Cooling Towers, Drift, and Legionellosis
CTI Journal, Vol. 31, No. 1
ABSTRACT
By itself, the presence of legionellae in a cooling tower is insufficient to predict the potential for disease transmission because other factors are involved. This paper will describe details about one of the important factors, cooling tower air emissions, by providing a comprehensive technical understanding of drift quantity, droplet distribution, and plume dispersion. By understanding these air emission details, ranges of Legionella bacteria concentration at distances from the tower can be estimated as a function of legionellae concentration in the tower water. This paper will also describe the ecology of the bacteria in cooling towers and the epidemiology of outbreaks attributed to cooling towers. Most importantly, the paper will discuss the correlation of the bacteria-exposure model described in this paper with the incidents of disease from previously studied outbreaks. The quantity of bacteria required to cause disease depends on several factors including the health of the individual and the exposure. A hypothetical example would be a situation where an individual could inhale 1 bacterium a week for fifty weeks with no ill effects, but develops disease when he inhales 50 bacteria in an hour. There is likely some exposure rate (inhalation of X bacteria / time) where the risk of disease may occur; the higher the exposure rate, the more likely the occurrence of disease. The inhalation rate (inhalation is the only means of transmission from cooling towers) depends strongly on two factors: 1) the concentration of bacteria in the ambient air in a particular area and 2) the time spent in that area.
INTRODUCTION
Legionellosis is a form of pneumonia caused from Legionella bacteria being inhaled or aspirated deeply into the lungs. Legionella is quite common in the environment and there are many steps from
‘present in the environment’ to ‘disease’. The accepted prerequisite for infection is the bacteria must be contained in droplets of water less than 5 microns in diameter. Larger droplets would not penetrate deeply enough into the lungs to cause infection. While there is no known infectious dose or alternatively safe level for the bacteria, because of its ubiquity in nature, most researchers believe that an infection requires the inhalation of tens if not hundreds of bacteria.
As might be expected with the widespread distribution of Legionella bacteria, there is a ‘normal’, low incidence of random cases of Legionellosis. With the low background rate, a cluster of disease would be statistically rare. When a cluster has occurred, there is often a specific source identified as the cause of the outbreak. Outbreaks of community-acquired Legionella have been attributed to specific spas, fountains, cooling towers, metal working fluids, misters and other sources.
With all of these sources except cooling towers, a very close proximity to the source was required for exposure. With cooling towers, exposures have been reported several kilometers away from the purported source. The exposure to disease connection is not fully understood. There are likely at least two competing processes occurring in the host: bacteria germination and host immune system response.
In a healthy, non-smoking individual, an exposure of thousands of bacteria is likely necessary before the immune system is overwhelmed; in others the exposure of a few bacteria may be sufficient. It is not the norm for a cooling tower to cause an outbreak of Legionnaires’ disease. There are hundreds of thousands of cooling towers in the US, many if not most containing some level of Legionella bacteria, yet there are only a handful of cooling-tower implicated outbreaks. A person’s exposure to Legionella bacteria from a cooling tower is based on a variety of factors:
1) The drift rate of the tower. Drift is the mechanically aspirated droplets of circulating water that are en trained into the effluent air stream.
2) The volume of air passing through the tower.
3) The dispersal of the exhaust air with ambient air (plume dilution).
4) The time spent in plume/air mix
5) The concentration of Legionella bacteria in the circulating water.
This paper will explore these factors and the effect they have on
risk of exposure.
COOLING TOWER DRIFT
Most tests on a specific tower design show a linear relationship between circulating water flow and drift within normal cooling tower operating air flow. Air flow rate will significantly affect tower drift
only at the extremes of the design. Because of this, drift is typically described as a percentage of circulating water rate. All modern cooling towers are or should be equipped with drift eliminators (DE). The DE force the exhaust air to make sharp turns before exiting.
The momentum of en-trained droplets carries the droplets to the DE surfaces where they coalesce and drip back into the tower. Cooling towers or drift eliminators may be evaluated for drift rates under controlled conditions. The standard test is the “Heated Bead Isokinetic (HBIK) Drift Test Procedure” described in the Cooling Technology Institute code ATC 1401. A portion of the exit air stream is drawn at the same speed and direction (isokinetically) into a collection device. The collection device consists of heated beads. Any drift that is pulled into the tower is dried on the beads.
