DRAFT NEW RESEARCH IDENTIFIES A METHODOLOGY FOR MINIMIZING RISK FROM AIRBORNE ORGANISMS IN HOSPITAL ISOLATION ROOMS Farhad Memarzadeh, Ph.D., PE. National Institutes of Health November 5, 1999 1. INTRODUCTION The patients in hospital isolation rooms constantly produce transmissible airborne organisms by coughing, sneezing or talking, which, if not under control, results in spreading airborne infection. Tuberculosis (TB) infection, for example, occurs after inhalation of a sufficient number of tubercle bacilli expelled during a cough by a patient [1]. The contagion depends on the rate at which bacilli are discharged, i.e., the number of the bacilli released from the infectious source. It also depends on the virulence of the bacilli as well as external factors, such as the ventilation flow rate. In order to prevent the transmission of airborne infection, the isolation rooms are usually equipped with high efficiency ventilation systems operating at high supply flow rate to remove the airborne bacteria from the rooms. Research has shown, however, ventilation rate is no guarantee of good control to the spreading airborne infection. Another means of minimizing the risk from airborne bacteria is to apply ultraviolet germicidal irradiation (UVGI). UVGI holds promise of greatly lowering the concentration of airborne bacteria and thus controlling the spread of airborne infection among occupants. UVGI achieves this through the drying and evaporation of the droplets carrying the bacteria colony. Comparing the clearance level of airborne bacteria achieved by increased ventilation flow rate or by applying UVGI may be useful in evaluating the efficiency of UVGI. The most widely used application of UVGI is in the form of passive upper-room fixtures containing UVGI lamps that irradiate a horizontal layer of airspace above the occupied zone. They are designed to kill bacteria that enter the upper irradiated zone, and are highly reliant on vertical room air currents. The survival probability of bacteria after being exposed to UVGI depends on the UV irradiance as well as the exposure time in a general form [1]: DRAFT 1 % Survival 100 e kIt (1) Where I t k = UV irradiance, μW/cm2 = time of UV exposure = the microbe susceptibility factor, cm2/μW.s Increasing room air mixing enhances upper-room UVGI effectiveness by bringing more bacteria into the UV zone. However, rapid vertical air circulation also implies insufficient exposure time. It can be understood that removing and killing bacteria in isolation rooms are greatly influenced by the flow pattern of ventilation air through parameters, such as: Ventilation flow rate Locations of air supplies/exhausts Supply air temperature Location of the UV fixture(s) Room configuration Susceptibility of the particular species of bacteria In order to achieve a better performance of UVGI as well as a higher removal effectiveness of the ventilation system, the airflow pattern needs to be fully understood and well organized. Therefore, it is necessary to conduct a system study on minimizing the risk from airborne organisms in hospital isolation rooms with all the important parameters being analyzed. Previous research has been almost entirely based on empirical methods [2-4], which are time consuming and are limited by the cost of modifying physical installations of the ventilation systems. The absence of U-V treatment systems also imposed limitations on previous research. Therefore, the design guidance for isolation rooms basically relied on gross simplifications without fully understanding the effect of the complex interaction of room airflow and U-V treatment systems. Computational Fluid Dynamics, CFD, (sometimes known as airflow modeling) has been proven to be very powerful and efficient in research projects involving parametric study on room airflow and contaminant dispersion [5-7]. In addition, the output of the CFD simulation can be presented in many ways, for example with the useful details of field distributions, as well as overviews on the effects of parameters involved. Therefore, CFD is employed as a main approach in this study [8]. The results of this study are also intended to be linked to a concurrent study into thermal comfort, uniformity and ventilation effectiveness in patient rooms. While the patient room is not exactly the same in terms of dimensions, the two studies share enough common features, for example, there is a single bed in the room, the glazing features are similar in each case, there are similar amounts of furniture in the room, etc, that the DRAFT 2 conclusions drawn from the patient room study will be viable in this study, and vice versa. 