National Academies Press: OpenBook

Research on the Transmission of Disease in Airports and on Aircraft (2010)

Chapter: SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft

« Previous: SESSION 2: Practical Case-Response Approaches to Investigating the Spread of Disease in Airports and on Aircraft
Page 15
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 15
Page 16
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 16
Page 17
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 17
Page 18
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 18
Page 19
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 19
Page 20
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 20
Page 21
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 21
Page 22
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 22
Page 23
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 23
Page 24
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 24
Page 25
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 25
Page 26
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 26
Page 27
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 27
Page 28
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 28
Page 29
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 29
Page 30
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 30
Page 31
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 31
Page 32
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 32
Page 33
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 33
Page 34
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 34
Page 35
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 35
Page 36
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 36
Page 37
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 37
Page 38
Suggested Citation:"SESSION 3: Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft." National Academies of Sciences, Engineering, and Medicine. 2010. Research on the Transmission of Disease in Airports and on Aircraft. Washington, DC: The National Academies Press. doi: 10.17226/22941.
×
Page 38

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

15 SESSIoN 3 Theoretical Modeling Approaches to Investigating the Spread of Disease in Airports and on Aircraft James S. Bennett, National Institute of Occupational Safety and Health (Presenter) Jennifer L. Topmiller, National Institute of Occupational Safety and Health Yuanhui Zhang, University of Illinois at Urbana–Champaign Watts L. Dietrich, National Institute of Occupational Safety and Health Qingyan (Yan) Chen, Purdue University (Presenter) Sagnik Mazumdar Michael W. Plesniak Stephane Poussou Paul E. Sojka Tengfei Zhang Zhao Zhang Byron Jones, Kansas State University (Presenter) Joan B. Rose, Michigan State University (Presenter) Mark H. Weir, Michigan State University summArizing exPosure PATTerns on commerciAl AircrAfT James S. Bennett (Presenter), Jennifer L. Topmiller, Yuanhui Zhang, and Watts L. Dietrich National Institute of occupational Safety and Health (NIoSH) research into the aircraft cabin environment began with a request from the fAA to study health effects among aircraft crew. A review of previous studies showed that female flight attendants may be at increased risk of adverse reproductive outcomes (1). Exposure assessments and epidemiologic studies in the areas of radiation and cabin air-quality studies followed (1–3). Difficulties in conducting studies in the passenger air- craft cabin environment during flight led to the decision that further work be done using realistic cabin mock-ups and computational fluid dynamics (CfD) to understand the behavior of any air contaminants present. The aircraft cabin environment is maintained during flight by the environmental control system (ECS). It is no small accomplishment to provide a safe atmosphere at cruise altitude—for example, 35,000 ft. In addition to pressurization, the ECS provides clean outside air to the cabin, which has a high-occupancy density compared with, for example, office buildings and classrooms. In newer aircraft, about 50% of the air supplied to the cabin has been recirculated and passed through a high- efficiency particulate air (HEPA) filter, with the remain- ing supply volume coming from the outside. The ECS is designed, as shown in figure 1, to use the length of the cabin as a plenum, so that air is supplied and exhausted at a velocity that is constant with respect to the length of the plane. Also, the direction of flow out of the supply and into the exhaust slots is in the seat row direction, perpendicular to the aisle. The movement of air between seat rows is thus minimized in the ECS design concept. While the airflow coming from the supply outlet can be considered two dimensional, the flow in the open space of the cabin is freer and somewhat turbulent, insofar as it is characterized by fluctuations in velocity (speed and direction). A flow can be deconstructed into its Reynold’s averaged velocity components: (1) where each instantaneous component, U(t), is the sum of a time average and a fluctuation with a time average of U t U u t( ) ( )= +

16 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT zero (4). Air contaminants, such as small droplets from an exhaled breath or a cough, are transported by the fluctuations, even though the average of the fluctuations is zero. The ECS, then, creates two competing processes, one that is intended and another that is perhaps impos- sible to avoid: (a) removal of potentially contaminated cabin air into the exhaust and replacement with clean air, and (b) movement of contaminants within cabin air by flow fluctuations. fluctuations are present, even in the hypothetical absence of obstructions, moving bodies, and thermal plumes. Airflow and contaminant transport research has taken place in collaboration with many expert partners (figure 2). The data generated by collaborations have been flow fields measured by experiments with realistic mock-ups or calculated by using CfD. The flow fields have con- sisted of velocity, turbulence parameters, and either gas or aerosol contaminant concentration. Airflow is a critical factor that influences air quality, disease transmission, and airborne contamination. FIGURE 1 Aircraft environmental control system design concept attempts to minimize the movement of air between seat rows. FIGURE 2 Aircraft Air Quality Partners: Sandia National Labs (SNL); University of Illinois (UI); Purdue University; Boeing Commercial Airplanes; Federal Aviation Administration (FAA); Kansas State University (KSU); University of Tennessee (UT); and American Society of Heating, Refrigerating, and Air Conditioning Engineers (ASHRAE).

17THEoRETICAL MoDELING APPRoACHES CfD simulations took place in collaboration with Boe- ing, Inc. (5, 6). At the University of Illinois, experiments in a five-row B767 mock-up delivered volumetric particle tracking velocimetry images of cabin flow seeded with helium bubbles and tracer gas (carbon dioxide) concen- tration fields generated by three source locations and three ventilation rates (7–9)., Sandia National Labs provided a massively parallel computing platform for the Boeing– NIoSH CfD simulations, including large eddy simulation. figure 3 provides snapshots of the Illinois, Boeing, and Sandia efforts. Sandia also provided advice and evaluation of the cabin airflow research and suggested that tracer gas experiments would be useful. Data for a real Boeing 747, including velocity and turbulence fields, were gathered by the University of Tennessee, at the fAA Aero-medical Research Institute. They also created detailed CfD simu- lations of the fluctuating cabin flow. NIoSH provided a review of the University of Tennessee report to the fAA. Kansas State University (KSU) was a pioneer in aircraft cabin research. KSU, along with Purdue University, has continued to advance the field in part through the fAA Center-of-Excellence for Aircraft Cabin Environmental (a) (c) (d) (b) ISO–surface for 1 measles/m^3 @ t - 1 sec FIGURE 3 (a) Boeing 767 mock-up at the University of Illinois; (b) large eddy simulation CFD model of a velocity field conducted by Boeing, NIOSH, and Sandia; (c) unstructured mesh for a Reynolds-Averaged Navier–Stokes (RANS) CFD model of a Boeing 767, conducted by Boeing; and (d) time evolution of an aerosol cloud from a point source, using a RANS CFD model of a Boeing 767. Research. KSU has a Boeing 767 mock-up with many seat rows and Purdue has done large-scale CfD simula- tions, including the wake effect of a moving body. Some collaborators, including KSU and Purdue, and NIoSH researchers were involved in research projects sponsored by the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) and development of an ASHRAE standard for aircraft cabin ventilation. Much work has been done, yet the role of ventila- tion in controlling disease transmission in aircraft cab- ins remains opaque. There is consensus that the issue is complex because of the many variables involved. figure 4 diagrams possible modes of transmission and variables discussed during the symposium. In an effort to pull immediately useful information from the detailed, high-quality studies done to date, a simple model and a modeling framework are presented here. The general aircraft-cabin air-contaminant transport effect (GAATE) model seeks to build exposure–spatial relationships between contaminant sources and recep- tors, quantify the uncertainty, and provide a platform for incorporating future studies. To put this model in context,