The tower water has a specific concentration of a tracer element and by measuring the quantity recovered from the beads, the quantity of drift can be determined. Less modern designs for drift eliminators are not as efficient as newer equipment. While an older design might result in drift rates up to 0.02%, all towers constructed in the last few years by the major manufacturers are much better. Typically, for cross-flow designs the drift rate will be less than 0.005% while because of use of higher efficiency eliminators, counter-flow designs routinely achieve 0.001%.
The newer cellular drift eliminators started being used in the late 1980’s with particular model lines changing over about a 15-year period. Around the early 1990’s the 0.005% drift rate for cross-flow towers became standard. The 0.001% drift rate for induced-draft counter-flow became standard also around 1990. For forced-draft counter-flow units, 0.001% didn't become standard until just a few years ago. Prior to the change, cooling tower drifts were typically
in the 0.02% or higher range. A typical 1,000-ton HVAC cooling tower nominally circulates 3,000
gpm water. At nominal conditions, the drift from a 1,000-ton cross-flow tower would be less than 0.15 gpm while the drift from a 1,000-ton counter-flow tower would be less than 0.03 gpm. These values should be routinely achieved by units as they are shipped from the factory. With a well maintained tower, these rates can be sustained for many years. However, there are things that happen in the field which can degrade the eliminators effectiveness.
A partial list of these problems follows:
1) Damaged drift eliminators. UV, hail, and improper handling can all damage drift eliminators. Damaged drift eliminators interfere with exhaust air flow. If there are gaps or holes in
the eliminator, then more air will pass through the open area at high velocity carrying significantly more en-trained water.
2) Clogged drift eliminators. In highly cycled water the entrained droplets contain a high quantity of dissolved solids. This can result in a gradual build up of minerals on the DE. As the minerals build up, the air is blocked in some areas and the air velocity increases in the open areas. As this velocity gets high enough, the amount of entrained water carried from the tower increases.
3) Misaligned or missing drift eliminators. If there are gaps in the eliminators, their effectiveness is severely reduced.
4) Damaged fill. While not as obvious as damaged drift eliminators, damaged or partially clogged fill will change the airflow to the DE and affect their efficiency
5) Obstructed inlet air. For the same reason as in #4, changing the airflow in the tower affects DE efficiency.
6) Water distribution. Improper water distribution may put too much water in one area resulting in very high drift from that part of the tower. Misaligned or over-pressured spray nozzles can also increase the amount of drift.
7) Use of surfactants in the chemical water treatment. By lowering the surface tension of the recirculating water, surfactants can cause water to form very small droplets. These small droplets are more easily carried by the air stream and are less effectively removed by the drift eliminators.
The drift that leaves the tower is in the form of small droplets.
The larger the droplet the more momentum it carries and the more effective the DE. The distribution of water drop sizes in the drift can be measured by water sensitive paper.
The paper is treated so that a droplet impinging on the paper will generate a well defined mark. The size of the stain is related to drop size 2 . This is a less exact method than the HBIK test but provides some information about droplet size distribution. This test is not effective for droplets less than 30 microns in size.
Many studies have been performed on the size of drift particles from a tower. Figure 1 shows the results from nine separate tests on drift eliminators performed by the manufacturer. The cumulative volume of drops is plotted as a function of the diameter of the drop in microns. Below each label on the X-axis is the number of droplets of that specific diameter per milliliter of water.
The drops per ml information is useful to see that, absent clumping, it would be very unusual for a single droplet to contain more than 1 Legionella bacterium even in a heavily contaminated system with a Legionella count per ml of 1,000. Because of this, parametric statistical analysis is valid for considering Legionella dispersion in the plume.
Figure 2 displays results from another droplet distribution test performed on two towers in-the-field 4. Also include in Figure 2 is a line at the 50 micron drop size. A 50 micron diameter particle is
1000 times larger than a 5 micron respirable particle. Droplets larger than this line are at least 1000 times larger than respirable size. Most of the volume of the drift as it leaves the tower is in droplet
sizes too large to be deeply inhaled. The bacteria contained in those droplets can cause disease only if the droplets evaporate to a respirable size before falling to the ground.