2. PURPOSE OF THIS STUDY The main purpose of this study presented in this paper is to: Use airflow modeling to evaluate the effects of the parameters such as: - Ventilation flow rate - Supply temperature - Exhaust location - Baseboard heating (in winter scenarios) on minimizing the risk from airborne organism in isolation rooms. Assess the effectiveness of: - Removing bacteria via the ventilation system, either through the particles sticking to the wall, or through ventilation through exhaust grilles - Killing bacteria with U-V. Provide an architectural / engineering tool for good design practice that is generally applicable to conventional isolation room use. 3. METHODOLOGY 3.1 Airflow Modeling Airflow modeling based on Computational Fluid Dynamics (CFD) [8] which solves the fundamental conservation equations for mass momentum and energy in the form of the Navier-Stokes Equations is now well established: ( ) div ( V grad ) S t Transient + Convection Diffusion = Source Where : V S DRAFT (2) = density = velocity vector = dependent variable = exchange coefficient (laminar + turbulent) = source or sink term 3 3.1.1 How is it done? Airflow modeling solves the set of Navier Stokes equations by superimposing a grid of many tens or even hundreds of thousands of cells which describe the physical geometry heat and contamination sources and air itself. Figures 1 and 2 below, show a typical research laboratory and the corresponding space discretization, subdividing the laboratory into tens or hundreds of thousands of cells. Figure 1. Geometric model of a laboratory Figure 2. Superimposed grid of cells for calculation The simultaneous equations thus formed are solved iteratively for each one of these cells, to produce a solution that satisfies the conservation laws for mass momentum and energy. As a result the flow can then be traced in any part of the room simultaneously coloring the air according to another parameter such as temperature. 3.1.2 Validation of Airflow Modeling Methodology The methodology for this research was extensively used in a previous publication by Memarzadeh [9], which considered ventilation design handbook on animal research facilities using static microisolators. In order to analyze the ventilation performance of different settings, numerical methods based on computational fluid dynamics were used to create computer simulations of more than 160 different room configurations. The performance of this approach was successfully verified by comparison with an extensive set of experimental measurements. A total of 12.9 million experimental data values were collected to CONFIRM the methodology. The average error between the experimental and computational values were 14.36 percent for temperature and velocities, while the equivalent value for concentrations was 14.50 percent. To forward the above mentioned research, several progress meetings were held to solicit project input and feedback from the participants. There were more than 55 international experts in all facets of the animal care and use community, including scientists, veterinarians, engineers, animal facility DRAFT 4 managers, and cage and rack manufacturers. The pre-publication project report underwent peer review by a ten (10) member panel from the participant group, selected for their expertise in pertinent areas. Their comments were adopted and incorporated in the final report. The publication "NIH ventilation design handbook on animal research facilities using static microisolators" was also reviewed by the American Society of Heating, Refrigeration, and Air Conditioning Engineers (ASHRAE) technical committee and accepted for inclusion in their 1999 Application handbook. 3.2 Representation of Bacteria Droplets 3.2.1 Basic Concept The basic assumption in this study is that the droplet carrying the bacteria colony can be simulated as particles being generated from several sources surrounding the occupant. These particles then are tracked for a certain period of time in the room. The evaporation experienced by the droplet is not simulated in this study. The dose that the particles received when travelling in the room along their trajectories is the summation of the dose received at each time-step calculated as: Dose = ∑ (dt*I) (3) Where, dt is the time step and I is the local UV irradiance. Then, based on the dose received, the survival probability of each particle is evaluated. Since the airflow in a ventilated room is turbulent, the bacteria from coughing or sneezing of the occupants in the room are transported not only by convection of the airflow but also by the turbulent diffusion. The bacteria are light enough, and in small enough quantities, that they can be considered not to exert an influence on airflow. Therefore, from the output of the CFD simulation, the distributions of air velocities and the turbulent parameters can be directly applied to predict the path of the airborne bacteria in convection and diffusion process. 3.2.2 Particle trajectories The methodology for predicting turbulent particle dispersion used in this study was originally laid out by Gosman and Ioannides [10], and validated by Ormancey and Martinon [11], Shuen et al [12], Chen and Crowe [13]. Experimental validation data was obtained from Snyder and Lumley [14]. Turbulence was incorporated into the Stochastic model via the - turbulence model [15]. The particle trajectories are obtained by integrating the equation of motion in 3 coordinates. Assuming that body forces are negligible with the exception to drag and gravity, these equations can be expressed as, mp du p dt 1 C D AP (u u p ) 2 u u (u u) (v v) (v v) (w w) (w w) m p g x (4.1) DRAFT 5 mp dv p dt 1 C D AP (v v p ) 2 u u (u u) (v v) (v v) (w w) (w w) m p g y (4.2) mp dw p dt 1 C D AP ( w w p ) 2 u u (u u) (v v) (v v) (w w) (w w) m p g z (4.3) dx p dt dy p dt dz p dt up (5.1) vp (5.2) wp (5.3) Where u, v, w: up, vp, wp: xp, yp, zp: gx, gy, gz: Ap mp CD dt DRAFT instantaneous velocities of air in x, y and z directions particle velocity in x, y and z direction