18 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT of the many variables presented in figure 4, the GAATE model involves only the three variables indicated by blue boxes. Thus, it provides exposure information. Knowledge of the infection risk to flight crews and passengers is needed to form a coherent response to an unfolding epidemic. An essential part of infection risk is exposure, and exposure may have an airborne com- ponent. The infection of flight attendants on Air China and Singapore Airlines with severe acute respiratory syn- drome (SARS) in 2003 is evidence of the risk faced by these workers, who in some situations find themselves in the role of first responders. Moreover, the Associa- tion of flight Attendants asked the fAA for protection from SARS. The goal of the GAATE model, then, is to provide useful information to authorities for addressing exposure incidents involving SARS, avian flu, H1N1, and other potentially lethal agents and to provide guid- ance to emergency response personnel. Methods The GAATE model can be thought of as a metamodel— that is, a model built from other models or studies. As such, the first step is solicitation of contaminant trans- port data for aircraft cabin environments from research partners. These data sets must be placed on a common footing and normalized to remove meaningless sources of variability. The large metadata set thus formed is ame- nable to statistical analysis. The model chosen currently is regression analysis, where the dependent variable is concentration gradient and the independent variable(s) describes location within the cabin. Variables that must be normalized are mass emission rate of the source and air change rate of the cabin. Put another way, the ratio of these two terms is held constant. In the current study, this normalization was achieved by dividing the measured concentration at a given seat loca- tion by a reference concentration (2) where CAVE is the spatial average concentration over all measurement locations and CS is the concentration measured nearest the source. As the cabin air is not well mixed, the inclusion of CS helps to make CREf more rep- resentative. The concentration variable used in the anal- Host infectivity Large particle Aerosol Contact FomiteNear air space Far air space Viability Behavior Dose Susceptibility Disease FIGURE 4 Aircraft cabin air quality research (blue high- light) in the context of disease pathways discussed at the symposium. C C CS REf AVE= + 2

19THEoRETICAL MoDELING APPRoACHES yses is then the ratio of the measured concentration to the reference concentration, CMEAS/CREf, (3) Thus far, the GAATE model has been applied to a data set from the University of Illinois. Measurements of carbon dioxide as a tracer gas were taken in a five-row Boeing 767 mock-up. Data were generated over three air change rates and three source locations, in which the measured outcome was the concentration at each of 35 seat locations. The concentrations measured at 2-s inter- vals were time-averaged over 1,000 s after the system had stabilized. No exhaust air was recirculated, and the gaspers were off. These data sets reflect an isothermal scenario. A CfD simulation was performed for the same set of conditions. These results were not included in the GAATE model, because they did not fit the same regres- sion equation as the experiments, which were considered more reliable. In principle, data generated by CfD are reasonable candidates. The regression equation had the following general form: (4) where Yi = observed quantity (contaminant or pathogen concentration); b0 and b1 = y intercept and slope of regression line, respectively; Xi = independent random variable; and ei = residual for the ith observation. C C C = MEAS REf FIGURE 5 Time slice of contaminant dispersion, source location, 2B: (a) measured and (b) simulated. (a) (b) Various functional forms were chosen to attempt a fit to the data by inspecting a plot of concentration versus distance from the source for all three source locations. Distinguishing between the seat letter coordinate direc- tion and the row number coordinate direction did not provide a better fit than using the simple variable of dis- tance, r. Results figure 5 shows the contaminant dispersion pattern at time T for both the experiment and the simulation. The concentration pattern in the experiment resembles iso- tropic diffusion, while in the simulation the pattern is formed more by directional convection. The specific form of Equation 4 that provided the best fit to the experimental tracer gas data was (5) The regression line shown in figure 6 has an intercept, b0, of 1.055 and a slope, b1, of 0.493. With an R 2 value of 0.476, it can be said that 47.6% of the variability in the concentration data is explained by the regression model. While the regression passed the normality test (P = .141), it failed the constant variance test, which is not surprising given that the concentration is more vari- able near the source. the analysis carries an uncertainty of 95%. this uncertainty applies in two different ways. b0 and b1 both have 95% confidence intervals (0.9906 # b0 # 1.1194 and 0.4204 # b1 # 0.5660), and these intervals are not independent, which is why the blue confidence bands in

20 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT figure 6 are curved. The red bands indicate uncertainty in prediction of the relation between C and ln(1/r) for any member of the population of r values. Put another way, the confidence band addresses the question of whether this regression line is the best one possible, while the pre- diction band addresses the value of this regression line as a predictive model. Because the concentration variability is greater nearer the source, a two-segment linear regression (figure 7) was also done to see if the fit could be improved. Both the slopes of the two lines and the breakpoint between them, r = 2.48 m, were determined in the regression. Thus, a physicality—the near-zone–far-zone distinction was identified by the statistical analysis. The freedom to adjust for this phenomenon increased the R2 value from 0.476 to 0.502, only a small improvement. Here also, the analysis passed the normality test (P = .375) but failed the constant variance test. The near source behavior is perhaps not well described by any kind of model based on the isotropic assumption. However, performing the regression on only the far-field data— >2.48 m from the source—actually lowered the R2 value. The benefit of more data points was apparently greater than the cost of the increased variance. ln (1/r)–2 –1 0 1 C on ce nt ra ti on –0.5 0.0 0.5 1.0 1.5 2.0 2.5 Regression line Tracer gas data 95% confidence band 95% prediction band e2 e 1 1/er (meters) FIGURE 6 Regression analysis of (source distance, concentration) data pairs, with 95% confidence and prediction bands. Radial distance (meters) 0 1 2 3 4 5 0.0 0.5 1.0 1.5 2.0 Regression line Tracer gas data Near field Far field 2.48 C on ce nt ra ti on FIGURE 7 Two-segment regression, with breakpoint between near and far fields.

