4
Bioaerosol Sampling Systems for Near-Real-Time Detection

A bioaerosol sampling system is the first stage of most detection systems for defense against attacks involving aerosolized biological agents. The basic components of an aerosol sampling system are an inlet, a size fractionation device that strips unwanted larger-sized particles and debris from the distribution, a concentrator that confines the particles in a smaller volume of air, and a collector that deposits the particles on a surface or in a liquid. The output of the sampling system varies (e.g., wet or dry, concentrated or not) depending upon the requirements of the detector.

For liquid-based detectors, the collector will involve an aerosol-to-hydrosol transfer stage that serves the function of providing a small liquid flow rate (e.g., 0.5 milliliters per minute) to the detector, and for one variation of B-cell technology (see Chapter 7),1 the aerosol is collected in the dry state prior to the addition of B cells for analysis. In the future, there may be optical devices that can speciate airborne organisms, in which case neither a liquid-based nor a dry collector would be needed, because the particles would be identified in the aerosol state.2

Commercial bioaerosol sampling systems are available from several companies, and the Department of Defense has developed prototype bioaerosol detection systems that include sampler technologies. However, none of these systems has been optimized for the detect-to-warn function considered in this report.

Consistent with the scenarios discussed in Chapter 2, this chapter examines the requirements of bioaerosol sampler technologies for both indoor and outdoor environments. Two distinct types of samplers are required for indoor occupied environments (e.g., subway stations, airports, arenas, and office buildings), depending on whether the sampling is done from HVAC ductwork or from rooms or other open areas. The sampler in the latter application will be referred to as an area sampler. A third type of sampling system is required for the outdoor ambient environment.

The design of samplers for use in the ambient environment is presently a much more serious challenge than that for occupied environments. If an ambient sampler is to be used for protection of frontline troops, it must be portable, have minimal logistical requirements (power, consumable supplies, and operator interactions), be able to acquire samples over a broad range of meteorological conditions (wind velocity, direction, temperature, and precipitation) and be unobtrusive. If it is to be used to monitor the bioaerosol content of the ambient environment in the vicinity of a military base, a critical building, or a

1  

J.D. Harper, MIT Lincoln Laboratory. Presentation to the committee on June 13, 2002.

2  

Systems are already being developed that can detect moieties, such as riboflavin, NADH, and tryptophan, in biological cells owing to their fluorescence characteristics. These are discussed in greater detail in Chapter 5.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases 4 Bioaerosol Sampling Systems for Near-Real-Time Detection A bioaerosol sampling system is the first stage of most detection systems for defense against attacks involving aerosolized biological agents. The basic components of an aerosol sampling system are an inlet, a size fractionation device that strips unwanted larger-sized particles and debris from the distribution, a concentrator that confines the particles in a smaller volume of air, and a collector that deposits the particles on a surface or in a liquid. The output of the sampling system varies (e.g., wet or dry, concentrated or not) depending upon the requirements of the detector. For liquid-based detectors, the collector will involve an aerosol-to-hydrosol transfer stage that serves the function of providing a small liquid flow rate (e.g., 0.5 milliliters per minute) to the detector, and for one variation of B-cell technology (see Chapter 7),1 the aerosol is collected in the dry state prior to the addition of B cells for analysis. In the future, there may be optical devices that can speciate airborne organisms, in which case neither a liquid-based nor a dry collector would be needed, because the particles would be identified in the aerosol state.2 Commercial bioaerosol sampling systems are available from several companies, and the Department of Defense has developed prototype bioaerosol detection systems that include sampler technologies. However, none of these systems has been optimized for the detect-to-warn function considered in this report. Consistent with the scenarios discussed in Chapter 2, this chapter examines the requirements of bioaerosol sampler technologies for both indoor and outdoor environments. Two distinct types of samplers are required for indoor occupied environments (e.g., subway stations, airports, arenas, and office buildings), depending on whether the sampling is done from HVAC ductwork or from rooms or other open areas. The sampler in the latter application will be referred to as an area sampler. A third type of sampling system is required for the outdoor ambient environment. The design of samplers for use in the ambient environment is presently a much more serious challenge than that for occupied environments. If an ambient sampler is to be used for protection of frontline troops, it must be portable, have minimal logistical requirements (power, consumable supplies, and operator interactions), be able to acquire samples over a broad range of meteorological conditions (wind velocity, direction, temperature, and precipitation) and be unobtrusive. If it is to be used to monitor the bioaerosol content of the ambient environment in the vicinity of a military base, a critical building, or a 1   J.D. Harper, MIT Lincoln Laboratory. Presentation to the committee on June 13, 2002. 2   Systems are already being developed that can detect moieties, such as riboflavin, NADH, and tryptophan, in biological cells owing to their fluorescence characteristics. These are discussed in greater detail in Chapter 5.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases site for a mass gathering, the logistical requirements can be relaxed; however, the sampler must still be able to cope with wind speed and direction, temperature, and precipitation effects. For the detect-to-warn (DTW) capability, emphasis is generally placed on avoiding false negatives (i.e., not underestimating the aerosol concentration) as opposed to attaining a representative sample. However, if there were to be an attack with a bioagent, it is likely that the DTW device that triggers the alarm would also be a very important source of data for retrospective determination of dose. As a consequence, there should be traceability between response of the DTW device and dose, which implies that the representativeness of the sample should be understood. Of course, in any real detector system, the cost and benefit of this added level of sophistication, as opposed to the more limited goal of determining the presence or absence of a threat and an indication of its location and magnitude, would have to be evaluated. Some of the components used in sampling systems involve critical dimensions, and sampling systems that are designed for use in the ambient environment can be subjected to conditions that may lead to degradation of some materials. As a consequence, the use of novel materials and manufacturing processes may be required during development and production of DTW bioaerosol samplers. PARTICLE SIZE CONSIDERATIONS In general terms, a sampling system must be able to collect aerosol particles sized such that they can most efficiently be deposited in the human respiratory system. Traditionally, this has meant a focus on particles between 1 and 10 μm aerodynamic diameter (AD).3 However, other considerations suggest that a focus on this size range may be too limited. If, for example, the aerosol generation processes likely to be used by an attacker produce the majority of aerosol particles in sizes outside the optimal inhalation band, more effective detection could be achieved if the sampled size band is more closely matched to the generator output. A thorough analysis of the advantages and disadvantages of extending the size range to smaller or larger sizes should include consideration of (1) the likely size distribution of threat agents as well as that of the biological and nonbiological aerosol backgrounds; (2) how the inclusion of a given size range would affect the sensitivity, specificity, and false alarm vulnerability for a given collector/sensor class; and (3) the feasibility of expanding the size range for the specific scenarios addressed for a given collector/sensor class. Also, contemporary detectors generally sense the number of cells in a particle as opposed to the number of particles. Because the number of cells increases with the cube of particle diameter for similar bioparticles, sampling particles larger than 10 μm AD may increase the sensitivity of the detection process. However, other factors such as background effects associated with sampling larger particles might counteract the benefit. Sampling from the Ambient Environment Particles with sizes much below 1 μm aerodynamic diameter (AD) are difficult to generate in large quantities from either liquid slurries or unground bulk powders. Moreover, if vegetative cells or virus particles were to be aerosolized as submicrometer-sized particles, environmental stresses would probably significantly reduce their viability due to the high surface/mass ratio of the particles. To illustrate the difficulty of generating micrometer-sized droplets, consider the release of an agent through 0.01 millimeter slits in a spray boom on an aircraft traveling at Mach 1. The mean droplet size generated by this process would be approximately 10 μm diameter.4 If the viscosity of the fluid were higher than that of water because of the presence of an agent the droplet size would be even larger. A cloud of 1 μm droplets could be produced by this process but it would require atomization of a dilute 3   Aerodynamic diameter of a particle of arbitrary shape and density is the size of a water droplet that will have the same sedimentation velocity in air. E.W. Stuebing, U.S. Army. Presentation to the committee on April 15, 2002. 