social distancing) [4] , while 9 thousands of airlines have halted their international and domestic services [5] . It appears that our 10 collective vulnerability to the societal and economic impacts of the COVID-19 pandemic is inevitably 11 increasing. 12 While the number of COVID-19 cases is still growing, the understanding of the transmission of 13 COVID-19 virus also continues to improve [6] . The latest situation report from WHO revealed that 14 the SARS-CoV-2 virus was mainly transmitted via exhaled droplets or by touching contaminated 15 surfaces [6] . Human-to-human transmission of COVID-19 was summarised into three routes (Mittal 16 et al., 2020) : large expelled droplet directly acted on the recipients' mouth or nose; physical contact 17 with the deposited droplets and subsequently transferred to respiratory mucosa; and aerosolised 18 droplet nuclei from index patients' expiratory ejecta (i.e. aerosol transmission). Although the 19 transmission model of COVID-19 seems to be similar to SARS, its basic reproduction number (R 0 up 20 to 6.47) was found much higher than SARS (2.9) [7, 8] . More devastatingly, the virus could be 21 contagious during the asymptomatic incubation period [9] . Swab tests have detected higher viral load 22 on the upper airways than throats of test patients with symptoms, and the viral load from 23 asymptomatic carriers could be as high as that from symptomatic cases [10] . Since human respiratory 24 activities, such as coughing, speaking and sneezing could release significant amounts of droplets and 25 aerosols, these asymptomatic carriers could unknowingly escalate the spread of pathogens through 26 J o u r n a l P r e -p r o o f Although numerical approach could be more favorable to imitate the real contaminants release 1 process, it was still either completely neglected or simplified as constant release in many existing 2 studies [20, 24] . This was mainly because that modelling contaminants transport in multi-coupled and 3 multi-scale cabin domains requires significantly high computational cost and thereby the 4 instantaneous release details of contaminants were compromised. However, this overlooked factor 5 could be a vital gap when investigating respiratory disease transmission (e.g. COVID-19) in airliner 6 cabins. 7 When respiratory disease viruses (e.g. SARS-CoV-2) are released from the host in forms of droplets 8 or aerosols by coughing or sneezing, these respiratory activities may also have significant influences 9 on the local-environment around occupants and affect the transport characteristics of released droplets. 10 Existing studies found that cough in a confined space could cause turbulent jets with Reynolds 11 number of 10 4 and generate so-called 'cough-jet' immediately after release [25] . The strong cough-jet 12 with high-velocity airflow could induce instability in the local airflow interface and could even have a 13 in an occupant's breathing zone also captured significant interference of local airflow by coughing 15 and speaking jets [27] . The importance of jet flow from respiratory activities have been well 16 recognised through existing studies in traditional enclosed spaces, however, its significance in a more 17 extreme enclosed space, the airliner cabin, is still underrated. Existing studies by Gupta et al. [12] in 18 an airliner cabin reported that cough-jet should be one of the primary factors in the airborne disease 19 transmission since peak concentration of contaminants immediately occurred after release and were 20 dominated by cough-jet. Their study revealed strong necessity of including the instantaneous release 21 process of contaminants via respiratory activities. However, due to the limits of computational 22 capacity at early stage, the passenger models in their study were over-simplified using regular blocks. 23 It was a commonly applied approach to reduce computational cost, until recent studies found that 24 over-simplifying manikin models could induce dead flow zones in the vicinity of manikin and would 25 cause inaccurate airflow pattern and contaminants concentration predictions in the micro-environment 26 of occupants [28] . These drawbacks from over-simplification could be further enlarged in cabin 27 environments due to extremely large number of passengers sitting closely. Therefore, it is worth the 1 efforts of applying manikin models with real human body features when investigating the 2 instantaneous release process of airborne disease and its transmission in passengers' local 3 environment. Also, due