21THEoRETICAL MoDELING APPRoACHES Discussion once a concentration–space relation is established, it can be applied in useful ways. With half the variability being explained by distance from the source, estimation using this simple model is widely applicable in the cabin environment, although the predictive power has quanti- fiable limitations. An interactive graphic tool was built using the idea that the relative exposure, taken here as the time average of normalized concentration, can be estimated for a source located anywhere in the Boeing 767 coach section. figure 8 shows this idea actualized with a Visual Basic program. By clicking on any seat in the cabin diagram, the exposure is calculated for the rest of the 10-row field. The figure is an example of the resul- tant field from one source location. An exposure map can be used to refine assumptions made about how far air contaminants such as small droplets travel in the cabin. Also, a case history and an exposure map may be used together to gauge infectivity by the airborne route. Moreover, if infectivity and expo- sure are both known, decisions about which passengers authorities should follow up with after a known expo- sure to a reportable disease are obvious. Conclusion The ability of the GAATE model to make a contribu- tion in such situations depends on its predictive power. Improvements in accuracy may come from inclusion of additional data sets. The scalability inherent in this approach paves the way to study additional aircraft types. Exposure to small droplets and postevapora- tion nuclei, even at a source distance of several rows, is readily apparent. The airborne pathway should then be considered part of the matrix of possible disease trans- mission modes in aircraft cabins, unless the pathogen has been proven nonviable in air. The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health. AdvAnce models for PredicTing conTAminAnTs And infecTious diseAse virus TrAnsPorT in The Airliner cAbin environmenT (PArT 1) Qingyan (Yan) Chen (Presenter), Sagnik Mazumdar, Michael W. Plesniak, Stephane Poussou, Paul E. Sojka, Tengfei Zhang, and Zhao Zhang In 2003, SARS affected more than 8,000 patients and caused 774 deaths in 26 countries across five continents within months after its emergence in rural China (10). A more recent disease, H1N1 A flu, affected about 40,000 patients across 76 countries within 1.5 months after its FIGURE 8 Example of use of the GAATE model interactive graphic: relative exposure to an air contaminant from a source in Seat 32B.

22 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT emergence (www.who.int/csr/disease/swineflu/updates/ en/index.html). These cases illustrate the dramatic role of globalization and air travel in the dissemination of an emerging infectious disease. other cases of airborne infectious diseases transmitted in airliners in recent years include tuberculosis, influenza, measles, and mumps. CfD is a very attractive tool to study the transmission of airborne contaminants in an airliner cabin as it is inex- pensive and flexible in changing thermofluid conditions inside the cabins compared with experimental measure- ments. The results presented here illustrate the potential of using CfD in modeling gaseous and particulate con- taminant transport inside airliner cabins. CfD was also used to model the SARS transmission case in Air China flight 112 from Hong Kong to Beijing in 2003 where a contagious passenger infected some 20 fellow passen- gers, as shown in figure 9 (11). Some seated as far as seven rows from the contagious passenger were infected. The movement of passengers and crew members may play a role in transmission. CFD Modeling The commercial CfD software fluent 6.2. (www.flu- ent.com) was used for the studies. The CfD model used a second-order upwind scheme and the SIMPLE algo- rithm. The renormalization group k-e model was used to simulate the turbulent flow inside the cabin mock-ups. Two different cabin geometries were used in this investigation to understand the effects of moving crew and passengers on contaminant transmission inside air- liner cabins. Initial CfD studies were done with a section of a four-row, twin-aisle cabin model as shown in figure 10a. The cabin section had 28 seats in four rows, repre- senting a section of economy-class cabin. The cabin was fully occupied. The air entered through linear diffusers at the ceiling level and was exhausted through outlets placed in the side walls close to the floor. The airflow rate in the cabin was 10 L/s per passenger. Box-shaped manikins were used to represent passengers. A moving person was modeled as a rectangular box of height 1.7 m and was assumed to move along the aisle. To investigate the effects of a moving person on contaminant trans- port in the cabin, two scenarios were considered: one in which the person walked continuously from the front to the rear end of the cabin without stopping and the other with intermittent stops of 5 s at each row. A second case used a 15-row, single-aisle cabin for studying SARS transmission in the flight from Hong Kong to Beijing in 2003 for Row 4 to 18 as shown in figure 9. figure 10b shows only one row of the cabin and the remaining rows are identical. The air entered the cabin through four linear diffusers: two placed at the ceiling above the aisle injected air downward and the other two at the side walls located below the storage bins injected air inward to the aisle. The total supply air- flow rate of 10 L/s per passenger was distributed equally among the four inlets. The air was exhausted through outlets on the side walls close to the floor. The conta- gious passenger sat in Row 11 of the 15-row cabin. Two contaminant release scenarios were considered: one with a pulsed release for 30 s and the other with a continuous release. The body moved along the aisle from the rear end of the cabin and stopped seven rows in front of the contagious passenger. The movement was simulated by using a combination of static and dynamic meshing schemes. for example, the computational domain of the four-row twin-aisle air- liner cabin was modeled using two separate geometries: a section for the aisle with the moving body and the other section for the rest of the cabin, as shown in figure 11. The meshes for the first section were dynamic; the remaining meshes were static. Hence, only 3.7% of the total meshes inside the domain were dynamic, which can reduce the computing costs for remeshing. The move- ment inside the 15-row, single-aisle model for the SARS transmission case was modeled similarly. FIGURE 9 A contagious passenger with SARS virus infected some 20 passengers on the flight from Hong Kong to Beijing in 2003 (11).

23THEoRETICAL MoDELING APPRoACHES CFD Modeling Results figure 12 shows the airflow pattern and airborne con- taminant concentration at 1 m above the cabin floor as the body moved continuously from the front to the rear end of the cabin. The results were for a contaminant released from Passenger 2A seated in the right window seat on the second row. The results at t = 0 s show the initial steady-state air velocity and contaminant distribu- tion before the body started moving. The airflow patterns illustrate that the flow disturbance created by the mov- ing person was rather local. The impact of movement on airflow on the left half of the cabin was minimal. The moving body created a low pressure zone behind it and hence air was induced from the sides. The moving body also pushed the air at its front. Hence, the body could carry the contaminant behind to the rear of the cabin. figure 13 shows the effect of an intermittently mov- ing body for the same contaminant source. The body stopped for 5 s in each row—that is, it stopped from 0.7 to 5.7 s in Row 2 and from 6.6 to 11.6 s in Row 3, which simulated a moving crew member who stopped at each row to provide service. The airflow pattern and contaminant concentration at 1 m above the cabin floor are shown at t = 0.7, 5.7, 6.6, and 11.6 s in the figure. The area near the contaminant source became heavily contaminated when the moving person stopped at Row 2, because it broke the near symmetric flow vortices at the cross section that aided in formation of the high-con- taminant-concentration zone. The intermittently moving body also enhanced the contaminant concentration level to passengers sitting near the aisle when it stopped at Row 3. When the moving person stopped, the highly contaminated air it carried at its back was pushed to the sides. Hence, the contaminant concentration can be higher than that with a continuously moving person. The results from the four-row, twin-aisle cabin show a significant impact of a moving person on contaminant transport. Thus, this investigation used the method to study why the SARS virus could be transported as far as seven rows away in the Air China 117 flight from Hong Kong to Beijing in 2003. figure 14 shows the contami- nant distribution at the breathing level in the Air China cabin for a pulse contaminant release from the infected passenger, such as a cough. The high-concentration zone (a) (b) FIGURE 10 Two different cabins used in the study: (a) section of four-row, twin-aisle cabin, and (b) one-row model of the 15-row, single-aisle cabin. FIGURE 11 Mesh layout of the four-row, twin-aisle cabin section.