4   R.D. Ingebo and H.H. Foster. 1957. Drop-size distribution for cross-current breakup of liquid jets in airstreams. NACA TN 4087.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases hydrosol with the liquid being volatile. Initial droplets containing 99.9 percent (v/v) of a volatile solvent (e.g., water) and 0.1 percent agent would significantly reduce the amount of agent that could be transported by the aircraft. Notwithstanding the above, for some specific applications particles that would be released in sizes smaller than 1 μm AD should be considered. Particles with sizes greater than 10 μm AD have traditionally been excluded from consideration in bioaerosol ambient sampling because deposition of those particles in the alveolar region of the human respiratory system is considerably less effective than deposition of particles smaller than 10 μm AD,5 and also because naturally occurring background interferents such as plant pollens and other debris contain a relatively high fraction of particles larger than 10 μm AD, which may influence some detector systems. However, for some ambient environment sampling applications, there is justification for considering the detection of bioaerosol particles with sizes larger than 10 μm AD. From the respiratory point of view, a particle size as large as 100 μm can be inspired into the oral cavity or nasal passages with an efficiency of 50 percent.6 In addition, the methods used in the preparation and release of a bioaerosol can lead to a considerable fraction of aerosol particles with sizes greater than 10 μm AD. For example, if a lyophilized powder were not finely ground or if either a solid or liquid agent were released in an unsophisticated manner (low energy input to the aerosolization process), most of the organisms could be associated with sizes larger than 10 μm AD. Figure 4.1 shows the size distributions of Bacillus subtilis spores that were aerosolized by a very simple process. The spores, in the form of a lyophilized, unground powder, were placed in an envelope that was processed through a mail-sorting machine, where the envelope was subjected to sudden pressure forces applied by high-speed rollers and belts. On a particle number basis, only 4 percent of the particles are associated with sizes larger than 7.1 μm AD; however, because the number of spores in a particle varies approximately with the cube of particle size, about 65 percent of the spores are associated with sizes larger than 7.1 μm AD (volume distribution in Figure 4.1). Current speciation detectors generally respond to the number of cells, and because the number of cells in similar particles increases with the cube of diameter, it may be desirable to collect the larger particles, assuming the absence of significant changes in other effects such as increasing background interferents. Sampling from Occupied Environments For sampling of aerosols in building environments, either from ductwork or from the occupied environment, particles with sizes outside the nominal range 1 to 10 μm AD should be considered. The upper size for occupied environment sampling applications should be selected after considering several factors such as the potential methods for aerosolization of threat bioagents, the loss of particles during transport from the site of aerosolization to the sampling location, the effectiveness of filters and air conditioning components in duct sampling applications, and a realistic assessment of being able to collect a sample of the aerosol particles and transport it to the detector. In general terms, aerosol particles with sizes much larger than about 30 μm are difficult to efficiently sample and transport using this sampling apparatus, so this size could be considered as a nominal upper limit for many applications. BIOAEROSOL SAMPLING FROM INDOOR AIR Interior sampling scenarios include sampling from ductwork and from occupied open areas such as subway stations, airports, arenas, and office buildings. An aerosol concentrator may not be required for 5   M. Lippman. 1977. Respirable dust sampling. In Air Sampling Instruments for Evaluation of Atmospheric Contaminants, 5th Ed. Cincinnati, Ohio: American Conference of Industrial Hygienists. 6   S.C. Soderholm. 1989. Proposed international conventions for particle size-selective sampling. Ann. Occup. Hyg. 22:301-320.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases building applications, because the unmodified concentration associated with an aerosol release may be sufficiently high to cause the detector to alarm. For example, if there were to be a release that effectively aerosolized 1 gram of bioagent containing 1012 spores into a building with a volume of 3 × 104 cubic meters (3 × 107 liters) and if the sampling were to take place from the fully mixed aerosol, a sampling flow rate of 28 liters per minute (1 cubic foot per minute, cfm) would lead to the collection of 106 spores per minute. Other situations could, of course, lead to either larger or smaller numbers of spores—for example, if the aerosol release were into the building air intake and if the sampler were to collect material from that duct, the integrated dose sensed by the detector during 1 minute could be several orders of magnitude higher than that sensed by a detector monitoring the air in the occupied environment. FIGURE 4.1 Size distribution of bioaerosol particles generated from a contaminated letter during a mail sorting operation. Sampling from Building Ductwork Bioaerosol monitoring in critical government and civilian buildings can be partially accommodated by extractive sampling from the ductwork. The fresh air intake ductwork is often at ground level and in a location that can easily be accessed by unauthorized personnel. Routine monitoring of the fresh air intake would provide a rapid warning of either a biological aerosol release at this susceptible location or intake of an externally released aerosol cloud. Also, in some buildings an effective detect-to-warn capability could be economically implemented by sampling recirculated air in the ductwork. Extractive sampling is needed for ductwork, and because the velocities in ductwork can range from 3 to 25 meters per second7 and the concentration and velocity profiles at a prospective sampling location in a duct may be irregular, acquisition of a meaningful air sample is a potentially difficult challenge. There are standardized methods for batch-sampling aerosol particles from stacks such as those codified by the EPA.8 The batch techniques compensate for irregularities in the concentration and velocity profiles by sequentially sampling at prespecified points on a geometrical grid across the stack cross section. At each traverse point on the grid, the velocity of gas at the inlet plane of the sampling nozzle is set to equal that of the undisturbed stack velocity at the particular point. This operational mode of equal sampling and flow stream velocities is called isokinetic sampling. It is intended to assure that the collected sample is 7   Carrier Air Conditioning Corporation. 1960. Carrier System Design Manual, Part 2, Air Distribution. Syracuse, N.Y. 8   For example, 40 CFR Part 60, Method 5.