CONCENTRATION OF LEGIONELLA IN COOLING TOWER EXHAUST
Cooling towers cool water by evaporation, thereby exchanging both heat and humidity to the air. The amount of air passing through a tower per gallon of water depends on both the design of the tower
and often on the heat load on the tower for towers equipped with multispeed fans.
The design mass flow rate of circulating water to cooling air is usually given as an l/g ratio with the both the liquid and gas amounts given in pounds. A quick review of manufacturer catalogs shows that induced-draft counter-flow towers are designed with an l/g ratio between 1.43 and 2.23.
Induced-draft cross-flow towers are designed with a lower l/g ratio of between 1.34 and 1.50. Both designs produce an equivalent amount of cooled water per fan HP, the counter-flow run lower air volumes at slightly higher pressure than cross-flow towers.
Forced-draft counter-flow towers typically run at range of l/g ratios between 1.10 and 1.65. Using a nominal specific volume of air of 14 cubic feet per pound, Table 1 shows the min and max concentration of drift in the exhaust air at the referenced drift rates. This value is then extrapolated to a nominal time that an individual would need to breathe undiluted cooling tower exhaust to, on average, inhale a single Legionella bacterium.
The values in Table 1 are based on full fan power. It is assumed that the drift rate as a percentage of the circulating water rate does not fall appreciably until the fan rate drops below 50%. Thus for low fan speeds, the concentration of drift, and hence Legionella bacteria, could be up to twice as high as is shown in the table.
In 1988 there was a Legionella outbreak at a Los Angeles retirement home 5 . The investigation of that outbreak identified a below ground forced-draft evaporative condenser as the source of the outbreak. While the specifics of the condenser were not included in the paper, Table 1 contains calculations using a typical evaporative condenser with 1988-style drift eliminators.
The investigation measured Legionella in the tower basin of 9,000 CFU/ml. The investigation also used impinger sampling data to estimate that there were 2.3 CFU/liter of air of Legionella in the condenser exhaust vent. That 2.3 CFU/l agrees very closely to the 2.8 CFU/liter that was calculated using the approach of Table 1.
PLUME DILUTION
Only a very small percentage (on the order of 1%) of the drift as it leaves the cooling tower is of respirable size, and hence able to cause an infection. The remainder of drift and contained Legionella consists of droplets greater than 5 microns.
Thus a person breathing undiluted exhaust air from a well-maintained cooling tower with what might be considered a moderate concentration (100 CFU/ml) of Legionella would need to be in the exhaust much longer (possibly 99 times longer) than the times indicated in Table 1 before he statistically would inhale a single Legionella bacterium deeply into his lungs.
The droplets can evaporate to respirable size once they travel a distance from the tower. When exhaust air leaves a cooling tower it forms a plume. The characteristics of this plume depend on a complex interaction of, among-st others, the following:
1. wind speed and direction
2. buildings and structures down wash
3. buoyant/dense plume behavior
4. gravitational settling
5. droplet evaporation and humidity condensation (phase changes)
6. surrounding terrain
7. ambient temperature and humidity
Because of the complexity of the flows around the cooling equipment we will not focus on the ‘near-field’ plume. In addition, the water droplets in this area are likely too large to inhale deeply into the lungs. We are defining this near-field to extend 20 fan diameters from the base of a tower.
This is the approximate distance that the plume from a ground-level tower would reach the ground under the simplified plume dispersion model shown below. There are several computational fluid dynamic programs that attempt to model plume behavior with varying success^6. Beyond some generalizations, the detailed prediction of cooling tower plume is beyond the scope of this paper.
With the previous caveat, there are some generalities that can be stated. For the basic condition we will assume that people are all located at ground level. The worst case scenario would then be a tower installed on or near the ground. One of three things can happen as the ambient air mixes with the cooling tower exhaust air^7:
1. If the air is very still, the exhaust may climb very high and be dispersed over a very large area before it reaches ground level.
2. If the ambient air is very turbulent, winds greater than 20 mph, the plume will be rapidly diluted and dispersed.
3. The third condition is a steady mild breeze of 5 to 10 mph. This condition is sufficient to bend the plume over and bring it to ground level, yet the plume will maintain some cohesiveness. This is the condition that will most likely bring a person in contact with contaminates from the cooling tower and is the condition that we will discuss in the remainder of this section.