to the high occupancy density and narrowed sitting space in airliner cabins, 4 each row of passengers sitting closely to each other could form their own enlarged local environments 5 and the entire cabin environment is mainly composed of many local environments. However, whether 6 the enlarged passengers' local environments could extend the duration of cough-jet effects or even 7 promote the transmission of airborne disease still remain unknown. These findings would be of great 8 interests to airliner manufacturers for new ventilation designs and optimisations. Therefore, it is 9 essential to understand the interactions between respiratory-induced jet flow and contaminants 10 transport in passengers' local environments. 11 Therefore, with the urgent need of assessing respiratory disease transmission in airliner cabins being 12 awakened from COVID-19 pandemics, this study carefully investigated one of the critical but long 13 overlooked factors, the instantaneous release process of respiratory-induced contaminants. Coughing, 14 one of the most common respiratory activities, was modelled in a three-row cabin of Boeing 737 to 15 study the impacts of cough-jet flow on the airflow field and contaminants transport in the local 16 environment of passengers. Cough contaminants were released with the jet flow from various 17 passengers at different locations, and the release processes were modelled using transient Eulerian-18 Lagrangian approach. Dominating parameters on contaminants transmission, such as the cough flow 19 field, droplets travel distance and size distributions were carefully studied. An economy cabin section was built based on one of the widely served medium-size commercial 23 aircrafts Boeing 737, as illustrated in Figure 1 . Since the focus of this study was on the local 24 environment of passengers (i.e. the region in the vicinity of seated passengers at the same row), a 25 three-row cabin section was found sufficient to capture the instantaneous developments of cough flow 1 and contaminants release process from passengers sitting in the second row. The distanced travel of 2 contaminants in a full-size economy cabin was not considered in this study. The ventilation inlets and 3 outlets were located on the side of cabin, as demonstrated in Figure 1 . Due to symmetric 3 by 3 seats 4 arrangement and ventilation design in the economy cabin, half of the cabin section including 9 5 computational thermal manikins (CTMs) and seats was modelled to save the computational cost. All 6 CTMs used in this study contain real human body proportions and detailed body features, since over-7 simplifying CTMs could lead to inaccurate predictions of human local environment [28]. These 8 CTMs were firstly obtained from 3D scans and were further optimised in our previous study [29] to 9 achieve well balanced computational cost and accuracy. The cabin domain and CTMs were discretised using unstructured tetrahedron mesh. Local mesh 13 refinement was applied in the front of sitting CTMs, while 10 prism layers with total height of 15mm 14 were created on each CTM's surfaces to capture gradient changes of local velocity, temperature, etc. The airflow field in the cabin domain was solved using the incompressible Navier-Stokes equations 4 with the Boussinesq approximation for the thermal buoyance flows. High-resolution advection 5 scheme was applied to achieve better robustness and accuracy of the advection terms, while the 6 SIMPLEC algorithm was employed for the velocity-pressure coupling. RNG k-ɛ model obtained high 7 reputation on predicting indoor airflows [32], although it migh overestimate contaminants deposition. 8 This study employed RNG k-ɛ model for modelling the turbulence in airflow field considering its 9 successful applications in the past studies [33] . The contaminants transport and their trajectories were 10 tracked using the Lagrangian framework. Significant forces including the drag force , the buoyance 11 force and the virtual mass force were considered and expressed in Equations (2) -(4). 12 In each eddy, the fluctuating eddy velocity is varied by the lifetime t e and the length L e of the eddy, as 3 expressed in Equations (7) and (8). 4 where C µ is the turbulent constant, k and ɛ are the local turbulent kinetic energy and dissipation, 7 respectively. 