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases representative, provided the concentration does not change during the period in which the batch sample is collected. Unfortunately, isokinetic sampling is not compatible with unattended operation of a near-real-time detection system for bioaerosols. Apparatus that could articulate the sampling nozzle would be complicated, and a pulse release could be significantly underestimated if the nozzle is extracting aerosol from a low-concentration region of the duct during the time of the release. Also, there can be significant losses of aerosol particles in the nozzles and transport lines of a batch sampling system. This is not a problem for batch monitoring of stacks, because the sampling protocol requires cleaning the sampling nozzle and transport tubes at the completion of a test. For continuous monitoring for detect-to-warn purposes, however, in situ cleaning of the sampling system is not an option. The EPA requires continuous emission monitoring of stacks and ducts in U.S. government facilities that can potentially emit significant quantities of radionuclides into the ambient environment.9 Until recently, EPA had prescribed use of the sampling protocol specified in a 1969 version of an American National Standard.10 For ducts larger than 152 mm diameter (6 in.), the ANSI-1969 standard recommended use of multiple sampling nozzles that would be operated isokinetically relative to the airstream. Rakes of such nozzles (Figure 4.2) were typically used to span a duct cross section. However, there are two problems with this approach: The nozzles of such rakes will not all be isokinetic because of natural spatial variations in a velocity profile, and, more importantly, substantial aerosol particle losses can occur on the inner walls of the nozzles.11 Tests with an ANSI-196912 nozzle operated isokinetically in an aerosol wind tunnel at 10 meters per second showed losses of 75 percent for 10-μm AD aerosol particles,13 and tests with a rake of nozzles in a nuclear stack showed only 41 percent of the radionuclide activity was associated with material collected on a sampling filter, with the remainder lost in the sampling system.14 Because of problems with the extractive sampling approach used in ANSI N13.1-1969, a more robust approach was developed for continuous emission monitoring of the stacks and ducts of the nuclear industry.15 This methodology is single-point representative sampling, whereby a sample is extracted at a location in the duct where both fluid momentum and contaminant concentration are well-mixed, as manifested by the uniformity of the velocity and contaminant concentration profiles. It is the recommended approach in a revision to the ANSI standard, ANSI N13.1-1999.16 An illustration of a single-point sampling system is shown in Figure 4.3. A shrouded probe17 that has both low internal wall losses and negates the need for isokinetic sampling is used for sample extraction. It provides representative aerosol samples from the single-point in a duct under conditions for which the sampling flow rate is constant, but the air velocity in the duct is variable. These systems are optimized for 9   40 CFR Part 61, Subparts H and I. 10   ANSI. 1969. Guide to Sampling Airborne Radioactive Materials in Nuclear Facilities. ANSI N13.1 New York: American National Standards Institute. 11   M.D. Durham and D.A. Lundgren. 1980. Evaluation of aerosol aspiration efficiency as a function of Stokes number, velocity ratio and nozzle angle. J. Aerosol Sci. 11:179-188. 12   American National Standards Institute designation. 13   B.J. Fan, F.S. Wong, C.A. Ortiz, N.K. Anand, and A.R. McFarland. 1992. Aerosol particle losses in sampling systems. In Proceedings of the 22nd DOE/NRC Nuclear Air Cleaning and Treatment Conference. CONF-9020823. M.W. First, ed., pp 310-322. Washington, D.C.: U.S. Department of Energy. 14   R.B. Schappel. 1961. An investigation of the solid particulate collection efficiency of the traverse-type stack probe. U.S. Atomic Energy Commission Research and Development Report Y-1372. Oak Ridge, Tenn.: Union Carbide Nuclear Company. 15   A.R. McFarland and J.C. Rodgers. 1993. Single-point representative sampling with shrouded probes. LA-12612-MS. Los Alamos, N.Mex.: Los Alamos National Laboratory. 16   ANSI. 1999. Sampling and monitoring releases of airborne radioactive substances from stacks and ducts of nuclear facilities. ANSI/HPS Standard N13.1-1999. McLean, Va.: Health Physics Society. 17   A.R. McFarland, C.A. Ortiz, M.E. Moore, R.E. DeOtt, Jr., and A. Somasundaram. 1989. A shrouded aerosol sampling probe. Environ. Sci. Tech. 23:1487-1492. S. Chandra and A.R. McFarland. 1997. Shrouded probe performance: Variable flow operation and effect of free stream turbulence. Aerosol Sci. Tech. 26:111-126.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases FIGURE 4.2 Rake of nozzles recommended in ANSI N13.1-1969 for continuous emission monitoring of nuclear facilities. transmission of aerosol particles with sizes less than or equal to 10 μm AD, as is illustrated in Figure 4.4; however, they have not been tested with aerosol particles as large as 30 μm AD. Aerosol transmission of a sampling nozzle or probe is the ratio of aerosol concentration at the exit plane of the nozzle to the undisturbed aerosol concentration of the flow stream at the location of the probe. Loss of aerosol particles in sample transport systems is a matter of importance in the building air monitoring scenario. Larger-sized aerosol particles (e.g., those with sizes greater than about 5 μm AD) can be inadvertently deposited on internal walls of nozzles by turbulent deposition and forces set up by fluid shear; on the walls of straight tubes by gravitational settling (nonvertical tubes) and turbulent deposition; and on the walls of tube bends and fittings by inertial forces. A software code such as Deposition18 or hand calculations19 can be used to estimate aerosol particle transmission through complex transport systems; conversely, the calculations can serve as a design tool for optimizing particle transport through a system with a given geometrical configuration. FIGURE 4.3 A generic system for extractive sampling of aerosols from a duct. 18   A.R. McFarland, A. Mohan, N.H. Ramakrishna, J.L. Rea, and J. Thompson. 2001. Deposition: An illustrated user’s guide. Report 6422/03/01/ARM. College Station: Texas A&M University. 19   J.E. Brockman. 1993. Sampling and transport of aerosols. Aerosol Measurments. K. Willeke and P. Baron, eds. New York: van Nostrand Reinhold.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases FIGURE 4.4 Sampling performance of a shrouded probe: a) Effect of duct velocity. Particle size = 10 µm AD. b) Effect of particle size. Wind speed = 12 m/s. SOURCE: S. Chandra and A.R. McFarland, A.R. (1997). Shrouded probe performance: Variable flow operation and effect of free stream turbulence. Aerosol Sci. Technol. 26:111-126. The Deposition software, which is acceptable methodology for demonstrating compliance with the rules of the Nuclear Regulatory Commission20 and ANSI N13.1-1999 for estimating aerosol particle losses in sampling systems, can be used to evaluate the effectiveness of systems for sampling variously sized bioaerosol particles in ductwork. Sampler System Components Preseparators. A preseparator may be used in a ductwork sampling system to preclude entry of debris such as lint particles into the subsequent components of the system. There are two main techniques that are used for stripping unwanted large-sized debris from the size distribution. First is a cyclonic separator, which employs vortex flow (Figure 4.5). Cyclones have the advantage that they can be operated for long intervals (i.e., several weeks) without requiring cleaning; however, they can be relatively large and therefore not only difficult to locate strategically but also to clean. The second technique is inertial impaction, with either a classical impactor or a virtual impactor (Figure 4.6). In these devices the aerosol is accelerated in a jet, which is directed toward a solid collection surface (Figure 4.6a) in the case of a classical impactor or toward a receiver nozzle (Figure 4.6b) in the case of a virtual impactor. For the classical impactor, the deposition of particles takes place in an area on the collection surface that is only slightly larger than the projected area of the acceleration jet, and as a consequence there can be a rapid buildup of dust on the collection surface. When subsequent particles are deposited on a dust-laden impaction surface, aerosolization of previously collected dust may take place. This can be ameliorated by fabricating the collection surface from porous media and soaking the media in oil prior to use. The oil will saturate the dust layer and minimize the aerosolization process; however, because of the restricted area over which the impaction takes place, there is still a need for frequent (every other week or so) cleaning of the collection surface. 20   U.S. Nuclear Regulatory Commission. 1993. Air sampling in the workplace. Regulation 8.25.