particle moving in x, y and z direction gravity in x, y and z direction cross-section area of the particle mass of the particle density of the particle drag coefficient time interval 6 The drag coefficient for a spherical particle, taken from Wallis [16], is: 24 13 CD 1 Re Re 16 and 0.5 C D 0.44 for Re 1000 (6) for Re 1000 (7) The Reynolds number of the particle is based on the relative velocity between particle and air. In laminar flow, particles released from a point source with the same weight would initially follow the air stream in the same path and then fall under the effect of gravity. Unlike in laminar flow, the random nature of turbulence indicates that the particles released from the same point source will be randomly effected by turbulent eddies. As a result, they will be diffused away from the steam line at different fluctuating levels. In order to model the turbulent diffusion, the instantaneous fluid velocities in the 3 Cartesian directions, u, v and w are decomposed into the mean velocity component and the turbulent fluctuating component as: u u u , v v v , w w w . Where ū and u’ are the mean and fluctuating components in x-direction. The same applied for y-, z- directions. The stochastic approach prescribes the use of a random number generator algorithm which, in this case, is taken from Press et al [17] to model the fluctuating velocity. It is achieved through using a random sampling of a Gaussian distribution with a mean of 0 and a standard deviation of unity. Assuming homogeneous turbulence, the instantaneous velocities of air are then calculated from kinetic energy of turbulence: u u N (8.1) v v N (8.2) w w N (8.3) Where N is the pseudo -random number, ranging from 0 to1, with 2k 3 0.5 (9) k is the turbulent kinetic energy. The mean velocities which is the direct output of CFD determines the convection of the particles along the steam line, while the turbulent fluctuating velocity, Nα, contributes to the turbulent diffusion of the particle. DRAFT 7 3.2.3 Particle Interaction Time With the velocities known, the only component needed for calculating the trajectory is the time interval (tint) over which the particle interacts with the turbulent flow field. The concept of turbulence being composed of eddies is employed here. Before determining the interaction time, two important time scales need to be introduced: the eddy’s time scale and the particle transient time scale. Eddy’s time scale: life time of an eddy, defined as l te= e N (10) 2 le where C a 3 (11) le is the dissipation length scale of the eddy is the turbulent kinetic energy, C is a constant in the turbulence model. The transient time scale for the particle to pass through the eddy, tr, estimated as t r ln 1.0 and CD u v w 2 2 2 (u p ) 2 (v p ) 2 ( w p ) 2 le 4 D 3 p u 2 v 2 w 2 (u p ) 2 (v p ) 2 ( w p ) 2 (12) (13) Where is the particle relaxation time, indicating the time required for a particle starting from rest to reach 63% of the flowing stream velocity. D is the diameter of the particle. The interaction time is determined by the relative importance of the two events. If the particle moves slowly relative to the gas, it will remain in the eddy during the whole lifetime of the eddy, te. If the relative velocity between the particle and the gas is appreciable, the particle will transverse the eddy in its transient time, tr, Therefore, the interaction time is the minimum of the two: tint = min(te,tr) DRAFT (14) 8 It should be noted that the values of k and are assumed constant for the duration of the lifetime of the eddy. This results in a tendency for particles close to wall surfaces to stick to the surfaces, because In this study, the interaction time is in the order of 1e-5 to 1e-6s. If the particle track time is set to 300s, some 60,000,000 time-steps need to be performed just to calculate the trajectory for one particle. 3.2.4 Note on Use of k and Values in Calculation It should be noted that the values of k and are assumed constant for the duration of the lifetime of the eddy. This results in a tendency for particles close to wall surfaces to stick to them, because the turbulent velocity component added to the fluid velocity, and via its’ drag gives the particle additional momentum. This allows the particle to overcome the velocity condition at the wall, namely that the velocity at the wall surface is zero. If the other extreme condition for k and was taken, namely, that they are updated for every time step and so vary within the eddy, then virtually no particles hit the wall. The reason for this is that, as the wall is approached, the value of k tends to zero, meaning that the turbulent velocity component also tends to zero, and the application time of any given random fluctuating velocity component is so short that the particle accumulates little additional momentum. Consequently the particles does not tend to hit the wall surface. This effect was confirmed in tests in which the values of k and were interpolated at every time step. Although particles got very close to the surface, very few actually reached the wall, and so, using the accounting system, were tagged as still being active at the end of the 300s time period. In reality, the behavior of the particles close to wall surfaces lies somewhere between these two extreme assumptions. However, there is currently no research data available regarding k and values that are generally applicable to all types of wall boundary conditions encountered in this study. 