8 The ventilated flow rate inside the cabin was set according to the American Society of Heating, were expelled simultaneously with the cough-jet from each passenger's mouth. Since the size of 2 droplets released by cough dispersed diversely, a full-size range of contaminants (from 2 µm to 1000 3 µm) was included in this study. The contaminants size distribution and number fraction of each size 4 group were set according to existing experimental measurement from real coughs [25], as 5 demonstrated in Figure 2b . The evaporation process was not considered in this study. The Lagrangian 6 particle tracking model was employed to continuously trace the motions of contaminants, while 5,000 7 trajectories were found sufficient to achieve consistent contaminants concentration in this three-row 8 cabin domain. Detailed number of trajectories for each contaminant size was set accordingly to its 9 number fraction and was listed in Table 1 . Contaminants were assumed to be fully deposited when 10 hitting walls (floors, seats, cabin walls, etc.) due to the factor that real cabin materials are highly 11 absorptive (e.g. wool or nylon carpet, leather upholstery and fabric). (µm) 2 4 8 16 24 32 40 50 Number 51 291 974 1598 879 427 245 116 Diameter (µm) 75 100 125 150 200 250 500 1000 Number 143 83 46 38 34 29 31 15 Transient simulations were conducted with adaptive time steps to capture the simultaneous 2 development of cough flow and contaminants release process. Single cough behaviour was simulated 3 each time by applying one pulse air jet with duration of 0.5 s. Very small initial time-steps of 0.01 s 4 were used at the beginning of the cough process with maximum increase ratio of 1.1. The total 5 tracking time was 10 s, which took over 25 hours to finish each simulation using a workstation with 6 40 CPU cores (2.8G Hz Intel Xeon) and 128 GB RAM. Three cases with cough flow were studies, in 7 which Passenger A (window seat), B (in the middle) and C (aisle seat) at the second row was 8 coughing in each case, respectively. In addition, another three cases were simulated without 9 considering the cough flow, while contaminants were still released by same passengers via the mouth. 10 Contaminants in these cases were released with initial velocity of 1 m/s, while their release locations, 11 numbers and size distributions remained same as the cough cases. Therefore, a total of six cases were 12 presented and compared in this study as listed in Table 2 . 13 3 Results and discussion 15 Before considering the cough flow from passengers, the predicted airflow field was firstly compared The maximum velocity of ascending thermal plume (approx. 0.4 m/s) from passengers in cabin 7 environment was found slightly higher than traditional indoor spaces with single sitting person 8 (approx. 0.3m/s) [28]. This enlarged thermal plume was mainly induced by closely sat passengers 9 with merged heat loads and limited spaces above passengers. Although passenger's thermal plume 10 became stronger in cabin environment, it was still affected by the cough-jet after considering the 11 cough flow. Figure 4b showed the local airflow distribution immediately after cough at t = 0.6 s, in 12 which the ascending thermal plume was clearly interrupted by the cough flow above passenger A's 13 breathing region and led to a relatively weaker ascending thermal plume above passenger's head. Due 14 to the breakup of thermal plume and the strong effects of the cough-jet, local airflow recirculation was 15 observed in front of the sitting passenger. Since the airflow was relatively quiescent in passenger A's 1 local environment, the effects of the cough-jet were significant and was suspected to last much longer 2 after the completion of coughing behaviour. based on the employed CTM in this study was higher than the measured mouth opening during 4 coughs (2 ± 0.5 cm) [40, 41] . This might cause underestimations of the normalised distance (i.e. x/D). 5 As shown in Figure 5b , the cough-jet quickly passed the near-field in approximately 0.2 s even before 6 the cough release process was completed. However, the cough-jet was fluctuating at x/D around 10 to 7 15, which indicated that the cough-jet was unable to leave the passenger's local environment or 8 travelled further to the far-field. In this case when passenger A was the focus, this trapped cough-jet 9 could be due to the limited upper space above passenger A or the narrow space of each row in 10 economy cabins. Contaminants were released with the cough process during the first 0.5 second. The cough 3 contaminants transport characteristics were firstly compared among three cases (cases 1, 2 and 3), in 4 which each of passenger in the second row was coughing. 