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases FIGURE 4.5 Cyclone separator: (a) three-dimensional view, and (b) section just below inlet. The virtual impaction concept lends itself well to the application of stripping large particles from a size distribution. Approximately 10 to 20 percent of the air that flows through the acceleration jet is drawn into the receiver jet, where it transports the large particles from the fractionation zone, along with 10 to 20 percent of the finer particles. Cyclone Preseparators. Empirical models have been developed that enable determination of design parameters for a cyclone that is to be operated at a given flow rate and that will provide a particular cutpoint size.21 For a cyclone with a single inlet and a geometrical configuration similar to that shown in Figure 4.5, the cutpoint is related to the size (body diameter) of the cyclone.22 A cyclone designed to have a cutpoint of 10 μm at a flow rate of 57 liters per minute would have a body diameter of 80 mm; however, if the cutpoint is increased to 30 μm AD, the corresponding body diameter would be 140 mm. Typically, a cyclone has a height about 4 times the diameter, so the 140 mm diameter cyclone would be about 0.6 meters high. The variation of collection efficiency of the cyclone with particle size is shown in Figure 4.7. Ideally a fractional efficiency curve such as that shown in Figure 4.7 would be a step function, where particles smaller than the cutpoint would pass through the cyclone and those larger than the cutpoint would be collected. However, a cyclone, as with most aerosol size-fractionators, collects some of the particles that would be desirable to transmit and transmits some of the particles that would be desirable to collect. From Figure 4.7, it may be noted that if the cyclone is designed to have a cutpoint of 30 μm AD, approximately 10 percent of 20 μm AD particles will be collected and 90 percent will penetrate through it. 21   The cutpoint is the particle size for which 50 percent of the aerosol particles are separated from a flow stream and 50 percent are retained in the flow stream. 22   B.E. Saltzmann and H.M. Hochstrasser. 1983. Design and performance of miniature cyclones for respirable aerosol sampling. Environ. Sci. Technol. 17:418-424.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases FIGURE 4.6 Inertial impactor aerosol fractionators: (a) classical impactor and (b) virtual impactor. Preseparation with a Classical Inertial Impactor. The principle of operation of a classical impactor is that an air jet, when directed onto a flat plate, will turn abruptly at the plate surface. Particles with sufficient inertia (i.e., larger sizes) will strike the plate, whereas smaller particles can be carried with the airstream away from the impaction zone. Whether or not impaction takes place is primarily a function of a parameter called the Stokes number, Stk, which may be considered to be the ratio of the inertial (centrifugal) force exerted on a particle in a curvilinear airflow field to the drag force that tends to resist particle motion perpendicular to the curved airstreamlines. It is the motion of a particle in the direction normal to an airstreamline that causes it to strike a wall and be removed from the flow stream; thus, the larger the Stokes number, the greater the probability that a particle will impact the collection surface. With reference to Figure 4.8, the fractional efficiency of a classical circular-jet impactor is shown as a function of the Stokes number, where it may be noted the cutpoint Stokes FIGURE 4.7 Variation of collection efficiency of a single inlet cyclone with particle size. Da is the aerodynamic diameter of a particle. SOURCE: M.E. Moore and A.R. McFarland. 1993. Performance modeling of single inlet aerosol sampling cyclones. Environ Sci. Technol. 27:1842-1848.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases number23 is about 0.24. Here, the Stokes number is defined as: (1) where C = Cunningham's slip correction,24 which has a value very nearly equal to 1 for particles greater than 10 μm AD; ρw = density of water; Da = aerodynamic particle diameter, Uj = velocity at the exit plane of the acceleration jet; μ = air viscosity; and, dj = diameter of the acceleration jet at its exit plane. The cutpoint Stokes number is approximately a constant, so for a fixed flow rate different cutpoint sizes can be achieved by varying the jet diameter. For a slit impactor (rectangular acceleration jet), the jet diameter is replaced by the slit width. The cutpoint Stokes number for a slit impactor is about 0.59. A classical impactor is compact and easy to construct; however, it can produce biased results unless cleaned frequently. Rebound of incident particles and aerosolization of collected deposits can be minimized by use of an oil-soaked impaction surface; however, fibers in the deposit will protrude into the airstream and can filter particles smaller than the cutpoint from the size distribution. Classical impactors are widely used for preseparation in routine ambient air sampling, where characterization of mass concentration is the goal of the sampling effort. The EPA has a standard method for sampling PM10 aerosol,25 which stipulates that a fractionator with a cutpoint of 10 μm AD shall be used to condition the aerosol prior to its collection. The ThermoAndersen Model 1200 sampler26 accomplishes this by first passing the airflow through an impactor and then collecting the residual aerosol FIGURE 4.8 Classical slit jet inertial impactor with a cutpoint of 0.8 μm AD: (a) geometry showing important parameters and (b) performance. W = 0.38 mm; velocity of jet = 21 m/s; S/W = 2.5; T/W = 2. 23   Stokes number for which the collection efficiency is 50 percent. 24   N.A. Fuchs. 1964. The Mechanics of Aerosols. New York: The Macmillan Company. 25   40 CFR Part 53. 26   ThermoAndersen, Inc., Smyrna, Ga.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases with a preweighed filter. Nine impactor jets operate in parallel to provide a sampling flow rate of 1,130 liters per minute. The EPA reference method for collection of PM-2.5 samples27 involves the use of a cup-shaped impactor to strip particle sizes greater than or equal to 2.5 μm AD from the size distribution at a flow rate of 16.7 liters per minute. Oiled collection surfaces are used in both the PM-10 and PM-2.5 impactors. Preseparation by a Virtual Impactor A virtual impactor reduces the cleaning problem associated with a classical impactor; however, it is more expensive to fabricate and it requires two flow control systems. MSP Corporation in Minneapolis has developed a preseparator that strips particles with sizes larger than 10 μm AD from the size distribution entering an aerosol concentrator.28 The flow rate into the preseparator is 330 liters per minute, of which 300 liters per minute then flow into the concentrator, while the remaining 30 liters per minute are exhausted from the system. Particle separation in a virtual impactor is primarily a function of the Stokes number and the fraction of the flow rate, f, that is drawn into the receiver port, as shown in Figure 4.9. Geometrical parameters—for example, whether the acceleration jet is circular or rectangular and, to a lesser extent, the ratio of the spacing between the acceleration jet and receiver nozzle to the characteristic dimension of the acceleration jet—will also affect the performance. With reference to Figure 4.9, the cutpoint Stokes number is about 0.58 for a rectangular jet virtual impactor with a flow rate f of 10 percent. When a virtual impactor is used as a preseparator, the particle stream that exits the fractionation zone through the receiver nozzle contains the debris that is to be discarded, and the fine particle stream contains the particles that are to be subjected to subsequent processing (i.e., concentration, collection, or analysis). Were it not for wall losses in a virtual impactor, the concentration of aerosol particles with sizes smaller than the cutpoint size in the fine particle stream would be approximately equal to the concentration of that size fraction in the sampled airstream. Wall losses in the fractionation zone that are based on numerical predictions (see Figure 4.9) are on the order of a few percent; however, virtual impactor preseparators with cutpoints of 30 μm AD have not been developed, and there may be other losses (e.g., gravitational) in the flow components approaching and leaving the virtual impactor that are not captured by the model used for calculating the data in Figure 4.9. Thus, the losses of the large particles (30 μm) could be significantly higher than those illustrated in Figure 4.9. Collector Technology As noted previously, a complete sampling system consists of inlet, preseparator, concentrator, and collector. Preseparators are discussed above. Because of the higher bioagent concentration anticipated with indoor attacks, sample concentrators will likely not be required, and discussion of concentrators is deferred to the outdoor monitoring section, below. The committee thus moves on to a discussion of collection technologies. Detection and identification systems that analyze samples in liquids, on dry surfaces, or in an ambient airstream require different collection technologies. Each of these is discussed below. Aerosol-to-Hydrosol Transfer For bioaerosol detect-to-warn systems that employ detectors that analyze samples in the liquid state, the sampler must efficiently transfer the aerosol into the hydrosol state. There are several devices that have been developed to accomplish aerosol-to-hydrosol transfer in a batch mode (e.g., the Spincon from Sceptor Industries in Kansas City, Missouri). However, for batch systems, the time constant associated 27   40 CFR Part 53; U.S. Environmental Protection Agency. 1997. National ambient air quality standards for particulate matter; final rule. Federal Register 62:38651-38752. 