The Filtration of Air

THE FILTRATION OF AIR

Removal of particulate matter from air together with control of humidity, temperature, and distribution comprise air conditioning. Solid and liquid par-ticles are most commonly arrested by filtration, although other methods, such as electrostatic precipitation, cyclones, and scrubbers, are used in some circum-stances. The objective may be simply the provision of comfortable and healthy conditions for work or may be dictated by the operations proceeding in the area. Some industrial processes demand large volumes of clean air.

In this section, we shall be concerned mainly with air filtration, the objective of which is the reduction in number or complete removal of bacteria. This is applied, with varying stringency, to several operations associated with pharmacy. Where sterilization is the objective and the presence of inanimate particles is of secondary importance, other methods, such as ultraviolet radia-tion and heating, must be added.

Bacteria rarely exist singly in the atmosphere but are usually associated with much larger particles. For example, it has been shown that 78% of particles carrying Clostridium welchii were greater than 4.2 x 10^-6 m. The average diameter exceeded 10 x 10^-6 m. On this basis, it has been suggested that air filters that are 99.9% efficient at 5 x10^-6 m are adequate for filtration of air supplied to operating theaters and dressing wards (Williams et al., 1961). On the other hand, filters used to clean air supplied to large-scale aerobic fermentation cultures must offer a very low probability that any organism will penetrate during the process. This became important in the deep-culture production of penicillin when the ingress of a single penicillinase-producing organism could be disastrous. Similarly, stringent conditions are laid down for the supply of air to areas where sterile products are prepared and handled.

The Mechanism of Air Filtration

A theoretical foundation for the filtration of air by passage through fibrous media was laid in the early 1930s by studies of the flow of suspended particles around various obstacles. In studies of the filtration of smokes (Suits, 1961; Hinds, 1999), it has been shown that a number of factors operate simultaneously in the arrest of a particle during its passage through a filter, although their relative importance varies with the type of filter and the conditions under which it is operated. These factors may be listed as follows:

·       Diffusion effects due to Brownian movement

·       Electrostatic attraction between particles and fibers

·       Direct interception of a particle by a fiber

·       Interception as a result of inertial effects acting on a particle and causing it to collide with a fiber

·        Settling and gravitational effects

FIGURE 11.6 Inertial capture of a particle by a fiber.

Air filters operate under conditions of streamline flow, as indicated by the streamlines drawn around a cylindrical fiber shown in cross section in Figure 11.6. It was assumed that capture of a particle takes place if any contact is made during its movement around the fiber. Once capture occurs, the particle is not re-entrained in the airstream and is deposited deeper in the bed. Support for this assumption has been found by using an atomized suspension of Staphylococcus albus and spores of Bacillus subtilis (Terjesen and Cherryl, 1947). Nevertheless, some fiber filters are treated with viscous oils, presumably to make capture more positive and to reduce re-entrainment.

If a particle remains in a streamline during passage around the fiber, capture will occur only if the radius of the particle exceeds the distance between streamline and fiber, a dimension dependent on the diameters of the particle and the fiber. This mechanism, termed “capture by direct interception,” is independent of the air velocity except in so far as the streamlines are modified by changes in air velocity.

Deviation of particles from streamlines can occur in a number of ways (Hinds, 1999; Reist, 1993). The chance of capture will increase if Brownian movement causes appreciable migration across streamlines, an effect only important for small particles (<5 x 10^-7 m) and low air speeds, when the time spent in the vicinity of a fiber is relatively large. These conditions also apply to capture, which is the result of electrostatic attraction.

The inertial mechanism depends on the mass of the particle, the fiber diameter, and the velocity of approach. The particle deviates from the streamline and follows the broken line shown in Figure 11.6. Capture occurs if the devia-tion, which increases as the mass and velocity of the particle increase, brings the particle into contact with the fiber.

The simultaneous operation of mechanisms, at least one of which demands low air speeds and fine particles for effectiveness and another which requires large particles traveling at high speeds, suggests that maximum penetration could occur at an intermediate air speed. Conversely, there is, for any given conditions, an optimal particle size for which the combined filtration effects are a minimum and penetration is a maximum. The former was confirmed with a variety of inanimate aerosols. A diagrammatic representation of the interaction of mecha-nisms was also given, and this is reproduced in Figure 11.7.

Similar effects were demonstrated for bacterial aerosols (Humphrey and Gaden, 1956). Estimated the efficiency with which a glass fiber mat collected B. subtilis spores atomized as particles just over 1 mm in radius. The results are presented in Figure 11.8. A theoretical approach to the removal of industrial dusts has been developed (Stairmand, 1950; Fuchs, 1964).

FIGURE 11.7 Interaction of the mechanisms of particle arrest.

FIGURE 11.8 Effect of airstream velocity on the removal of bacterial spores by a filter.