SIMPLIFIED PLUME DISPERSAL MODEL – Under mild winds, plumes will generally expand in both the height and width direction with a typical angle of expansion of approximately 6.0°. A simplistic model of a cooling tower, sitting on the ground with a unit discharge characteristic dimension ‘F’ (this will be approximately equal to the fan diameter on an induced-draft tower) is shown in Figure 3.
The plume is dispersed from an area equal to the discharge characteristic dimension squared (F2). As the plume spreads, the concentration of the cooling tower exhaust is diluted by the ratio of the cross-sectional areas. When the plume touches the ground, the plume expansion changes from a square to a rectangle since the ground blocks the plume from further downward expansions. The model is particularly inaccurate in the ‘near field,’ before the plume touches the ground. This area is essentially being ignored because the drift droplets as they leave the tower are too large to be respirable.
Time and distance are required for the droplets to be reduced to 5 microns or less in size. For modeling Legionella dispersion in the far-field, we assume, conservatively, that all Legionella bacteria are in respirable-sized droplets. Due to the momentum of exhaust air as it leaves the tower, the plume will tend to rise up before bending down. To account for this rise, the height of the plume when it bends is set at twice the characteristic discharge dimension. For typical factory-assembled, HVAC cooling towers set on the ground this is a reasonable assumption.
Figure 3 details the near-field plume dispersion of the simplified model. With this model, the plume touches the ground at a distance D0 = 2F/ tan (6 o) = 20 F. At this point the cross-sectional area of the plume will be (5F)^2, 25 times the area as the plume left the tower. We are calling this D0 the characteristic distance and using multiples of this distance as a simplified way to describe plume dilution independent of cooling tower size.
We have arbitrarily set the point where the plume touches the ground, D0 , as the end of the near-field. As the plume continues to expand beyond this point, it is constrained from expansion in the vertical direction by the ground. We assume no such constraint in the horizontal direction.
Figure 4 illustrates how these assumptions affect the plume dilution. At 2x D0 the cross-sectional area is 63 F^2, at 3x D0 the area is 117 F^2 , and at 4x D0 the area is 187 F^2 .
The ratio of the cross-sectional area of the plume at a specific distance to the cross-sectional area of the plume as it leaves the tower (F^2) is a measure of the dilution of the cooling tower exhaust by ambient air at that point.
The characteristic distance, D0, that the plume travels before it touches the ground is a function of the discharge dimension. Larger units exhaust higher off the ground and the near-field of the plume extends for a greater distance. Table 2 lists this distance for some common sized units used in factory-assembled towers as well as some multiples of this distance.
COMPARISON OF SIMPLIFIED MODEL WITH EPA SCREENING MODEL – The EPA has several plume modeling programs for estimating air quality impact of stationary sources. One of these programs, SCREEN3^8, was used to validate the simple model described in this paper.
Data was input for ground level pollutant dilution from two cooling towers, one with a 12-foot diameter fan and operating at ½ fan speed; the other with an 8-foot diameter fan operating at full speed. The program automatically chooses the wind and atmospheric conditions that produce the highest concentration at a given distance.
From these worst-case concentration values, the plume dilution was calculated. The values of the SCREEN3 worst case dilutions are plotted against the simplified model dilutions in Figure 5. There is very good agreement between the two models.
Both these models assume no other structures in the immediate area. The close agreement of the simplified model with the EPA model helps to validate the reasonableness of this simplified approach.
The simplified model allows a quick consideration, independent of tower size, of the plume dilution at a distance from the tower. The model fails at very short distances (less than 20 fan diameters the
simplified model assumes zero ground-level concentration) and at very long distances. For the intermediate distance in an open area it has some use.
BACTERIA CONCENTRATION IN PLUME
We can now combine the bacteria concentration in the drift from Table 1 with the plume dilution after it touches the ground from Figure 4 to determine how many hours someone must be at several
distances from the tower in order to, on average, inhale a single Legionella bacterium.
These calculations assume:
1. A mechanically well-maintained tower (drift eliminators, fill, nozzles, etc.)
2. A moderate Legionella contamination of the circulating water of 100 CFU/ml
3. Ground-level tower (worst-case)
4. The worst-case l/g (at full speed) for typical modern cross-flow towers and for typical modern counter-flow towers (in terms of drift concentration in the cooling tower exhaust).