28   F.J. Romay, D.L. Roberts, V.A. Marple, B.Y.H. Liu, and B. Olson. 2002. A high-performance aerosol concentrator for bioaerosol agent detection. Aerosol Sci. Tech. 36:217-226.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases could navigate through the sampling system to the detector, or lint that could be inadvertently deposited on critical internal surfaces and cause biased sampling. Such sampling systems are widely used in laboratories where nuclear materials are handled or processed. The Alpha Sentry, made by Canberra Industries in Meridian, Connecticut, is used to continuously monitor for the presence of alpha-emitting radionuclide aerosol particles. A similar apparatus, but with no entrance screen, is manufactured by Thermo Eberline in Santa Fe, New Mexico. Both of these systems allow transmission from the room environment to a collection filter of at least 80 percent of 10 μm AD aerosol particles. Fractionation of larger particles occurs because of a sharp 90-degree turn that the airflow must negotiate to enter a small (5 mm) gap between the collection filter collector and a planar detector. Such a bend would not be needed in a bioaerosol sampler. If an inlet is to be fitted with a fractionator with a prescribed cutpoint, either a cyclone or an impactor (classical or virtual) could be used to scalp the larger particles. An impactor is more compact than a cyclone and would generally be preferred. The inlet should provide a sample flow rate of at least 30 liters per minute, which will transport aerosol particles in the range 1 to 30 μm AD to a collector such as an aerosol-to-hydrosol stage or a dry collector, depending on the requirements of the detector. For most occupied environment applications, the use of an aerosol concentrator would not be warranted since the device would increase size, noise, and cost. As noted in the discussion on ductwork sampling, however, some of the contemporary detectors are designed to be coupled directly with concentrators. The near-real-time radionuclide aerosol detection systems of the nuclear industry have focused on aerosols that will penetrate to the thoracic region of the human respiratory system and not on particles with sizes as large as 30 μm AD. It is anticipated that new inlet designs will need to be developed to accommodate sampling of these larger bioaerosol particles. An important aspect of sampling in the occupied environment is the optimal placement of samplers. In general, the greater the number of samplers, the higher the probability of detection; however, the availability of resources for system procurement and system operation dictate that the number of samplers in any occupied environment must be small. The U.S. Nuclear Regulatory Commission38 and the U.S. Department of Energy39 have developed guidelines for workplace sampling that suggest that consideration be given to airflow patterns when selecting locations for sampler placement. However, the advice is not specific but simply states that tracer tests (e.g., visible smoke) should be performed to determine the airflow patterns prior to selecting sampling locations. A study was conducted at Los Alamos National Laboratory40 on the placement of continuous air monitors in a nuclear laboratory, where the monitors are designed to provide the detect-to-warn function for workers who might be exposed to transuranic aerosols as a result of accidental releases. In a room with 6 to 12 air changes per hour and with 12 detectors (simulated by optical particle counters), the times for detection of an aerosol puff were on the order of 2 minutes, even though the optical particle counters had response times of 10 seconds. Clearly, in buildings where the density of samplers would be much less than that of these experiments, judicious sampler placement is paramount for detection to occur within a several-minute time frame. 38   U.S. Nuclear Regulatory Commission. 1993. Air sampling in the workplace. NUREG-1400. Washington, D.C.: Office of Nuclear Regulatory Research. 39   U.S. Department of Energy. 1999. Air monitoring guide for the use with Title 10 CFR 935, Occupational Radiation Protection DOE G441.1-8. Washington, D.C.: Office of the Assistant Secretary for Environmental Safety and Health. 40   J.J. Whicker, P.T. Wasiolek, and R.A. Tavani. 2001. Influence of room geometry and ventilation rate on airflow and aerosol dispersion: Implications for worker protection. Health Physics 82:52-63. J.J. Whicker, Los Alamos National Laboratory. Presentation to the committee on April 15, 2002.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases SAMPLING BIOAEROSOLS FROM AMBIENT AIR The requirements for outdoor (ambient) air samplers are considerably different from those for building ducts and occupied environments because the concentrations of biological agents to be detected in outdoor release scenarios are likely to be significantly lower those in than indoor release scenarios. Suppose one agent-containing particle per liter of air (ACPLA) is to be detected in the ambient environment with a hydrosol-based detector that can sense the presence of 103 organisms per milliliter of liquid. If the air sampling system were to provide a hydrosol flow rate of 0.5 milliliter per minute, then the air sampling rate would need to be at least 500 liters per minute. This flow rate is about 10 times greater than that which would typically be considered for building applications and simply reflects the different minimum detection levels anticipated for buildings and the ambient environment. In general terms, an ambient sampler will have three major components: an inlet that accommodates a preseparator, an aerosol concentrator, and a collector. For some designs, the latter two functions are combined through use of a wetted-wall cyclone. Two weather-related phenomena must be dealt with in the design of inlets. Precipitation must be excluded and the sampling performance must not be degraded by variations in wind speed. Entry of rain can be precluded by use of a sloped roof at the aerosol entrance section, as is illustrated in Figure 4.12, which shows a commercial inlet41 that was modified by EPA to minimize entrance of rain into the sampler body.42 This inlet has a bug screen, a flow decelerator in the shape of an inverted cone, and an internal impactor with a cutpoint of 10 μm AD.43 Although the inlet shown in Figure 4.12 is designed for a sampling flow rate of 16.7 liters per minute, there are other commercially available inlets with much larger flow rates, up to 1,130 liters per minute.44 Snow can be eliminated from the sampled flow stream in the inlet by the internal fractionator, but fog droplets, which can have a substantial fraction of mass in the size range of interest,45 will be transmitted through the inlet and flow into other regions of the sampling system, where they will either be collected or perhaps evaporate. Because the total water content of fog is typically 0.1 to 0.2 grams per cubic meter,46 considerable water could be collected in the preseparator, by the collector, or in other regions of the sampler. The performance of some omnidirectional inlets can be significantly affected by wind speed. Inertially affected aerosol particles can be lost in an inlet as a result of curvature of the streamlines when the flow turns the corners to pass through the rain-protective elements and when the aerosol takes on a vortex flow pattern. Because the phenomena that induce losses are related to airstreamline curvature, the losses can be characterized as Stokes number dependent. The Stokes number is proportional to the square of a particle's size and to velocity (e.g., wind speed) and is inversely proportional to the characteristic dimension of the part of the inlet system under consideration (e.g., the diameter of the flow exit port in the inlet). Almost all of the development effort related to ambient inlets has been directed toward the goal of engineering systems that will mimic the performance of the extrathoracic region of the human respiratory system by stripping particles with sizes larger than 10 μm AD. Little information is available on the performance of inlets for particles of 30 μm AD. McFarland et al.47 tested an EPA-approved total 41   Model 246b, ThermoAndersen, Inc., Smyrna, Ga. 42   M.P. Tolocka, T.M. Peters, R.W. Vanderpool, F.L. Chen, and R.W. Weiner. 2001. On the modification of the low flow-rate PM10 dichotomous sampler inlet. Aerosol Sci. Tech. 34:407-415. 43   B.Y.H. Liu and D.Y.H Pui. 1981. Aerosol sampling inlets and inhalable particles. Atmos. Env. 15:589-600. A.R. McFarland and C.A. Ortiz. 1984. Characterization of Sierra-Andersen PM-10 inlet model 264B. Report 4716/02/02/84/ARM. College Station: Texas A&M University Air Quality Laboratory. 44   Model 1200 inlet, ThermoAndersen, Inc., Smyrna, Ga. 45   Fuchs, 1964. See note 24 above. J.H. Seinfeld and S.N. Pandis. 1998. Atmospheric Chemistry and Physics: From Air Pollution to Climate Change. New York: John Wiley & Sons. 46   Seinfeld and Pandis, 1998. See note 45 above. 47   A.R. McFarland, C.A. Ortiz, and C.E. Rhodes. 1980. Characterization of sampling systems. The Technical Basis for a Size