24 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT was initially within two rows of the infected passenger, which appears to be in good agreement with common sense because the flow in the longitudinal direction should be small. When a person moved along the aisle, the wake could carry the contaminant to seven rows in front of the infected passenger, where the body stopped its movement. The contaminant carried in the wake was then distributed to the passengers seated near the aisle. A similar phenomenon was observed for the scenario with a continuous contaminant release. The CfD results showed that body movement may have caused the trans- mission of SARS pathogen from the infected passenger to fellow passengers seated as far as seven rows away on the Air China flight from Hong Kong to Beijing in 2003. Thus, CfD modeling appears to be a powerful and effective tool for predicting airborne contaminant trans- port in airliner cabins. Because CfD models use approxi- mations, the predictions should always be validated with high-quality experimental data. CFD Model Validation It is expensive and time-consuming to conduct experi- mental measurements of airborne contaminant concen- t = 0 s t = 0.7 s t = 1.6 s t = 2.5 s FIGURE 12 Velocity distribution and contaminant transport trends from the move- ment of a person in the four-row, twin-aisle cabin mock-up.

25THEoRETICAL MoDELING APPRoACHES trations in a full-scale airliner cabin with passengers. Hence, this study used a 1/10th-scaled, water-based experimental test facility consisting of an upside-down cabin mockup as shown in figure 15a. The cabin was made by a transparent semicircular pipe 45 cm in diam- eter and 2.44 m long. The mock-up, fully submerged in a water tank, was equivalent to a cabin with 28 rows of economy-class seats. The interior of the modeled cabin was empty so no seats and passengers were modeled. To simulate the ECS, water was injected through an overhead duct of the inlet diffuser assembly. To achieve a uniform inflow in the cabin, the water entered a set- tling chamber through 23 pipe fittings and was then supplied to the cabin through 48 elongated openings cut along the length, where a T-shaped diffuser diverted the fluid laterally to both sides of the cabin cross sec- tion. Water was extracted from two outlets located near the side walls of the cabin at floor level. To simu- late a moving person, an automated mechanism placed above the experimental facility traversed the moving body (0.02 m thick × 0.05 m wide 3 0.17 m tall) along the longitudinal direction of the cabin. Particle image velocimetry (PiV) was used to measure the velocity dis- tribution inside the water tank. The camera and laser were positioned to capture cross-sectional and longi- tudinal flow images. The corresponding CfD model t = 0.7 s t = 5.7 s t = 6.6 s t = 11.6 s FIGURE 13 Velocity distribution and contaminant transport trends from an intermit- tently moving person in the four-row, twin-aisle cabin mock-up.

26 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT was built for the water model as shown in figure 15b. The model was constructed to simulate as close to the experimental model as possible. Thus, the inlet started at the water supplying pipe to eliminate the difficulties in specifying inlet conditions in the cabin. figure 16a shows the measured mean flow fields at frames 4 and 7, which were acquired when the body moved 8.25 and 15.5 cm, respectively, past the laser sheet. A strong downwash in the wake of the moving body was observed, which is produced by the two sym- metric eddies around the top corners. As the two eddies approached the cabin floor, they spread to the sides and dissipated. The disturbance created by the moving body diminished very rapidly after this process. figure 16b shows the corresponding computed flow fields. Side- by-side comparison indicates that the CfD model was able to qualitatively predict the development of the two eddies. The predicted core size, flow pattern, and struc- ture are in reasonable agreement with the experimental values, although noticeable differences exist with respect to vortex aspect ratio. figure 17a shows only a small area of the measured flow due to the limited image size captured by the PiV. The comparison between the measured and computed velocity in the midsection along the longitudinal direc- tion in figure 17 shows reasonable agreement between the two results. flow recirculation due to flow separa- tion could be observed from the results. However, the t = 30 s t = 32.6 s t = 35.2 s t = 36 s Infected passenger FIGURE 14 Contaminant transport process from a person’s movement along the aisle with a pulse release of contaminant from the infected passenger in the single-aisle SARS transmission cabin mock-up.

27THEoRETICAL MoDELING APPRoACHES longitudinal flow computed behind the moving body is much stronger than that measured, with overprediction of longitudinal momentum transfer. This result may be due to less momentum transfer in lateral directions, resulting in vertically elongated eddy rings in the cabin cross section. overall, the CfD model can capture the fundamental flow mechanisms found in such a simu- lated cabin. Conclusions CfD, a powerful tool for predicting the transport of airborne contaminants in airliner cabins, shows that the movement of a person could have a significant effect. The movement of a person may have resulted in the spread of SARS virus to passengers seated far from the contagious passenger on Air China flight 112 from Hong Kong to (a) (b) Outlets T-shaped slots Traverse mechanism Body Cabin Overhead duct of inlet diffuser Inlet location Cabin Body FIGURE 15 (a) Small-scale experimental test facility of the cabin mock-up, and (b) CFD model of the test facility. Frame 4: Measured Frame 4: Computed Frame 7: Measured (a) Frame 7: Computed (b) FIGURE 16 (a) Measured and (b) computed mean flow fields at Frames 4 and 7 from movement inside the small-scale cabin mock-up.

28 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT Beijing in 2003. CfD results should always be validated with high-quality experimental data, as CfD models use many approximations. By using the measured velocity fields obtained from a small-scale, water cabin mock-up, CfD modeling can capture the fundamental flow fea- tures, although discrepancies exist between the measured and computed results. Acknowledgments This study was funded by the fAA office of Aerospace Medicine through the National Air Transportation Cen- ter of Excellence for Research in the Intermodal Trans- port Environment under a cooperative agreement. Although the FAA sponsored this project, it neither endorses nor rejects the findings of this research. This information is presented in the interest of invoking tech- nical community comment on the results and conclu- sions of the research. AdvAnce models for PredicTing conTAminAnTs And infecTious diseAse virus TrAnsPorT in The Airliner cAbin environmenT (PArT 2) Byron Jones (Presenter) The results from three aircraft cabin contaminant dis- persion studies are presented. These studies address dis- persion of gaseous contaminants, solid particles, and bacteria in an aerosolized liquid. All the studies were conducted in an aircraft cabin mock-up. Each study was conducted with somewhat different goals. However, all the studies are intended to support the development and validation of mathematical and computational models of dispersion in aircraft cabins. An attempt is made here to compare the data from the three studies. Cabin Mock-Up The aircraft cabin mock-up facility (figure 18) used in these studies is located in the Aircraft Cabin Environ- ment Research Laboratory at KSU. It is based on the geometry of a Boeing 767 but is intended to represent a midsize wide-body aircraft in general. The mock-up cabin is 9.45 m (31 ft.) long with 11 rows of seats. The seat spacing is 825 mm (32.5 in.) per row and the seats are seven across in a 2-3-2 configuration. The air inlet diffusers are from a Boeing 767 aircraft as is the air distribution system that supplies the diffusers. The air supply design for this aircraft consists of two linear slot diffusers extending the length of the cabin near the cen- ter ceiling of the cabin, each blowing outward. The inlet airflow is uniform along the length of the cabin. The uni- formity of this airflow was experimentally verified for both sides. Air exits the cabin through continuous floor- level exhausts on both sides of the cabin. The mock-up is equipped with coach seats from a Boeing 767 aircraft, and each seat is occupied by a thermal manikin with a heat output of 75 W. The manikins do not breathe or perspire. All inlet air is conditioned and passes through HEPA filters before entering the cabin. There is no recir- culation. The total airflow rate to the cabin was 660 L/s (1,400 ft3/min) for all data presented. (b) (a) FIGURE 17 Velocity in the midsection along the longitudinal direc- tion: (a) measurement and (b) computations.