3.3 Calculation Procedure The calculation procedure for each case consists of 4 steps: Computing the field distribution of fluid velocity, temperature and turbulent parameters Adding the UV distribution into the result field with the specified fixture location and measurement data. Specifying the source locations from where a specified number of particles are released. Performing computational analysis to calculate trajectory for each particle for up to 300s. The output of the analysis includes: DRAFT 9 - The number of particles being removed by ventilation varying with time (for every 60s) The number of particles sticking on walls varying with time (for every 60s) The number of viable particles varying with time (for every 60s) The number of particles killed by UV dosage varying with time (for every 60s) The percentage of surviving particles in the room varying with time (for every 60s The number of particles in different dose bands (for every 60s) 4. MODEL SET UP 4.1 CFD Models Figure 3 shows the configuration of the isolation suite being studied. The suite consists of three rooms, being connected through the door gaps between them: the main isolation room, bathroom and the vestibule. The main room is equipped with 4 slot diffusers near the window and a low induction diffuser on the ceiling. Three extreme weather conditions that affect the supply temperature are considered: Peak load: Maximum summer day solar loading for South facing isolation room: External temperature is 31.5°C (88.7°F). Peak T: Maximum summer day external temperature 35°C (95°F) without solar radiation in the room. Minimum T: Minimum winter night temperature of -11.7°C (10.9°F) The variation of the ventilation parameters involves: Supply flow rate ( 2 ACH- 16 ACH) Weather condition: summer or winter (supply temperature) Ventilation system: - Low exhausts - High exhausts - Low exhausts with baseboard heating for winter cases 20 cases with two low exhausts in the isolation room, as listed in Table 1, were studied to evaluate the influence of the supply flow rate and temperature on the particle tracking and killing. In order to examine the effects of ventilation system change, 10 cases were run with high exhausts and the combination of baseboard heating and low exhausts (see Table 2). Figure 4 shows the locations of the diffusers, exhausts and the baseboard heating in the main room. DRAFT 10 The UV lamp fixture is located on the partition wall between isolation room and vestibule 7.5’ (2.29m) from the floor with total lamp rating 36W. The plan view of the UV field generated by the lamp is shown in Figure 5. The UV intensity is assumed to be constant over the 5” (1.27e-2m) height of the lamp. bathroom UV fixture Isolation room vestibule Figure 3. Configuration of the isolation suite diffusers High exhausts Baseboard heating Low exhausts Figure 4. DRAFT Ventilation system in the isolation room 11 Figure 5. Plan view of UV field through lamp. Values in W/ cm2. Main Isol. Room (cfm) Sup. Exh. 62 42 125 105 “ “ 187 167 “ “ 250 230 “ “ Case Weather Condition ACH Case1 Case2 Case3 Case4 Case5 Case6 Case7 Min. T Peak T Min. T Peak T Min. T Peak load Peak T 2 4 “ 6 “ 8 “ Case8 Case9 Case10 Case11 Case12 Min. T Peak load Peak T Min. T Peak load “ 10 “ “ 12 “ 312 “ “ 375 Case13 Case14 Case15 Case16 Case17 Case18 Case19 Case20 Peak T Min. T Peak load Peak T Min. T Peak load Peak T Min. T “ “ 14 “ “ 16 “ “ “ “ 437 “ “ 499 “ “ DRAFT Bathroom (cfm) Vestibule (cfm) Sup. 0 “ “ “ “ 0 “ Exh. 100 “ “ “ “ “ “ Sup. 180 “ “ “ “ “ “ Exh. 150 “ “ “ “ “ “ “ 292 “ “ 355 “ 0 “ “ 0 “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ 26.5 12.2 17.5 25.8 14 “ “ 417 “ “ 479 “ “ “ “ 0 “ “ 0 “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ 18.4 25.3 15.3 19.1 25 16.3 19.5 24.8 Supply T 37.2 9.2 30 13.8 27.7 9.6 16.1 12 Table 1 20 cases with variation in supply flow rate and temperature Main Isol. Room (cfm) Sup. Exh. Bathroom (cfm) Sup. Exh. Vestibule (cfm) Sup. Exh. Suppy Temp (°C) Change in Ventilation system 150 25.8 Baseboard Heating “ “ “ “ 23.9 23.5 “ “ “ “ 23.3 “ Case Weather Condition ACH Case21 Min T 2 62 42 0 100 180 Case22 Case23 “ “ 6 12 187 375 167 355 “ “ “ “ Case24 “ 16 499 479 “ “ Case25 Peak T 4 125 105 “ “ “ “ 9.2 High exhausts in Main room Case26 Case27 Case28 Case29 Case30 Min T Peak T Min T Peak T Min. T “ 10 “ 16 “ “ 312 “ 499 “ “ 292 “ 479 “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ “ 30 17.5 25.8 19.5 24.8 “ “ “ “ “ Table 2 4.2 10 cases with variation of ventilation system Model for Bacteria Killing The bacteria are simulated as 100 particles released from 27 discrete source locations above the bed in a 3 x 3 x 3 array. The distance between the array release points in the vertical direction was 3’ (0.91m). The 2700 particles were tracked for 300s or until they were removed from the room by ventilation system or stuck to the wall. The percentage survival is dependent on exposure to UV dose, defined as Dose = Exposed time * UV Irradiance (15) in different patterns due to the room condition, the relative humidity, and the susceptibility of the species of the bacteria. In this report, the probability of survival was calculated using the experimental data shown in Figure 6. DRAFT 13 1 0.1 0.01 0.001 0 Figure 6. 