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases FIGURE 4.12 (a) An omnidirectional inlet with a bug screen and an internal fractionator, and (b) sampling performance of the inlet, flow rate equals 16.7 liters per minute. suspended particle sampler,48 which does not have an internal fractionator, and reported that the cutpoint was approximately 30 μm AD at a wind speed of 8 kilometers per hour. To accommodate a particle cutpoint as large as 30 μm AD, new inlet designs need to be developed. Performance degradation of an inlet associated with larger particle sizes can be reduced by use of larger dimensions, or perhaps by use of a more effective means for decelerating the airflow inside the inlet before the point where the flow becomes entrained in the vortex zone. AEROSOL CONCENTRATORS The most critical component of existing ambient sampling systems is a concentrator that will enhance the concentration in the 1 to 10 μm AD range by as much as 1,000. SCP Dynamics of Minneapolis has developed a family of aerosol concentrators that are based on the principle of virtual impaction. A typical system is the SCP Model 1001, which samples air at a flow rate of 1,000 liter per minute and concentrates particles in the range of 2.5 to 10 μm AD into a flow stream of 1 liter per minute. MSP Corporation, also of Minneapolis, developed a virtual impactor that samples at a flow rate of 300 liters per minute and concentrates aerosol particles in the range of 2.5 to 10 μm AD into a coarse aerosol particle exhaust flow rate of 1 liter per minute.49 The JBPDS has a cyclone concentrator that samples air at a flow rate of 780 liters per minute and concentrates the particulate matter into a hydrosol flow of 1 milliliter per minute, which is then subjected to near-real-time bioanalyses. In the 1960s, Aerojet-General Corporation in Rancho Cordova, California, developed a 1,000 liter per minute vertically oriented glass cyclone that uses an upstream water spray to wet the internal surface of the cyclone. The hydrosol is then aspirated from the bottom of the cyclone.50     Specific Particulate Standard. Parts I and II. Pittsburgh, Pa.: Air Pollution Control Association. 48   40 CFR Part 50, Appendix B. 49   F.J. Romay, D.L. Roberts, V.A. Marple, B.Y.H. Liu, and B. Olson. 2002. A high-performance aerosol concentrator for bioaerosol agent detection. Aerosol Sci. Tech. 36:217-226. 50   J.M. Macher. 1999. Bioaerosols: Assessment and control. Cincinnati, Ohio: American Conference of Industrial Hygienists.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases If the particle size range of interest for detect-to-warn ambient aerosol sampling systems is taken to be from 0.5 to 30 μm AD, effort will need to be devoted to develop new inlet and concentrator systems that will accommodate this extended range. Although both the air sampling flow rate and the flow rate of an enriched aerosol or a hydrosol stream are driven by the application and the performance of the detector, these numbers will typically be approximately a sampling flow rate of 500 liters per minute and a coarse particle flow rate of 1 liter per minute or a hydrosol flow rate of approximately 1 milliliter per minute. Performance of Virtual Impactor Aerosol Concentrators The effectiveness of a concentrator can be considered in terms of two parameters: the fractional penetration or efficiency and the concentration factor. For some detection concepts there can be problems with background contaminants in the submicrometer range, e.g., the fluorescence of diesel particulate matter. Because a cyclone essentially collects particles larger than the cutpoint size and exhausts particles smaller than the cutpoint with the airstream, the cyclone offers the advantage that it not only concentrates the particles of interest (sizes larger than the cutpoint) but also eliminates small background contaminants. On the other hand, in a virtual impactor, the concentration of particles with sizes smaller than the cutpoint is approximately the same in the coarse particle flow stream as it is in the ambient air. However, if the coarse particle flow stream is collected by a classical impactor or a cyclone prior to analysis, small background particles will be eliminated. For a virtual impactor concentrator, the fractional efficiency is the ratio of the particle mass flow rate associated with a given small interval of particle size in the exhaust stream of a concentrator to the particle mass flow rate for the same size interval in the sampled airstream. Inadvertent losses of aerosol particles on the internal walls of a concentrator reduce the fractional penetration. In the case of a cyclonic concentrator, particle collection on the wetted internal walls of the device is the desired goal, so fractional efficiency is used to characterize performance. Fractional efficiency is the difference in value between unity and the fractional penetration. The concentration factor is the ratio of aerosol mass concentration associated with a given small interval of particle sizes in the exhaust stream of the concentrator to the concentration of that same size interval in the inlet stream. Were it not for internal losses of aerosol particles, an ideal concentrator would produce a concentration factor for the particle size range of interest (particles larger than the cutpoint) that would be equal to the ratio of the volumetric flow rates of the inlet and outlet (containing particles of interest) flow streams. Haglund et al.51 tested an SCP Model 1001 in an aerosol wind tunnel and reported the fractional penetration. Peak penetration of 78 percent is associated with a particle size of about 4 μm AD. Lower efficiencies are associated with smaller particles because of the fractionation characteristics of a virtual impactor, and lower efficiencies are associated with larger particles because of internal wall losses, which primarily occur in the nozzles and fractionation zones of the latter stages of the multistage system. This apparatus was designed to concentrate aerosol particles in the range 2.5 to 10 μm AD, and the mean penetration over that size interval is 48 percent. Because the ratio of flow rates for the device is 1,000:1 and the mean penetration is 48 percent, the mean concentration factor over the range 2.5 to 10 μm AD is 480. If an aerosol-to-hydrosol stage were added to the concentrator, it would transfer 25 percent of the particles (wall losses of 75 percent) in the 1 liter per minute flow rate at the exit port to a liquid flow rate of 0.5 milliliters per minute, and the overall concentration factor would be 240,000. From the manufacturing point of view, it becomes increasingly difficult to fabricate virtual impactors as the cutpoint is reduced. Consider a slit virtual impactor for which the cutpoint is to be reduced by varying the slit width. Total flow rate would be held constant by maintaining a constant velocity, achieved by adjusting the slit length to compensate for changes in the slit width. The cutpoint is primarily dictated 51   J.S. Haglund, S. Chandra, and A.R. McFarland. 2002. Evaluation of a high volume aerosol concentrator. Aerosol Sci. Tech. 36:690-696.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases by the Stokes number, and, as shown in Equation 1, the required slit width varies as the square of cutpoint size. To achieve a cutpoint of 2.5 μm AD will typically require a slit width of about 0.67 mm (0.03 in.); however, a cutpoint of 0.7 μm would imply a slit width of about 0.06 mm (0.002 in.). The latter width is extremely difficult to achieve with contemporary machining techniques. Bergman52 patented a virtual impactor with multiple parallel slits that could be made from silicon, which has a cutpoint of 2.5 μm AD, and Haglund et al.53 reported on a circumferential slit virtual impactor with a cutpoint of 1 μm AD. The circumferential impactor is fabricated on a lathe where close tolerances can be achieved; however, a cutpoint of 0.7 μm AD pushes the current limits of machining. Also, for virtual impactors with small slit or nozzle dimensions, there is a problem of cleaning, particularly for the latter stages of multistage devices where the concentration at the entrance plane is high.54 Indeed, if the design cutpoint were 0.5 μm AD, the slit widths would be comparable in size to the upper end of the 0.5 to 30 μm range, which suggests the nozzle could be plugged by some of the very particles it is designed to transmit. Performance of Cyclonic Concentrators The Aerojet General cyclone is designed to sample air at a rate of 1,000 liters per minute and to concentrate the particulate matter into a liquid flow rate of 1 milliliter per minute, thereby providing an ideal aerosol-to-hydrosol concentration factor of 106. Fractional efficiency for the cyclone is about 50 percent over the range of 1 to 10 μm AD.55 Correspondingly, the average concentration factor over that size interval is approximately 500,000.56 However, those test data were collected by comparing the aerosol concentration in the outlet airflow stream to that in the inlet airflow stream and do not