The Design, Operation, and Testing of Air Filters

Granular beds, fibrous media, and “absolute filters” prepared from cellulose and asbestos are used for high-efficiency air filtration. With fibrous and granular filters, the fractional reduction in particle content is assumed to be the same through successive incremental thicknesses of the filter. We may, therefore, rewrite equation (2).

where C represents the number of particles entering a section of thickness dx. The constant, k, is a measure of the filter’s ability to retain a particle and is a complex function of fiber diameter, interfiber distance, and operational air velocity. Integration between inlet and outlet conditions gives

The use of this log penetration effect in filter design has been described else-where (Gaden and Humphrey, 1955). If a certain filter thickness is capable of retaining 90% of the entering particles, then if 10⁶ particles enter, 10⁵ will pen-etrate. If six thicknesses are used, then the relation above predicts that only one particle will penetrate. The log penetration effect has been confirmed for fibrous filters and for granular beds (Humphrey and Gaden, 1956; Cherry et al., 1951), respectively. It must be stressed, however, that both fibrous and granular filters present passages very much greater than the fine particles they remove. Abso-lute sterility or absolute filtration at a certain particle size cannot be achieved. However, design variables, such as the fiber diameter, the density with which fibers are packed, the thickness of the filter, and the air speed. For example, these variables may be varied to give air that, for a given input contamination, is, with a high statistical probability, sterile.

In an early study, Terjesen and Cherryl used a bacterial aerosol and a Bourdillon slit sampler to test the suitability of filters for air sterilization (Terjesen and Cherryl, 1947). They showed that 0.075-m slabs of slag wool composed of fibers, most of which were less than 6 x 10^-6 m and compressed to a suitable density, gave sterile air when operated for fifteen days at a face velocity of 0.152 m/sec. A similar efficiency was found for filters composed of glass fibers of similar diameters (Cherry et al., 1951). Resin-bonded filter mats composed of glass fibers, 12 x 10^-6 to 13 x 10^-6 m in diameter, have also been described. A number of these mats assembled to give a filter 0.304 m deep were effective in the removal of bacteria.

Bacteria may be effectively removed by passing air through deep granular beds of activated carbon, alumina, and other materials. Table 11.1 gives data on the efficiency of alumina in a bed 0.381 m deep for the removal of Serratia marcescens from air (Sykes and Carter, 1954). The effect of two design variables, granule size and air speed, is illustrated.

The extremely hazardous nature of radioactive dusts has promoted the design of high-efficiency air filters for use in establishments where such mate-rials are handled. These filters may be used for any application requiring extremely pure air. The evolution of filters that remove 99.995% of particles in the range 1 x 10^-7 to 5 x 10^-7 m has been described by White and Smith (White and Smith, 1960). A medium in paper form was constructed from cellulose and asbestos. This could be pleated round corrugated spacers to give a large filtering area in a relatively small space. A paper composed of very fine glass fibers was later developed, which resisted temperatures of up to 773 K and could, there-fore, be sterilized.

TABLE 1 Removal of Serratia marcescens with a 0.3-m Bed of Alumina Granules

The general object of design in all filters is the virtual certainty of removing the particles under consideration with a medium offering minimal resistance to the flow of air. Unlike liquid clarifiers, air filters become more efficient with time because accumulation of particles restricts the passages through the medium. This deposition causes an increase in the pressure differential required to maintain a given flow rate. When the filter has become laden with a certain amount of dust, it must be cleaned or replaced. The life of high-efficiency air filters may be lengthened by passing the air first through a coarse or “roughing” filter, which removes the larger particles.

The use of bacterial aerosols as tracer organisms to test the efficiency of filters has already been described. Other tests with inanimate dusts are more generally used for the evaluation of filter performance. For general ventilation purposes, two tests are specified. The first determines gravimetrically the capacity of the filter to hold dust and still function satisfactorily. A standard dust of 5 x 10^-6 or 26 x 10^-5 m is passed into the filter until a specified increase in air flow resistance occurs. The second test, which is also applicable to high-efficiency filters, determines the fraction of a methylene blue aerosol that passes through the filter under given conditions. The aerosol is generated by atomizing a 1% aqueous solution of methylene blue. The droplets dry to give a cloud of particles, 90% of which are below 2 x 10^-7 m. The test is, therefore, extremely stringent. The cloud is passed through the filter at a constant rate (10^-3 m³/min) and then through a strip of porous paper, which collects any methylene blue particles that have penetrated. The stain due to the dye, after intensification in steam, is compared with a series of similar stains that correspond to known volumes of unfiltered air. Thus, if 60 x 10^-3 m³ of filtered air gives a stain that matches that produced by 1.2 x 10^-5 m³ of unfiltered air, the penetration is 0.02% (Green and Lane, 1957). Both tests are fully described in BS 2831:1957. An alternative method of evaluating penetration employs a cloud produced by the atomization of a solution of sodium chloride. After passage through the filter, part of the air is passed through a hydrogen flame. The intensity of the sodium flame produced is estimated with a photoelectric cell.