5 6 Figure 8 Contaminants trajectories after release by passengers at t = 1.5 s 7 Figure 8 showed the predicted trajectories of contaminants from four size groups (i.e. 0 -10 µm, 10 -8 50 µm, 50 -100 µm, and >100 µm). It was noticed that the transport of smaller contaminants (below 9 50 µm) was jointly dominated by the cough-jet and ventilated airflow. These small contaminants were 10 quickly ejected horizontally due to the strong jet effects of cough flow. Meanwhile, due to 11 J o u r n a l P r e -p r o o f insignificant weight of small contaminants, the vertical travel of these small contaminants heavily 1 relied on the local ventilated flow (i.e. the counter-clockwise air recirculation). As a result, small 2 contaminants released by passenger A was travelling with higher upward components, whereas same 3 groups of contaminants tended to travel further downward when passenger C was coughing. While the 4 cough-jet effect remained significant on contaminants sizing from 50 µm to 100 µm, the ventilated 5 flow started to loss its dominating effects on this size group due to increased weight and inertia of 6 larger contaminants. For large contaminants (> 100 µm), the cough-jet still had effect on the 7 contaminants transport although it was clearly weakened. On the other hand, the impacts from 8 ventilated flow was almost neglectable as contaminants were settling down significantly due to weight. With the effects of cough-jet on contaminants trajectories being clearly visualised, the travel time of 2 contaminants were further compared with and without considering the cough flow. Figure 9 plotted 3 the travel time of contaminants against their number fractions in each size group. It was clearly 4 observed through comparisons that the peak of number fractions from all three size groups were 5 clearly delayed after considering the cough flow, which indicated longer contaminants travel time. 6 Specifically, the cough-jet had strongest effects on extending travel time of contaminants below 10 7 µm, in which most of contaminants were travelling longer than 10 s with cough flow. A similar 8 pattern was also found on contaminants between 10 to 100 µm, in which the cough-jet stimulated 9 higher fraction of contaminants to travel longer than that without cough flow. Although the cough-jet 10 effects were found less significant to the large size group (> 100 µm) comparing to the other two 11 groups, the maximum travel time was still considerably increased nearly 50% after considering the 12 cough flow. It was also an interesting finding that contaminants released by passenger A revealed 13 different travel time pattern than those from passengers B and C in the cases without cough flow. This 14 was mainly due to the factor that the local airflow was relatively quiescent near the window seat 15 passenger and thereby small contaminants would suspend longer in the air. However, after 16 considering the cough flow, the jet effects from cough was more significant than the local ventilated 17 airflow and eventually led to a similar travel time pattern of contaminants released by different 18 The average residence time of each contaminant size in the air was further compared, as shown in 20 Figure 10 . For extremely small contaminants, the increase of residence time after considering cough 21 flow was not dramatic. This was because that the ventilated flow had already caused long residence 22 time of these contaminants due to their extremely small weight and inertia. However, for larger 23 contaminants (e.g. 50 µm, 75 µm and 100 µm) with considerably increased weight and inertia, the 24 influences from ventilated flow were very minimal, whilst the impacts from cough-jets became more 25 obvious and led to a significant increase of residence time on these contaminants with cough flow. 26 The cough-jet was found having significant impacts to contaminants up to 100 µm, these impacts 27 gradually reduced with the further increase of contaminant size and became minimal with 1 contaminants larger than 500 µm. Therefore, comparing the ventilated flow, the cough flow could 2 have considerable impacts to a larger size range of contaminants up to 200 µm. Also, the average 3 residence time of contaminants released by passenger A was higher than the other two cases, although 4 the travel time pattern became similar after considering the cough flow. On other words, contaminants 5 could suspend longer in the local environment of window seat passengers who had even less space 6 than middle and aisle passengers. To better understand the causes of longer residence time after considering