150 300 450 600 750 900 1050 1200 1350 Dosage (W/ cm2s) Survival Fraction vs. Dosage for M. tuberculosis (ASHRAE Transactions 1999, V.105 Pt. 1) PS = a *exp (-k x) (16) a=100 k=0.00384 Where 4.3 a: PS: x: k: coefficient from curve fitting Survival probability UV Dose susceptibility Model for Impingement of Particles on Solid Surfaces When a particle is calculated to have hit a solid surface, the particle is defined to stick to the surface, and is effectively classed as ‘killed’. The particle is no longer considered to receive any further UV dose. DRAFT 14 5. RESULTS The results are presented in graphical format showing the status of the 2700 particles for the tracking period considered (300s). The particle status indicates the effectiveness of the ventilation system and UVGI. There are four particle statuses considered here: Status 1 Status 2 Status 3 Status 4 Vented out (considered killed) Particle stuck to wall (considered killed) Airborne, killed by UV (killed) Airborne, not killed (viable) The results from the particle tracking are presented in 15 charts showing: The number of particles being removed by ventilation varying with time (for every 60s) The number of particles sticking on walls varying with time (for every 60s) The number of viable particles varying with time (for every 60s) The number of particles killed by UV dosage varying with time (for every 60s) The survival fraction of particles varying with time (for every 60s) 5.1 The number of particles being removed by ventilation varying with time Figures 7 to 10 show the number of particles removed by ventilation varying with time for several parametrical changes. Figures 7 and 8 show the variation with ACH for the winter (with no baseboard heating) and summer conditions respectively. The winter cases, Figure 7, show a bigger variation in the number of ventilated particles than the summer cases. This is because there is generally poorer mixing at low ACH for winter cases with no baseboard heating than for summer cases. The results indicate that the winter cases are more effective at venting particles at higher ACH than summer cases. However, the ventilation of particles is not the whole story – particles are also removed via surface deposition and killing through UVGI. In particular, the primary means of removal in the summer cases is actually through the deposition of particles on wall surfaces (as demonstrated in Section 5.2), and there are actually fewer airborne particles in the summer cases to ventilate. Figure 9 shows the variation in vented particles with time based on exhaust location. The result indicates that the high level exhaust is more effective than the low level exhausts in removing particles through ventilation for the particle release points considered in this study. Finally, Figure 10 displays the variation in removed particles in selected winter cases with or without baseboard heating. The plot indicates that more particles are ventilated when baseboard heating is used, a result of an increase in mixing in the room. DRAFT 15 Number of particles vented out varying with time for different ACH (Winter, low exhausts) Number of particles 1200 1000 2 ACH 4 ACH 800 6 ACH 600 10 ACH 8 ACH 12 ACH 14 ACH 400 16 ACH 200 0 30 60 90 120 150 180 210 240 270 300 Time in second Figure 7. Number of vented out particles with ACH change (Winter) (Note 1) Number of particles vented out varying with time for different ACH (Summer, low exhausts) Number of particles 700 600 4 ACH 6 ACH 8 ACH 10 ACH 12 ACH 14 ACH 16 ACH 500 400 300 200 100 0 60 120 180 240 300 Time in second Figure 8. Number of vented out particles with ACH change (Summer) DRAFT 16 Number of vented out particles varying with time for Low / High exhausts (Winter) 1600 4 ACH (L) Number of particles 1400 4 ACH (H) 1200 16 ACH (L) 16 ACH (H) 1000 800 600 400 200 0 60 120 180 240 300 Time in second Figure 9. Number of vented out particles with exhaust location change (Winter) Number of vented out particles varying with time with/without B. Heating 800 Number of particles 700 2 ACH 2 ACH (BH) 12 ACH 12 ACH (B.H.) 600 500 400 300 200 100 0 60 120 180 240 300 Time in second Figure 10. DRAFT Number of vented out particles with /without Baseboard Heating 17 5.2 The number of particles sticking on wall varying with time Figures 11 to 13 show the number of particles sticking on wall varying with time for several parametrical changes. Figures 11 and 12 show the variation with ACH for the winter (with no baseboard heating) and summer conditions respectively. Figure 11, which shows the winter cases, indicates again that the mixing is poor and velocities are generally lower for these cases compared with Figure 12, which shows the summer case equivalent plot. The very good mixing in the summer cases, coupled with higher momentum flows, results in much greater deposition. Figure 13 shows the effect of adding baseboard heating to selected winter case ACH. In particular, the mixing is dramatically increased in cases in which baseboard heating is used, resulting in greater deposition of particles to external walls and inner surfaces. Number of particles sticking on wall varying with time for different ACH (Winter, low exhausts) Number of particles 1600 2 ACH 1400 4 ACH 1200 6 ACH 1000 8 ACH 10 ACH 800 12 ACH 600 14 ACH 400 