29THEoRETICAL MoDELING APPRoACHES Description of Experiments The first set of experiments used carbon dioxide (Co2) tracer gas to measure contaminant dispersion. The Co2 tracer gas was mixed with helium to generate a mix- ture with a molecular weight equal to that of air. The tracer gas was at the same temperature as the cabin air when injected. As Co2 is much denser than air, negative plume buoyancy gives distorted results if these measures are not taken to ensure neutral buoyancy. Calculations and experimental results show that turbulent diffusion is several orders of magnitude greater than molecular diffusion, so the molecular diffusion is expected to be a negligible consideration in these experiments. The tracer gas was injected continuously at low velocity through a vertical tube in the center of the right or left aisle at a height of 1.2 m (48 in.) as shown in figure 18. The air was sampled through a seven-port sample tree that can be seen near the front of the cabin in figure 18. All measurements reported are at a height of 1.5 m (60 in.). Air was sampled from one port at a time for a minimum of 30 min before proceeding to the next port. once all ports were sampled, the entire tree was moved to the next location. The second set of measurements used talcum powder as a representative solid particle contaminant. The peak number density for this powder occurred at approxi- mately 1.5 µm and the data presented are for the total particle numbers between 0.5 and 5.0 µm. Injecting solid particles in a controlled manner without disrupting the cabin airflow is difficult. To accomplish this feat, a “puff generator” was developed. A measured amount of tal- cum powder was placed in a small cup. A small cop- per tube connected to a source of pressurized air was directed downward at the cup. The airflow was turned on and off quickly with a solenoid valve to generate a short but intense puff of air that aerosolized the talcum powder without generating a large airflow. figure 19 shows seven of the devices being tested simultaneously. for the experiments, the injection occurred in Row 2 and was done simultaneously at all seven seats in the row. Particle concentration was measured with a TSI 3321 aerodynamic particle sizer (APS) with the instrument placed in the seat as shown in figure 20. A straight tube was used to collect air samples at a height of 1.18 m (46.5 in.). Before the talcum powder was injected, the APSs were monitored to verify that the air was free of particles and the count rate was near zero. Data were then collected for 15 min after injection, at which time the counts had returned to near zero. The data reported here are the 15-min sums. The third set of measurements used aerosolized Lac- tococcus lactis as a surrogate bacteria. The bacteria were aerosolized by using a handheld mister (figure 21a) and the mist was released around head height of the seated “passengers.” Collection plates were located on top of the seat backs as shown in figure 21b. The collection plates were opened for 30 min for collection after the L. lactis was released. Additionally, air samples were taken at selected locations. The data presented here include only the collection plates. Controls were also run with no bacteria aerosolized to verify that near-zero counts FIGURE 18 Aircraft cabin mock-up. FIGURE 19 Solid particle injection. FIGURE 20 Solid particle measurements.

30 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT were obtained and thus all counts measured could be attributed to the aerosolized L. lactis. Longitudinal Dispersion These three sets of experiments were conducted with dif- ferent objectives, and now an attempt is made to com- pare the results from all three studies. figure 22 shows the seat and row numbers used to identify measurement locations. for the tracer gas measurements, the tracer gas was injected at Row 6 and measurements were made along the entire centerline. for the solid particle measurements, the particles were injected at Row 2. one APS was located in Seat 3D for all experiments and was used as a reference. A second APS was placed, in turn, in each of the D seats for Rows 4 to 11. for the bacteria measurements, the aero- solized bacteria were sprayed along the front of the cabin, generally in the Row 1 area. Measurements were taken at each seat but, for the purpose of this presentation, only the data for the three center seats are reported. figure 23 presents the tracer gas data. The data were deduced as follows: Cn = (cm – co)/(Vt /Vv) where Cn = normalized concentration at a location (nondi- mensional); cm = measured concentration at a location [parts per million (ppm)]; co = concentration in the ventilation air supplied to the cabin (ppm); Vt = volumetric flow of the tracer gas, Co2 only (L/s, ft3/min); and Vv = volume flow rate of ventilation air (L/s, ft 3/min). Data were collected at 178-mm (7-in.) intervals but are grouped by row for ease of comparison with the par- ticle and bacteria data. The results are asymmetric, with the tracer gas in the right aisle tending to go rearward and the tracer gas in the left aisle tending to go forward. A clear drop-off with distance along the centerline is observed in both cases. figure 24 presents the particle measurement results. Each data point represents a separate experiment, and for each data point shown the total number of counts at that location was divided by the total number of counts at the reference APS in Seat 3D. Thus, the value at Row 3 is automatically 1 but is not shown. The drop-off with dis- tance is similar to what was observed with the tracer gas. figure 25 shows the bacteria measurements results. Here the measurements for all three center seats for the row are averaged. The data are normalized based on the (a) (b) FIGURE 21 (a) Release of bacteria and (b) collection of bacteria. A B C D E F G 1 2 3 4 5 6 7 8 9 10 11 FIGURE 22 Row and seat identification for cabin mock-up.