provide information on whether the particles were actually transferred to the water. The water collection fluid in the cyclone does not uniformly wet the internal walls of the cyclone but rather tends to form rivulets on the internal cyclone wall, so the calculated concentration factor would be an overestimate of the hydrosol collection efficiency. There are two important differences between existing virtual impactor and cyclonic concentrators. First, a virtual impactor concentrates particles with sizes greater than approximately its cutpoint. However, it also transmits smaller sizes at concentration values that are essentially equal to those in the sampled air. In contrast, the collection efficiency of a cyclone tends to zero for small particles. Second, and more important, is that with present technology, cyclones with submicrometer cutpoints are practical; however, virtual impactors with cutpoints smaller than about 2 μm AD, which have low internal wall losses and are easily cleaned, are not yet achievable on a commercial scale. Novel Concentrators Research is being conducted on two nonclassical approaches to aerosol concentration, ultrasonic and electrostatic. The ultrasonic approach involves subjecting an aerosol stream to a standing sound wave57 that causes the aerosol particles to concentrate at the nodes. The physics of acoustic motion of aerosol particles has been studied in the context of ultrasonic coagulation;58 however, the application to 52   W. Bergman. June 11, 2002. Low pressure drop, multi-slit virtual impactor. U.S. Patent 6,402,817. 53   J.S. Haglund, S. Hari, H. Irshad, Y.A. Hassan, and A.R. McFarland. 2002. Bioaerosol sampling and collection. Presentation at Scientific Conference on Obscuration and Aerosol Research. Aberdeen Proving Ground, Md., June 25. 54   J.S. Haglund, S. Chandra, and A.R. McFarland. 2002. Evaluation of a high volume aerosol concentrator. Aerosol Sci. Tech. 36:690-696. 55   A.R. McFarland and H.W. Davis. 1998. Wetted wall biological aerosol sampling system. Report 4716/01/03/98ARM. College Station, Tex.: Texas A&M University. 56   McFarland and Davis, 1998. See note 55 above. 57   M.J. Anderson, R.S. Budwig, K.S. Line, and J.G. Frenkel. 2002. Ultrasonic concentration of aerosol particles. Scientific Conference on Obscuration and Aerosol Research, Aberdeen Proving Ground, Md., June 25. M. McDonnell, Dstl. Presentation to the committee on June 12, 2002. 58   Fuchs, 1964. See note 24 above.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases aerosol concentrators is still of an investigatory nature. Electrostatic concentration is based on first imparting a unipolar charge on the aerosol particles and then focusing the particles in an electric field.59 Both numerical and experimental techniques are being used to test the concept at the United Kingdom Porton Down facility. This concept has the potential to concentrate aerosols with a low expenditure of power, because the energy for causing particle motion is input directly to the aerosol particles rather than to the airstream, and it would not require special heating during cold weather. Also, electrostatic devices have the potential to concentrate submicrometer-sized aerosol particles. Ideal Power to Draw Air Through a Concentrator The consumption of power is an element of concern, particularly for those detection systems placed in the field and operated with portable electrical generation equipment. Experience with the JBPDS system has been that the aerosol concentrator uses more power than any other subsystem.60 However, even if a concentrator were to be employed in an occupied environment, in general the higher the power consumption, the noisier the system, so for either field or indoor applications minimization of power consumption should be a goal. The logistical requirements for placement and operation of a remote sampler can be dramatically affected by the power requirements. The sampler of the JBPDS requires 400 watts during nonfreezing conditions at a flow rate of 780 liters per minute,61 and it would require an additional 520 watts if the entire sampled airstream needs to be heated to prevent freezing when the outdoor air is at a design operational condition of −28°C. In the field, the JBPDS is connected to a 3-kilowatt generator that weighs approximately 90 kilograms (200 pounds). If the power requirements were less, not only would the weight of the generator be less, but also the generator would have a lower fuel consumption and thus lower logistical requirements. The actual power requirement to draw air through the JBPDS concentrator is considerably in excess of the ideal power requirement, where the latter is the power to overcome frictional losses in the device. A specific measure of the ideal power is the pressure drop of the flow as it passes through a concentrator, where the pressure drop can be considered as equivalent to ideal power divided by sampling flow rate. The pressure drop value affords an opportunity to compare the ideal power consumption of two concentrators in a manner that is independent of airflowrate through the concentrator. For the JBPDS system, the pressure drop is 7.0 kPa (28 inches of water), which for a flow rate of 780 liters per minute, would require an ideal power of about 90 watts. The large difference between the actual (400) and ideal power consumption is primarily due to inefficiencies in the blower-motor combination. A more efficient blower-motor is needed for ambient sampling systems. The ideal power requirement of a cyclonic sampling system depends strongly on the design cutpoint. Suppose a cyclone with a cutpoint of 2 μm AD is used to sample the ambient air at a flow rate of 500 liters per minute. If it is desired to sample that same flow rate with a geometrically similar but smaller cyclone that has a cutpoint of 0.5 μm AD, the ideal power would increase from 190 pascals (0.75 inches of water) to 5.1 kilopascals (20 inches of water). Multiple small cyclones operated in parallel could be used to reduce the pressure loss; however, that would complicate the aerosol-to-hydrosol transfer process. The pressure loss in virtual impactors can be kept low by use of extended slit lengths to accommodate both desired flow rate and cutpoint. In general, the pressure loss varies as the square of the velocity at the exit plane of the acceleration jet, but if the velocity is held approximately constant as the design cutpoint is reduced, the pressure drop will not be significantly affected. 59   S.R. Preston, T.G. Foat, M.D. Walker, and J.M. Clark. 2002. The design for an electrostatic aerosol collector. Joint Service Scientific Conference on Chemical and Biological Defense Research. Hunt Valley, Md. November 21. 60   T. Moshier, U.S. Army. Presentation to the committee on December 18, 2001. 61   R.S. Black and M.J. Shaw. 2002. Development of the wetted wall cyclone for the Joint Biological Point Detection System. Scientific Conference on Obscuration and Aerosol Research. Aberdeen Proving Ground, Md., June 25.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases The pressure loss across virtual impactors depends on another consideration. The airstream from the acceleration jet reaches near-stagnation conditions at the entrance of the receiver nozzle, because only a small part (10 to 20 percent) of the total flow is drawn into that port. As a consequence, there will be recovery of pressure at the inlet of the coarse particle receiver, which reduces the pressure loss to near zero for that portion of the stream. Thus, as an approximation, the power loss in a virtual impactor (whether the impactor jets are circular or rectangular) is only the loss associated with the major, or fine-particle flow, of each stage. Power Consumption to Prevent Freezing of Liquid Concentrators such as the JBPDS system and the Aerojet General cyclone, which combine aerosol concentration with aerosol-to-hydrosol transfer, must have provisions for prevention of liquid freezing if the concentrators are used in ambient sampling applications. The simplest approach to precluding freezing is to heat the airstream before it enters the concentrator; however, that can require considerable power. If the design outdoor temperature is −28°C and the airstream is heated to 5°C, the power requirement is 2 to 3 watts per liter per minute of airflow. The JBPDS and the Aerojet General cyclones would require 520 and 670 watts, respectively, to accomplish this heating. In contrast, virtual impactor concentrators, when used in series with aerosol-to-hydrosol transfer stages, require little extra power to prevent freezing of the liquid in the AHTS. In the case of a system such as the SCP Model 1001 concentrator, which has an airflow rate of 1 liter per minute at the exit port, only 2 to 3 watts would be needed to heat the airstream from −30°C to +5°C. AEROSOL-TO-HYDROSOL TRANSFER STAGES SCP Dynamics, Inc., has developed a batch-type AHTS that is used with a virtual impactor that has an aerosol inlet flow rate of 1,000 liters per minute and a coarse-particle flow rate of 20 liters per minute. The AHTS collects the particulate matter in the coarse-particle airflow into 40 milliliters of liquid. This virtual impactor and AHTS both have cutpoints of about 2.5 μm AD. There are wall losses in both the virtual impactor and the aerosol-to-liquid collector, precluding efficiency values much greater than about 25 percent over the range 2.5 to 10 μm AD.62 If the penetration values can be increased, it may be possible to modify the SCP system to operate on a continuous flow basis. Two AHTS devices utilize circular jet impactors to deposit the particles from a 1 liter per minute aerosol flow rate into a liquid film that flows at a rate of 0.5 milliliters per minute. The cutpoints of the two devices are 0.8 and 2.5 μm AD. The liquid film forms on a porous surface through which the liquid is transpirated. FIGURE 4.13 Collection efficiency of a combined concentrator using aerosol-to-hydrosol transfer stages (AHTS). The system is the JBPDS main sampler, which has an airflow rate of 780 liters per minute and a liquid flow rate of 1 milliliter per minute. 