16 ACH 200 0 30 60 90 120 150 180 210 240 270 300 Time in second Figure 11. DRAFT Number of particles sticking on wall with ACH change (Winter) (Note 2) 18 Number of particles sticking on wall varying with time for different ACH (Summer, low exhausts)) Number of particles 2500 4 ACH 6 ACH 8 ACH 10 ACH 12 ACH 14 ACH 16 ACH 2000 1500 1000 500 0 60 120 180 240 300 Time in second Figure 12. Number of particles sticking on wall with ACH change (Summer) Number of particles sticking on wall varying with time with/without B. Heating Number of particles 2500 2 ACH 2 ACH (BH) 12 ACH 12 ACH (B.H.) 2000 1500 1000 500 0 60 120 180 240 300 Time in second Figure 13. Number of particles sticking on wall with /without Baseboard Heating DRAFT 19 5.3 The number of viable particles varying with time Figures 14 to 17 show the number of viable particles varying with time for several parametrical changes. Figures 14 to 16 show the variation with ACH for the winter (with no baseboard heating), winter (with baseboard heating) and summer conditions respectively. The winter cases with no baseboard heating, Figure 14, show a bigger variation in the number of viable particles than the summer cases. Further, the number of viable particles is always higher for the winter cases with no baseboard heating than the summer cases. The main reason for this result can again be traced to the poor mixing conditions for the winter cases with no baseboard heating. In particular, the particles are less likely to be removed through ventilation, deposition on surfaces or killed by UV dosage because of the mixing. A point of interest here is the connection between Figure 15 and the concurrent study on the thermal comfort and uniformity in a typical patient room. Figure 15 shows that there is little benefit in increasing the ACH beyond 6 ACH – the curve for this case and that of the 12 ACH case is very similar. In the patient room study, a value of 6 ACH was found to provide very good thermal comfort and uniformity for winter cases with baseboard heat. The effect of exhaust location on selected winter cases is displayed in Figure 17. The results show that the high exhaust is more effective than the low exhaust for the particle release points considered in this study. Number of viable particles varying with time for different ACH (Winter, low exhausts): No Baseboard Heating Number of particles 3000 2 ACH 2500 4 ACH 6 ACH 2000 8 ACH 10 ACH 1500 12 ACH 14 ACH 1000 16 ACH 500 0 60 120 180 240 300 Time in second Figure 14. DRAFT Number of viable particles with ACH change (Winter) (Note 3) 20 Number of viable particles varying with time for different ACH (Summer, low exhausts) 1800 4 ACH Number of particles 1600 6 ACH 8 ACH 1400 10 ACH 1200 12 ACH 1000 14 ACH 16 ACH 800 600 400 200 0 60 120 180 240 300 Time in second Figure 15. Number of viable particles with ACH change (Summer). Number of viable particles varying with time (Winter, B.Heating) 2000 1800 Number of particles 1600 1400 1200 2ACH 1000 6ACH 12 ACH 800 16 ACH 600 400 200 0 60 120 180 240 300 Time in Second Figure 16. DRAFT Number of viable particles with exhaust location change (Winter). 21 Number of viable particles varying with time for Low / High exhausts (Winter) 3000 4 ACH (L) Number of particles 2500 4 ACH (H) 16 ACH (L) 2000 16 ACH (H) 1500 1000 500 0 60 120 180 240 300 Time in second Figure 17. 5.4 Number of viable particles with exhaust location change (Winter). The number of killed particles varying with time Figures 18 to 19 show the number of killed particles varying with time for ACH. Figures 18 and 19 display the variation for the winter (with no baseboard heating) and summer conditions respectively. The winter cases, Figure 18, do not show such a large variation with ACH as the other status parameters, and the results are closer to the summer results. The interesting aspect to these results is that the plot lines for the different values of ACH cross each other in several places for both the winter and summer cases. This makes it difficult to specify the most effective ACH value in conjunction with the UV beam based on this parameter alone. The survival fraction of the viable particles, presented in the next section, is a more effective means of determining this value. DRAFT 22 Number of particles 400 350 300 250 200 150 100 50 0 Number of killed particles varying with time for different ACH (Winter, low exhausts): No Baseboard Heating 2 ACH 4 ACH 6 ACH 8 ACH 10 ACH 12 ACH 14 ACH 16 ACH 60 120 180 240 300 Time in second Figure 18. Number of killed particles with ACH change (Winter) (Note 4) Number of killed particles varying with time for different ACH (Summer, low exhausts) 400 4 ACH Number of particles 350 6 ACH 8 ACH 300 10 ACH 12 ACH 250 14 ACH 200 16 ACH 150 100 50 0 60 120 180 240 300 Time in second Figure 19. DRAFT Number of killed particles with ACH change (Summer) 23 5.5 The number of killed particles varying with time Figures 20 to 22 show the survival fraction for viable particles varying with time for several parametrical changes. Figures 20 and 21 show the variation with ACH for the winter (with no baseboard heating) and summer conditions respectively. The winter cases, Figure 20, show that there is a distinct line between survival percentages of around 90%, and those of around 80%. This line appears at the intersection between 8 and 