the cough flow, the travel 3 motion characteristics of each contaminant size were compared with and without cough flow. Figure 4 11a illustrated the travel distance (D) versus displacement (d) before and after considering the cough 5 flow from passenger A. It can be clearly noticed that the travel displacement was almost the same 6 J o u r n a l P r e -p r o o f with and without the cough flow, which indicated that the contaminants did not travel further away 1 from passengers. However, by looking into the distance comparisons, significant increases of distance 2 were observed on contaminants up to 200 µm with cough flow. Also, the difference between the 3 displacement and distance was further enlarged because of the cough-jet, which indicated increases of 4 contaminants suspensions and recirculation in the local environment of passengers. The ratio of 5 distance (D) and displacement (d) at each contaminant size was then shown in Figure 11b . 6 Contaminants suspension was not significant without the cough flow, as the local environment of 7 passenger A was relatively quiescent. On the other hand, when contaminants were released with 8 cough flow, severe suspensions were found at contaminants up to 100 µm with travel distance more 9 than doubled than the displacement. Therefore, comparing to the ventilated flow, the cough-jets could 10 have considerable impacts on a much wider size range of contaminants (up to 200 µm). Also, the effective duration of the cough-jets could be essential on the transport characteristics of 4 cough contaminants. The above outcomes had revealed a stronger impact of cough-jets on 5 contaminants travel distances than that on contaminants travel displacements. Therefore, the ratio of 6 travel distance intervals to displacement intervals over time could be a clear indicator when the 7 cough-jets were having significant impacts on the contaminants field. Seven sizes of contaminants 8 from 4 µm to 500 µm were plotted in Figure 12 for comparisons. The ratios of travel distance 9 intervals to displacement intervals were sharply increased during the cough process (0.5 s) and 1.5 10 seconds after the completion of cough process. These dramatic ratio increases indicated a dominating 11 impact from strong cough flow on the transport characteristics of all sizes of cough contaminants. The 12 cough-jets effect was found eased in the next two seconds (up to t = 4 s) with the plotted ratios 13 gradually reduced. The ventilated flow took over and dominated flow after 4 seconds when the cough-14 jets were almost fully dispersed. Also, the results showed that small contaminants had higher ratio of 15 travel distance to displacement intervals than the large ones in the first 4 seconds, which revealed 16 more suspensions from small cough contaminants. Overall, the effective cough-jets could last at least 17 4 seconds in passengers' local environment and could affect the transport characteristics of 18 contaminants up to 150 µm during the entire period. The time-dependent cough flow and contaminants released from passengers sitting at the window seat, 4 in the middle and at the aisle seat in a medium-size economy cabin section were investigated in this 5 study. The effects of cough-jet in passengers' local environments were focused, while the cough-jets 6 impacts on local airflow field and instantaneous cough contaminants transport characteristics were 7 carefully evaluated. Outcomes from this study were concluded as follows: 8 (1) The cough flow could severely break up the ascending thermal plume of seated passengers, 9 even though the thermal plume became stronger in the cabin environment due to merged heat 10 loads from passengers sitting closely in the same row. Additional local airflow recirculation 11 was found in the front of sitting passengers due to the strong interferences of cough-jets to 12 ventilated flow and the breakup of thermal plume. Although the cough behaviour finished in 13 half second, the effects of cough-jets could last and remain influential for at least two seconds. 14 Due to the limited seating space in the economy cabin, the released cough flow could be 15 locked in the local environments (i.e. near and intermediate fields) of passengers. 