62   Kesavan and Doherty, 2001. See note 32 above.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases Collection efficiency as a function of particle size is shown in Figure 4.13 for the AHTS with the 0.8 μm AD cutpoint. The data in Figure 4.14 are for two types of experiments: One type was based on a comparison of the aerosol concentration of monodisperse liquid droplets upstream and downstream of the AHTS; the second was based on a comparison of the number of solid polystyrene aerosol particles collected per unit of time in the output liquid with the number of particles per unit time that enter the device in the aerosol state. Results of the two experiments compare favorably and show that average efficiency over the range of interest (1 to 10 μm AD) is greater than 90 percent. The temporal response of the 0.8 μm AD cutpoint AHTS is shown in Figure 4.14, where the concentration of 2.3 μm polystyrene beads in the liquid was monitored as a function of time. The AHTS was subjected to a step increase of aerosol at a time of 0 minutes and to a step decrease at an elapsed time of 16 minutes. The time constant, which is the time FIGURE 4.14 An aerosol-to-hydrosol transfer stage: (a) design, (b) collection efficiency, and (c) time response.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases required for the hydrosol concentration to reach 63.2 percent of its equilibrium value following a step change, is about 0.8 minutes for either a step increase or a step decrease. FINDINGS AND RECOMMENDATIONS Below, the committee's findings and recommendations for sampling of bioaerosols in the context of detect-to-warn systems are presented for the three principal applications discussed in this chapter: ductwork, occupied environments, and ambient sampling. Ductwork Sampling The Department of Energy (DOE) confronted the need to continuously monitor emissions of radionuclides from stacks and ducts that could potentially emit significant amounts of radionuclides. To fulfill this requirement, the concept of single-point representative sampling was developed, and that concept is embodied in an ANSI standard method for extractive sampling of emission points in the nuclear industry. Finding 4-1: The current ANSI standard provides methodology for extractive sampling of ducts; however, that work focused on an upper size cutpoint of 10 μm AD, whereas for bioaerosol sampling the upper size cutpoint of interest may be 30 μm AD. Recommendation 4-1: The methodology given in ANSI N13.1 should be reviewed with reference to extraction of bioaerosol samples from building ductwork, and a comparable document should be prepared that would provide guidance to designers and users of detect-to-warn systems. Studies should be carried out to determine the advantages and disadvantages of extending the sampling capability to a cutpoint of 30 μm AD. Finding 4-2: While the ANSI standard stipulates that samples must be extracted from locations where the velocity and contaminant concentration profiles are uniform, it does not provide guidance for selecting such locations on an a priori basis. Recommendation 4-2: Numerical and experimental studies should be conducted to develop criteria for nozzle siting. Finding 4-3: While work has been done on developing the individual components of bioaerosol sampling systems (e.g., nozzles, transport systems, and collectors), integration of the systems still needs to be accomplished. Recommendation 4-3: The development of integrated, turnkey sampling systems should be supported for aerosol particles that are automatic, robust, and require little maintenance. Studies should be conducted on the advantages and disadvantages of developing systems that would extend the range for extraction, transport, and collection of aerosol particles, from 1 to 10 μm AD to 0.5 to 30 μm AD. Finding 4-4: The time constants of current continuous flow aerosol-to-hydrosol devices are on the order of 1 minute or more, which is too long for DTW needs. Current ATHT stages have flow rates as low as 0.5 milliliters per minute; however, that is probably higher than would be desired from some DTW applications, especially those intended for use in the field. Recommendation 4-4: Developmental efforts should be supported for ATHT stages that have shorter time constants and smaller liquid flow rates.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases Occupied Area Sampling Area sampling for protection of the occupied environment is a greater challenge at the present time than ductwork sampling. Because the probability of detection of a bioaerosol using samplers in an occupied area increases with the number of samplers, such devices must be inexpensive, unobtrusive, robust, require little maintenance, and not be prone to generating false alarms. The DOE uses near-real-time samplers in laboratories where radionuclides are handled, and the experience gained in that program should be of benefit in efforts to monitor occupied environments for bioaerosols. Finding 4-5: There is currently no bioaerosol sampling system that is compatible with the occupied environment application. Recommendation 4-5: In the long term, after low-cost, reliable detection systems are developed, effort should be directed to developing compact, fully automatic, low-cost, reliable, and unobtrusive area sampling systems. Studies should be conducted on the advantages and disadvantages of developing equipment with representative sampling of particles in the range 0.5 to 30 μm AD, where the values of 0.5 and 30 are to be regarded as cutpoint sizes. Finding 4-6: Guidance is lacking on strategies for the placement of samplers to obtain the optimum trade-off between detection time and number (cost) of samplers in occupied environments. Recommendation 4-6: Guidelines for optimum sampler placement should be developed for occupied environment applications. Ambient Sampling The military has undertaken an extensive effort to provide troops in the field with bioaerosol detection systems, which led initially to BIDS and now to JBPDS. The sampling and detection systems contain an inlet (with preseparation capabilities), a concentrator, and a detector. Finding 4-7: The aerosol size range of most present interest is 1 to 10 μm AD. Recommendation 4-7: A study should be conducted to evaluate the merits of considering sampling systems capable of detecting a 0.5 to 30 μm size range of particles. Finding 4-8: The performance of current design ambient sampling inlets can vary with wind speed. Recommendation 4-8: Design models should be developed that will allow users to construct optimized inlets that will provide robust performance in spite of variations in meteorological conditions and that will allow selection of a desired cutpoint size. Finding 4-9: The currently available virtual impactor concentrators have cutpoint sizes ≥2 μm AD, and it may be desirable to reduce the cutpoints to 0.5 μm AD. However, it may not be possible to accomplish this, because very small slit widths (or jet diameters for circular jet virtual impactors) are needed. Cyclone concentrators with cutpoints of 0.5 μm AD can be constructed with currently available technology; however, a single cyclone will consume considerable power and it will require more heating to prevent liquid freezing in cold weather. Also, the cyclone does not permit delivery of an aerosol sample to the detector, i.e., it is only compatible with hydrosol delivery. New concepts such as acoustic and electrostatic devices are being investigated; however, considerable work needs to be done before practical investments are realized.

OCR for page 46
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases Recommendation 4-9: High-capacity, easily cleanable, inexpensive, robust devices with submicrometer cutpoints and with the ability to transmit or collect (cyclone) particles as large as 30 μm AD should be developed. Use of new materials and fabrication techniques should be considered. Finding 4-10: An aerosol concentrator can require the expenditure of considerable power. Recommendation 4-10: The development of more efficient blowers and concentrators should be supported.