10 ACH. This indicates that, when baseboard heating is not included in the ventilation system, the inclusion of UV is less effective at low ACH, the cases it is intended to help most. The summer cases, Figure 21, indicate that there is no real variation in survival fraction with ACH – all values are equally as effective. The survival percentage for all these cases is around 80%, indicating a consistent advantage in the inclusion of UV lamps in the room. Finally, Figure 22 shows the effect of including baseboard heating for selected winter cases. The plot shows further evidence of the advantages in using baseboard heating in winter cases. In particular, the survival percentage is 10 – 30 % lower for the baseboard heating case at low ACH, and consistently 10% lower for the high ACH case. Survival fraction varying with time for different ACH (Winter, low exhausts) 2 ACH 4 ACH 6 ACH 8 ACH 10 ACH 12 ACH 14 ACH 16 ACH 1 0.9 Survival fraction 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 60 120 180 240 300 Time in second Figure 20. DRAFT Survival fraction with ACH change (Winter) 24 Survival fraction varying with time for different ACH (Summer, low exhausts) 1 4 ACH 6 ACH 8 ACH 10 ACH 12 ACH 14 ACH 16 ACH 0.9 survival fraction 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 60 120 180 240 300 Time in second Figure 21. Survival fraction with ACH change (Summer) Survival fraction varying with time with / without B. Heating (Winter, low exhausts) 1 0.9 2 ACH 2 ACH (BH) 12 ACH 12 ACH (B.H.) survival fraction 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 60 120 180 240 300 Time in second Figure 22. DRAFT Survival fraction with /without Baseboard Heating 25 Notes Note 1: Figures 7 to 10 show the number of particles that have been ventilated via the exhausts in the room varying with time. These particles are not used in the calculation of the average UV dosage for the remaining, viable particle population. Note 2: Figures 11 to 13 show the number of particles that have impinged on the walls of the room or any internal surface, such as the bed, wardrobe, etc., varying with time. These particles are not used in the calculation of the average UV dosage for the remaining, viable particle population. Note 3: Figures 14 to 17 show the number of viable particles varying with time. Viable particles are defined as the particles that are: - not vented out not deposited on walls or internal surfaces not killed by UV Note that only viable particles contribute to the average UV dose in calculating the percentage of surviving particles. Note 4: Figures 18 to 19 show the number of killed particles varying with time. The number of particles classed as killed is calculated as follows. 1/ The code determines the number of viable particles, and records the UV dose (irradiance in W/cm2 x period of exposure in seconds) experienced by each individual particle. 2/ At the conclusion of the time interval, an average total dose for the viable particle population is calculated. 3/ The average population UV total dose is used as the "It" term in Equation 1 to determine the percentage of survival for the particle population. 4/ The number of killed particles is then: Number of killed particles = Number of viable particles * (1 – (survival percentage for population/100)) At the beginning of the next time interval, the particles which are tagged as being killed are no longer included in the calculation of the survival percentage, irrespective of whether they are vented out or are deposited on a wall. The tagged particles are those which have the highest individual UV total dose. DRAFT 26 Note 5: Figures 20 to 22 show the survival fraction of the particle population varying with time. The survival fraction is calculated with Equation (16) using steps 1/ to 3/ in Note 3. Note 6: time. This plot shows the total dose bands of all airborne particles varying with Airborne particles are defined as the particles that are: - not vented out not deposited on a wall may or may not be killed by UV. The bands are calculated as follows: The code determines the number of airborne particles, and records the UV total dose (irradiance in W/cm2 x period of exposure in seconds) experienced by each individual particle. In order to help understanding how the particles are classified, Table 3 lists the particle numbers in different status at the end of every minute for Case10. Table 3: Vented out Wall deposition Airborne Viable % Surviving Killed Budget Table for 2700 Particles (Case 10) End of Min 1 43 1044 End of Min 2 170 1531 End of Min 3 253 1818 End of Min 4 298 2004 End of Min 5 325 2126 1613 1613 (1) 81.0 (2) 306 (3) 999 821 (4) 81.6 151 629 432 76.4 102 398 215 77.7 48 249 115 69.6 35 The summation of airborne, vented out and wall deposition at the end of any minute is 2700. (1) As this calculation is at the end of the first time interval, all airborne particles are assumed to be viable. (2) The average UV total dose for the viable particle population is used in the calculation of the survival percentage. (3) Number of killed particles = Number of viable particles * (1 – (survival percentage for population/100)) Number of killed particles = 1613*(1-(81.0/100)) = 306 DRAFT 27 (4) The number of viable particles does not match the number of airborne particles from the second interval onwards. This is because the airborne total can include a number of killed particles from the previous time interval that have not been vented or deposited on a wall surface. 