31THEoRETICAL MoDELING APPRoACHES Row 3 data, similarly to the particle data. Again the drop- off with distance is similar to the tracer gas and particles. figure 26 presents all three sets of data on the same graph. The tracer gas data were divided into three groups identified as A, B, and C in figure 23. Groups A and B (right aisle) were normalized by dividing by the aver- age value at Row 7 and Group C was normalized by dividing by the average value at Row 5. Rows 5 and 7 then, effectively, became the equivalent of Row 3 for the particle and bacteria data sets and the drop-off with distance is plotted accordingly in figure 26. Although there is quite a bit of scatter, especially for the tracer gas results, quantitatively similar trends are observed for all three data sets. Bacteria values appear to start high but drop off more rapidly with distance than the particles and tracer gas. The particles may drop off more quickly than the tracer gas. Lateral Dispersion The injection for the tracer gas and for the solid particles is the same as for the longitudinal dispersion. Tracer gas measurements were made from side to side for Rows 5 to 9. for the particles, measurements were made only for Rows 4 and 7. for the bacteria, releases were made at Seats 6B, 6D, and 6f and measurements were collected at all seats. Because of the differences in the experimental setup for each study, it is not possible to directly com- pare lateral dispersion results for the three data sets as was done for longitudinal dispersion. The results for each are presented in turn. figures 27 and 28 present the tracer gas results. The peak concentration is offset from the injection location in the lateral and longitudinal directions. The rearward shift for the right aisle injection and the forward shift for the left aisle injection are evident. There is a clear drop-off with distance across the aircraft at a given row that is similar to what was observed in the longitudinal direction but no direct comparison is made. figure 29 presents the solid particle results. As the injection was uniform across Row 2, this experiment does not give a direct measure of lateral dispersion. The distri- bution at Row 4 is very nonuniform. Much of the lateral nonuniformity has disappeared by the time the particles get back to Row 7, which should not be surprising. 0 0.5 1 1.5 2 2.5 3 0 1 2 3 4 5 6 7 8 9 10 11 Row C on ce nt ra ti on Right aisle injection Left aisle injection A B C FIGURE 23 Tracer gas longitudinal dispersion data. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1 2 3 4 5 6 7 8 9 10 11 Row R el at iv e ex po su re FIGURE 24 Solid particle longitudinal dispersion data (Row 3 level normalized to 1).

32 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT figures 30, 31, and 32 present the bacteria results for the three different release locations. Counts above 400 are considered off scale for this method and are indicated as 400, as shown in Rows 4 and 5 in fig- ures 30 and 31 and in Rows 3, 4, and 5 in figure 32. Although there are some local peculiarities (e.g., Row 3 for the Row 6D release), in general the dispersion pat- terns are pretty much as expected for the Row 6B and 6f releases, and the drop-off in counts across the cabin is clear and reasonably consistent in the region of the release. There was a tendency for forward movement at all three release locations. It is far more pronounced for the left-side release than for the right-side release, which is consistent with the right–left differences found with the tracer gas. Discussion and Conclusions It was observed, particularly with the longitudinal data, that the various forms of contaminants behave similarly with respect to dispersion. The relative bacteria concen- trations appear to drop off more quickly with distance than those for the tracer gas and solid particles. There are at least two potential explanations. first, the bacteria may have a limited life span when airborne. only viable bacteria are counted. Thus, in addition to being removed by ventilation as they disperse through the cabin, some of them may become nonviable before they reach the more distant parts of the cabin. It is also possible the col- lection plates preferentially collect larger droplets as they are oriented vertically and would catch falling droplets. The larger droplets may settle out of the airflow before they reach the more distant parts of the cabin. Neverthe- less, these data combined give a reliable quantification of the far-field dispersion of contaminants and provide a basis for developing or validating dispersion models. The far field may be thought of as the region that is more than about two rows or seats in any direction from the point of release. The data also give some insight into the behavior in the near field (two seats or fewer from the point of FIGURE 25 Bacteria longitudinal data (referenced to Row 3; average of Seats C, D, and E). 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 1 2 3 4 5 6 7 8 9 10 11 Row R el at iv e ex p o su re Particles Bacteria Tracer gas A Tracer gas B Tracer gas C FIGURE 26 All longitudinal dispersion data combined (centerline dispersion). Bacteria Referenced to Row 3 0 0.2 0.4 0.6 0.8 1 1.2 1 2 3 4 5 6 7 8 9 10 11 Row R el at iv e ex po su re

33THEoRETICAL MoDELING APPRoACHES 0 1 2 3 4 5 6 7 –100 –80 –60 –40 –20 0 20 40 60 80 100 Location (inches) N or m al iz ed c on ce nt ra ti on Row 5 Row 6 Row 7 Row 8 Row 9 Seat center Injection A B C D E F G FIGURE 27 Tracer gas lateral dispersion data, right-aisle injection (Row 6). 0 1 2 3 4 5 6 7 –100 –80 –60 –40 –20 0 20 40 60 80 100 Location (inches) N or m al iz ed c on ce nt ra ti on Row 5 Row 7 Row 8 Seat center Injection A B C D E F G FIGURE 28 Tracer gas lateral dispersion data, left-aisle injection (Row 6). 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 –100 –80 –60 –40 –20 0 20 40 60 80 100 Distance from center (inches) R el at iv e ex p o su re Row 4 Row 7 A B C D E F G FIGURE 29 Solid particle lateral dispersion data (normalized to Seat 3D).

34 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT release). In this region, the dispersion is dominated by local airflow patterns and concentrations are dominated by plumes of high concentration from the source or plumes of low concentration from the supply air. Evi- dence of large, three-dimensional flow structures is evi- dent in all three data sets. Also, there is evidence that these structures are chaotic. for example, the tracer gas data in figure 23 have poor repeatability in the vicin- ity of the injection, but they have good repeatability at locations well removed from the injection. This chaotic nature makes it difficult to model and predict concentra- tions in the near-field region. Acknowledgments The following individuals conducted the research that generated the results presented: • Jeremy Beneke, graduate student; • Jie Chen, graduate student; • Helmut Hirt, assistant professor of biology; • mohammad Hosni, professor of mechanical engi- neering; • Francis Noonan, laboratory manager; and • P. J. A. Priyadarshana, postdoctoral researcher. 0 50 100 150 200 250 300 350 400 1 2 3 4 5 6 7 8 9 10 11 Row C ou nt s Seat A Seat B Seat C Seat D Seat E Seat F Seat G FIGURE 30 Bacteria lateral dispersion data, Seat 6D release. 0 50 100 150 200 250 300 350 400 1 2 3 4 5 6 7 8 9 10 11 Row C ou nt s Seat A Seat B Seat C Seat D Seat E Seat F Seat G FIGURE 31 Bacteria lateral dispersion data, Seat 6B release.