16 (2) The cough flow was found having significant effects on the residence time of cough 1 contaminants by stimulating the contaminants travelling longer in the air. While ventilated 2 flow started to lose its dominating influences on intermediate-size contaminants (from 50 µm 3 to 100 µm) due to their considerable weight and inertia, the strong cough-jets still remained 4 significant impacts on these contaminants and led to a significant increase of their residence 5 time (up to near 50%). The cough-jets effects were gradually reduced on large-size 6 contaminants (up to 200 µm) and became very minimal on extremely large contaminants over 7 500 µm. 8 (4) The cough flow was found having long and effective impacts on contaminants transport up to 19 four seconds, which was 8 times longer than the duration of cough and contaminants release 20 (0.5 s). Also, comparing to the ventilated flow, cough flow had considerable impacts to a 21 much wider size range of contaminants (up to 200 µm) due to its strong jet-effects. Therefore, 22 when investigating infectious respiratory diseases (e.g. SARS-CoV-2) transmitted via airway 23 defensive behaviours (e.g. coughing, sneezing and speaking), it is essential to consider these 24 strong jet-effects during the release process. The 6 global impact of the coronavirus pandemic The Department of Health Australian, National Framework For Communicable Disease Control Australian Government Department of Health, Commonwealth of Australia COVID-19 Dashboard by the Center for Systems Science and 10 Effectiveness of the measures to flatten the epidemic 13 curve of COVID-19. The case of Spain Passenger Traffic during the COVID-19 Coronavirus Outbreak and its Socio-Economic Impact World Health Organisation, Novel Coronavirus(2019-nCoV) -Situation Report World Health 18 Novel coronavirus 2019-nCoV: 20 early estimation of epidemiological parameters and epidemic predictions The reproductive number of COVID-19 is 23 higher compared to SARS coronavirus Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Tie Song, Jianfeng He SARS-CoV-2 Viral Load in Upper Respiratory Specimens of Infected Patients Evaluation of airborne disease infection risks in an airliner cabin 10 using the Lagrangian-based Wells-Riley approach Transport of expiratory droplets in an aircraft cabin Epidemiologic and 15 clinical characteristics of 91 hospitalized patients with COVID-19 in Zhejiang, China: a retrospective, 16 multi-centre case series In-flight Transmission Cluster of COVID-19: A Retrospective Case Series Local and overall thermal comfort in an aircraft cabin 21 and their interrelations Experimental validation of local exhaust strategies for improved IAQ in 23 aircraft cabins Mitigation of cross-contamination in an aircraft cabin via localized 25 exhaust Global airflow field distribution in a cabin mock-up 27 measured via large-scale 2D-PIV Study on the carbon dioxide lockup phenomenon in aircraft cabin 1 by computational fluid dynamics Effects of passenger thermal plume on the transport and distribution 3 characteristics of airborne particles in an airliner cabin section Thermal environment around 6 passengers in an aircraft cabin Flow and contaminant 8 transport in an airliner cabin induced by a moving body: Model experiments and CFD predictions Effect of human movement on 11 airborne disease transmission in an airplane cabin: study using numerical modeling and quantitative 12 risk analysis Computational fluid 14 dynamics modeling of transport and deposition of pesticides in an aircraft cabin Violent expiratory events: on coughing 17 and sneezing Effect of artificial mucus properties on the 19 characteristics of airborne bioaerosol droplets generated during simulated coughing Study on the initial 22 velocity distribution of exhaled air from coughing and speaking Evaluation of manikin simplification methods for CFD 24 simulations in occupied indoor environments The simplification of computer simulated persons (CSPs) in CFD models of 26 occupied indoor spaces Perspective A Method for Uniform Reporting of Grid Refinement Studies Comparison of different k-ε models for indoor air flow computations Assessment of turbulence models and air 7 supply opening models for CFD modelling of airflow and gaseous contaminant distributions in 8 aircraft cabins Accurate 10 and high-resolution boundary conditions and flow fields in the first-class cabin of an MD-82 11 commercial airliner Convective and radiative heat transfer coefficients for 15 individual human body segments Experimental investigation of the 17 human convective boundary layer in a quiescent indoor environment Flow dynamics and characterization of a cough Reynolds number effects within the development region of a 21 turbulent round free jet Flow dynamics and characterization of a cough Human Cough as a Two-Stage Jet and Its Role in Particle Transport