6. SUMMARY The results from the cases studied show: The number of particles vented out of the room increases with ACH. The variation with ACH is more pronounced for winter cases with no baseboard heating than summer cases, because low ACH cases have poorer mixing. Cases with high exhaust grilles vent out more particles than low exhaust systems for the particle release points considered in this study. The number of particles ventilated out from the room is lower when baseboard heating is used. The reason for this is that particles are more prone to sticking to the wall when baseboard heating is included because of increased mixing in the room. The number of particles sticking to the walls or internal surfaces increases with ACH. The variation is more pronounced in winter cases with no baseboard heating because of poor mixing at low ACH. The number of viable particles parameter clearly shows the advantages of using baseboard heating. The results show that there is little advantage in increasing the ventilation rate in the room beyond 6 ACH for summer cases, or winter cases with baseboard heating. This value is also consistent with the results of a concurrent study examining thermal comfort and uniformity in patient rooms. In particular, this study suggests that the optimum ventilation rate for similar winter conditions as considered here is 6 ACH to provide good levels of thermal comfort and uniformity. This value is also suitable for summer condition cases. The number of viable particles in the room is generally lower for high exhaust systems compared with low exhaust system cases. Distinct trends for the number of killed particles are more difficult to determine. In particular, the lines for the number of killed particles cross each other in several places. UVGI does result in the killing of a significant percentage of the viable particles in the room. DRAFT In particular, as seen by the Table 3 example, UVGI kills around 20 – 28 30% of the viable particles in the room. The exception is for winter cases with no baseboard heating and low ACH– here the UVGI only kills around 10% of the viable particles. The addition of baseboard heating results in much better kill UVGI rates irrespective of ACH. Baseboard heating should therefore be used in winter cases, especially at low ACH. Irrespective of the addition of UVGI in the room, the primary mechanism for particle removal is through deposition on wall surfaces. 7. REFERENCES 1. Federal Register “Draft Guidelines for Preventing the transmission of Tuberculosis In Health –Care Facilities, Second Edition; Notice of Comments Period”, Vol.58, No.195 (1993). 2. J.C. Chang, S.F. Ossoff, D.C. Lobe,M.H. Dorfman, R.G. Quall, J.D. Johnson, “UV Inactivation of Pathogenic and Indicator Microorganisms”, All.Environ. Microbiol. Vol.49, pp.1361-1365 (1985). 3. J.M. Macher, L.E. Alevantis, Y-L. Chang, K-S, Liu, “Effect of Ultraviolet Germicidal Lamps on Airborne Microorganisms in Outpatient Waiting Room”, Appl. Occ. Environ.Hyg, Vol..7, pp.505-513 (1992). 4. V.D. Mortimer and R.T. Hughes, “The Effects of Ventilation Configuration and Flow rate on Contaminant Dispersion”, American Industrial Hygiene Association Conference and Exposition (1995). 5. Z. Jiang, F. Haghighat, Q. Chen, "Ventilation performance and indoor air quality in workstations under different supply air systems: a numerical approach", Indoor + Built Environment, 6, 160-167 (1997). 6. Z. Jiang, Q. Chen, F. Haghighat, "Airflow and Air Quality in Large Enclosures", ASME Journal of Solar Energy Engineering Vol.117, pp.114-122. (1995). 7. F. Haghighat, Z. Jiang, Y. Zhang, "Impact of ventilation rate and partition layout on VOC emission rate: time-dependent contaminant removal", International Journal of Indoor Air Quality and Climate, Vol.4, pp.276-283 (1994). 8. FLOVENT Reference Manual 1995, Flomerics, FLOVENT/RFM/0994/1/1 9. Memarzadeh, F., “Ventilation Design Handbook on Animal Research Facilities Using Static Microisolators”, National Institutes of Health, Office of the Director, Bethesda (1998). DRAFT 29 10. D. Gosman, E. Ioannides, “Aspects of Computer Simulation of Liquid-Fuelled Combustors”, AIAA 19th Aerospace Science Meeting 81-0323, 1 – 10 (1981). 11. A. Ormancey, J. Martinon, “Prediction of Particle Dispersion in Turbulent Flow”, PhysicoChemical Hydrodynamics 5, 229 – 224 (1984). 12. J-S. Shuen, L-D. Chen, G. M. Faeth, “Evaluation of a stochastic Model of Particle Dispersion in a Turbulent Round Jet”, AIChE Journal 29, 167 – 170 (1983). 13. P-P. Chen, C.T. Crowe, “On the Monte-Carlo Method for Modelling Particle Dispersion in Turbulence Gas-Solid Flows”, ASME-FED 10, 37 – 42 (1984). 14.W.H. Snyder, J.L. Lumley, “Some Measurement of Particle velocity Autocorrelation Functions in Turbulent Flow”, J. Fluid Mechanics 48, 41 – 71 (1971). 15. Alani A, D. Dixon-Hardy D, M. Seymour, “Contaminants Transport Modelling”, EngD in Environmental Technology Conference (1998). 16. G.B.Wallis, “One Dimensional and Two Phase Flow”, McGraw-Hill, New York (1969). 17. W.H. Press, S.A. Teukolsky, W.T. Vetterling, B.P. Flannary, “Numerical Recipes in FORTRAN”, Second Edition, ISBN 0 521 43064 X, Cambridge University Press, Cambridge (1992). DRAFT 30