35THEoRETICAL MoDELING APPRoACHES The results presented are from research funded, in part, by the fAA office of Aerospace Medicine through the National Air Transportation Center of Excellence for Research in the Intermodal Transport Environment under a cooperative agreement. The research was also funded, in part, by the KSU Targeted Excellence Program. Although the FAA sponsored this project, it neither endorses nor rejects the findings of this research. This information is presented in the interest of invoking tech- nical community comment on the results and conclu- sions of the research. chArAcTerizing The risk of Tuberculosis infecTion in commerciAl AircrAfT by using QuAnTiTATive microbiAl risk AssessmenT Joan B. Rose (Presenter) and Mark H. Weir on May 12, 2007, a man infected with multidrug- resistant Mycobacterium tuberculosis (XDR-TB) trav- eled from Atlanta, Georgia, to Paris, france, and then from Prague, Czech Republic, to Montreal, Canada, on May 24, 2007. These flights lasted about 8 h each. The Centers for Disease Control and Prevention (CDC) attempted to address the risk of infection for the approx- imately 80 people who sat in the five rows surrounding the infected man during the flights. TB transmission to nearby passengers during a flight to Hawaii in 1994 had been previously reported (12); it has been acknowledged by the CDC that this risk is low (no estimate of how low has been given), but the consequence of infections could be high because of the rare, drug-resistant type of TB the patient had. The combination of this individual’s travel, global transmission of SARS, and now the potential for trans- mission via various new types of influenza strains (bird and A/H1N1) has reignited concern about the likelihood of disease transmission in commercial aircraft and the scientific uncertainties in addressing the risk. Several issues and questions are associated with transmission of disease during air travel: 1. Widespread geographic movement of infected indi- viduals to new communities, 2. Spread of disease to fellow passengers during the flight, 3. The types of pathogens primarily associated with disease transmission on airplanes, 4. Understanding the level of risk associated with air flight, and 5. Implementing sound and effective policies to pre- vent disease transmission during air travel. Quantitative microbial risk assessment (QMRA) is an approach that can be used to address these issues. QMRA as an integrated science is expanding and follow- ing a National Academies’ approach (13) (figure 33). It is now possible to address the hazards and model expo- sures and, via a dose–response relationship, characterize the risk to a greater confidence level than was previously possible (14). The bounds of transmission for specific hazards have been in place, but the traditional understanding of trans- mission must be reexamined. Respiratory pathogens are transmitted by coughing or sneezing and enteric 0 50 100 150 200 250 300 350 400 1 2 3 4 5 6 7 8 9 10 11 Row C ou nt s Seat A Seat B Seat C Seat D Seat E Seat F Seat G FIGURE 32 Bacteria lateral dispersion data, Seat 6F release.

36 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT pathogens are transmitted by the fecal–oral route, but it is important to acknowledge that these pathogens may also be transmitted via fomites and contaminated hands. In addition, the role of the contaminated environment needs to be further explored. Infectious disease hazards enter an airplane environ- ment through three major sources: 1. Infected individuals, 2. Contaminated food, and 3 Contaminated water. It is known, through at least one small study that investigated sewage from airplanes, that infected passen- gers are traveling (15). This study found enteric viruses in 45% of the sewage on international flights. Water on airplanes remains a potential source of pathogen risk. In the United States alone, there are 63 air carriers and 7,327 aircraft public water systems that will need to be monitored and potentially treated. The use of bottled water has reduced the risk from aircraft water systems, but safety depends then on the efficiency of bottling facilities and water safety programs in place in various countries. food-borne disease onboard air- craft is being addressed through hazard analysis and crit- ical control point programs for companies that prepare airline food. However, infected passengers, particularly those who are still excreting pathogens but may not have any major symptoms, are difficult to assess and control. QMRA for TB Transmission on Airplanes Jones and colleagues used the QMRA approach to address the TB scenario; a number of research needs and knowl- edge gaps needed to be addressed (16). The assumption of equal risk throughout the airliner cabin has been debunked; the transport was shown to be modeled by using Markov chains rather than the more complex and time-consuming CfD. This situation also showed the power of having a known dose–response model, which can give a risk level specific to XDR-TB. These advance- ments have allowed for a more complete picture of the risks to passengers and cabin crew members. Generally, risks ranged from 1/10,000 to 1/100,000 on flights in the United States but over an 8-h flight risk, estimates were 0.104 infection per 169 susceptible passengers, which is equivalent to 6.2 infections per 10,000 exposed suscep- tible persons. The main drivers of the risk were shown to be cough- ing rate and active infection and numbers of bacilli. Current understanding of the excretion of pathogens needs to be expanded, whereby much of the quantita- tive information typically has been gathered from peer- reviewed articles from the 1950s and 1960s. further advancements in molecular methods may allow for more accurate determination of excretion rates. This informa- tion is very important to the risk assessment, as the level of pathogens to which a person is exposed depends on the amount of pathogen excreted from the infected indi- vidual. These levels can vary based on the medium being measured, such as sputum compared with saliva. There- fore, advancements in determining the level of secretion should include all fluids, which would be a concern for the exposure route, such as sputum, saliva, and coughing and sneezing overall for the respiratory route. The current model, which has been used to model air- flow indoors, is simplistic in important areas; even the more sophisticated CfD simulations keep all the peo- ple in the cabin stationary. No movement is allowed in the simulations, which can alter the results of the final destination of the pathogens. Also, most of the models that attempt to model the transport of pathogens and particulate matter in airliner cabins are based on indoor air models. Yet airliner cabins more closely resemble confined spaces than the traditional indoor air environ- ments such as office building rooms and other rooms in buildings. A greater concern is the movement of cabin crew and passengers. This oversight can be critical to understanding the true risks posed by an infected passen- ger due to air current movements, which will be shifted from the baseline (which would be no one moving about in the cabin). The other concern with modeling the indoor air envi- ronment is the viability question. Current indoor air models and those designed specifically for the interior cabin do not allow for including whether the pathogen is viable. This question is important, especially if the transport model is going to assist in designing sampling strategies. The viability question does not likely have a straightforward answer. This problem may go beyond the natural environmental survival of the pathogen and may be a function of the transport and means by Hazard ID Characterization Management ExposureDose response FIGURE 33 Microbial risk assessment framework.

37THEoRETICAL MoDELING APPRoACHES which the pathogen has been introduced to the indoor environment. Therefore, the viability questions should be answered by considering the scenario as well as the location where the transport will take place. This situ- ation will allow for better overall risk estimates as well as risk-based surface sampling strategies after release of the pathogen. The final research need recognized from this scenario surrounded the surface sampling and decontamination plans. These two concerns are connected. A primary concern is decontamination of the interior of an airliner. Ideally, the decontaminant would not pose a risk to damaging the electronics or structure of the airliner and cabin. Some decontaminants, such as chlorine dioxide, are aggressive oxidizers, which makes them good dis- infectants but also damages multiple surfaces; residuals are not possible due to the human health risk as well as the chemistry of the disinfectant. In the case of TB, even XDR attachment of the pathogen to fomites is not a major concern as hand-to-self transmission is not known to occur. However, this issue is greater for other patho- gens such as norovirus and methicillin-resistant Staphy- lococcus aureus. The latter issue leads to the issue of surface sampling, which raises concerns about the target that should be monitored for (indicators versus pathogens) in addition to methods and detection limits. Sampling strategies should be established to determine whether the decon- tamination scheme has worked, the level of pathogens remaining is acceptable, and the risk posed is accept- able. Sampling is typically done by swabbing surfaces and using a rapid detection method such as a molecular tool. These methods are specific and rapid but would have to be tailored to the environment and the pathogen of interest. Research to Inform Risk for the Airline Industry There are two major research programs that would assist in building an improved understanding of the risks of disease transmission on aircraft. • Characterization of hazards associated with travel- ing on airplanes: Surveillance of sewage, water, and key fomites (touched and nontouched) for contamination (using Escherichia coli and pathogen-specific quantita- tive polymerase chain reaction) on airplanes would allow one to examine quantitatively the numbers of passengers infected (sewage assessment) and evaluate water and sur- faces addressing the key exposure pathways. This method would assist in developing adequate monitoring policies (of people, food, water, and surfaces) and disinfection. • Assessment of risk and integration of air trans- port models with QMRA: While there are sophisticated particle—and in some cases microbial—transport mod- els being developed in aircraft, they cannot adequately address risk as hazard-specific survival and dose–response have not been integrated with the partial assessment of exposure. Use of a QMRA framework would allow one to examine quantitative risks with a yardstick that would put the disease transmission during air travel into perspective (1/10,000, which has been deemed accept- able for drinking water). Acknowledgments This work was supported by the Center for Advanc- ing Microbial Risk Assessment (www.camra.msu.edu/), funded by the U.S. Environmental Protection Agency Sci- ence to Achieve Results Program, and the U.S. Depart- ment of Homeland Security University Programs. Investigators Joan B. Rose, Michigan State University, rosejo@msu. edu; Syed Hashsham, hashsham@msu.edu; Charles N. Haas, Drexel University, haas@drexel.edu; Patrick Gurian, pgurian@drexel.edu; Rosina Weber, rweber@ cis.drexel.edu; Charles P. Gerba, University of Arizona, gerba@ag.arizona.edu; Chris Choi, cchoi@arizona.edu; Joseph N.S. Eisenberg, University of Michigan, jnse@ umich.edu and James Koopman, jkoopman@umich.edu; Mark Nicas, University of California, Berkeley, mnicas@ berkeley.edu; and David Wagner, dave.wagner@nau.edu. Institutions Michigan State University (lead); Carnegie Mellon Uni- versity; Drexel University; Northern Arizona University; University of Arizona; University of California, Berkeley; and University of Michigan. References Waters, M., T. f. Bloom, and B. Grajewski. The NIoSH/1. fAA Working Women’s Health Study: Evaluation of the Cosmic-Radiation Exposures of flight Attendants. Health Physics, Vol. 79, No. 5, 2000, pp. 553–559. Grajewski, B., M. Waters, E. A. Whelan, and T. f. Bloom. 2. Radiation Dose Estimation for Epidemiologic Studies of flight Attendants. American Journal of Industrial Medi- cine, Vol. 41, 2002, pp. 27–37. Whelan, E. A., B. Grajewski, E. Wood, L. Kwan, M. 3. Nguyen, T. M. Schnorr, E. A. Knecht, and J. S. Kes- ner. feasibility Issues in Reproductive Biomonitoring of

38 RESEARCH oN THE TRANSMISSIoN of DISEASE IN AIRPoRTS AND oN AIRCRAfT female flight Attendants and Teachers. Journal of Occu- pational and Environmental Medicine, Vol. 44, 2002, pp. 947–955. Hinze, J. o. 4. Turbulence, 2nd ed. McGraw-Hill, New York, 1975. Lin, C., R. H. Horstman, M. f. Ahlers, L. M. Sedgwick, 5. K. H. Dunn, J. L. Topmiller, J. S. Bennett, and S. Wirogo. Numerical Simulation of Airflow and Airborne Pathogen Transport in Aircraft Cabins. Part 1: Numerical Simula- tion of the flow field. ASHRAE Transactions, Vol. 111, Part I, 2005, pp. 755–763. Lin, C., R. H. Horstman, M. f. Ahlers, L. M. Sedgwick, 6. K. H. Dunn, J. L. Topmiller, J. S. Bennett, and S. Wirogo. Numerical Simulation of Airflow and Airborne Pathogen Transport in Aircraft Cabins. Part 2: Numerical Simula- tion of Airborne Pathogen Transport. ASHRAE Transac- tions, Vol. 111, Part i, 2005, pp. 764–768. Sun, Y., Y. Zhang, A. Wang, J. L. Topmiller, and J. S. 7. Bennett. Experimental Characterization of Airflows in Aircraft Cabins. Part I: Experimental System and Mea- surement Procedure. ASHRAE Transactions, Vol. 111, Part II, 2005, pp. 45–52. Wang, A., J. S. Bennett, Y. Zhang, K. H. Dunn, and J. 8. L. Topmiller: Tracer Study of Airborne Disease Transmis- sion in an Aircraft Cabin Mock-Up. ASHRAE Transac- tions, Vol. 112, Part ii, 2006, pp. 697–705. Zhang, Y., Y. Sun, A. Wang, J. L. Topmiller, and J. S. 9. Bennett. Experimental Characterization of Airflows in Aircraft Cabins. Part II: Results and Research Recommen- dations. ASHRAE Transactions, Vol. 111, Part ii, 2005, pp. 53–59. Joseph, S. M. P., Y. Y. Kwok, D. M. E. o. Albert, and 10. S. Klaus. The Severe Acute Respiratory Syndrome. New England Journal of Medicine, Vol. 349, No. 25, 2003, pp. 2431–2441. 11. olsen, S. J., H. L. Chang, T. Y. Cheung, A. f. Tang, T. L. fisk, S. P. ooi, H. W. Kuo, D. D. Jiang, K. T. Chen, J. Lando, K. H. Hsu, T. J. Chen, and S. f. Dowell. Transmis- sion of the Severe Acute Respiratory Syndrome on Air- craft. New England Journal of Medicine, Vol. 349, No. 25, 2003, pp. 2416–2422. Kenyon, t. A., S. E. Valway, w. w. ihle, i. m. onorato, 12. and K. G. Castro. Transmission of Multi-drug Resistant Mycobacterium tuberculosis During a Long Airplane flight. New England Journal of Medicine, Vol. 334, No. 15, 1996, pp. 933–938. National Research Council, Committee on Improving Risk 13. Analysis Approaches. Science and Decisions: Advancing Risk Assessment. National Academies Press, Washington, D.C., 2009. Haas, C. H., J. B. Rose, and C. P. Gerba (eds.) 14. Quanti- tative Microbial Risk Assessment. John Wiley and Sons, New York, 1999. Shieh, Y.-S. C, R. S. Baric, and M. D. Sobsey. Detection 15. of low levels of Enteric Viruses in metropolitan and Air- plane Sewage. Applied and Environmental Microbiology, Vol. 63, No. 1, 1997, pp. 4401–4407. Jones, R. M., Y. Masago, T. Bartrand, C. N. Haas, M. 16. Nicas, and J. B. Rose. Characterizing the Risk of Infection from Mycobacterium tuberculosis in Commercial Passen- ger Aircraft Using Quantitative Microbial Risk Assess- ment. Risk Analysis, Vol. 29, No. 3, 2009, pp. 355–365.

Next: SESSION 4: Experimental Bench Science Approaches to Investigating the Spread of Disease in Airports and on Aircraft »
Research on the Transmission of Disease in Airports and on Aircraft Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB Conference Proceedings 47: Research on the Transmission of Disease in Airports and on Aircraft is the summary of a September 2009 symposium. The symposium examined the status of research on or related to the transmission of disease on aircraft and in airports, and the potential application of research results to the development of protocols and standards for managing communicable disease incidents in an aviation setting. The symposium also explored areas where additional research may be needed.

An article on this report was included in the January-February issue of TR News.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!