5. Fans operating at ½ speed (worst-case doubling the drift concentration from full fan speed).
6. All of the Legionella bacteria that leave the tower become respirable (worst-case).
Table 3 indicates that in a modern tower that was acceptably maintained there is little chance of inhaling multiple Legionella bacteria unless one spent extensive time close to the tower under a worst scenario wind condition or if the cooling tower were sited very near a building fresh air intake.
LEGIONELLA IN COOLING TOWER WATER
THE ECOLOGY OF LEGIONELLA – Legionella have a unique ecology compared with other bacteria that live in water. It is now well understood that Legionella in the environment grow as intracellular parasites of free-living amoebae and other protozoa. Rowbotham9 first demonstrated the ability of Legionella to replicate within freshwater and soil amoebae as early as 1980, and since then this phenomenon has been confirmed by many investigators using Acanthamoeba, Naegleria, and Hartmanella amoebae, and the ciliated protozoan Tetrahymena 10. Most authorities agree that this intracellular replication not only plays a vital role in the amplification of Legionella in the environment, but is also the unique pathogenic ability that enables Legionella to infect humans via the intracellular replication with monocytes and macrophages.
In an environmental habitat such as a cooling tower, most of the amoebae reside as part of the biofilm on the solid surfaces, rather than free in the water. This complex ecosystem contains a wide variety of slime-producing bacteria that colonize the surfaces, along with higher organisms such as amoebae and other protozoa that graze on the bacteria as a food source.
Legionella interact with the amoebae in the biofilm, blocking the killing and digestion process of the amoebae, and replicating to large numbers within the food vacuole or vesicle inside of the amoebae. Eventually the amoeba is killed and the Legionella are released to find new hosts. Some of the bacteria (including Legionella) and amoebae in the biofilm migrate from the surface into the free-flowing water and are distributed to other biofilm locations.
It is these water-borne (referred to as planktonic) Legionella, along with other bacteria and amoebae, that are released into the air from the cooling tower in the drift. Rowbotham8,11 has described the replication cycle of Legionella within the amoebae and first noticed the release of small vesicles full of Legionella at certain stages. He hypothesized that, while intact amoebae (10-40 ìm diameter) are generally too large to be inhaled into the lungs, the inhalation of small (<5 ìm) vesicles packed with Legionella could provide an infectious dose in a single inhalable particle 10.
More recently, Berk and colleagues 12 have more convincingly demonstrated the production of respirable vesicles (2-6 ìm in diameter) containing live Legionella from Acanthamoeba. Berk 11 estimated that each vesicle could contain between 20 and 200 bacteria, while Rowbotham10 calculated numbers in the range of 365-1,483 Legionella for a 5 ìm diameter vesicle. These membrane bound vesicles would also protect the Legionella from desiccation during the airborne dissemination from the tower.
Thus, from a disease transmission perspective, the cooling tower drift that exits the tower would contain a mixture of free Legionella, clusters of Legionella within respirable amoebae vesicles, and intact amoebae containing Legionella. In would seem obvious that the respirable vesicles would provide the highest risk to humans since they can be inhaled into the lungs, with a single vesicle providing a potentially infectious dose of Legionella.
THE STANDARD CULTURE METHOD FOR LEGIONELLA – The gold standard for detection and quantitation of Legionella in cooling towers or other environmental water samples is the standard culture method originally described by the CDC13. In this procedure, samples (with and without acid treatment to reduce the other heterotrophic bacteria in the water) are diluted and portions plated on selective and non-selective agar media. Any Legionella-like colonies that appear after appropriate incubation are confirmed as Legionella (species and serotype) with standard confirmation procedures. Using the assumption that each colony originated from a single bacterium, the number of Legionella in the water can be calculated and recorded as “colony forming units” (CFU) per ml or liter of original sample.
To date, a number of organizations have published protocols for the culture of Legionella from environmental samples including an international standard (ISO 11731) 14. Culture of Legionella from environmental samples is technically demanding and successful testing requires a microbiology laboratory that is experienced in the detection of this bacterium. There are no programs to certify the proficiency of environmental laboratories for their ability to culture Legionella.
In addition, variations which can be associated with procedures such as filter concentration or acid pretreatment (to kill non-Legionella bacteria) can dramatically affect the number of Legionella detected by these procedures. CDC will be initiating a proficiency testing program for environmental laboratories culturing Legionella in 2009 which should help with the standardization of these practices. Until that time, the only way to ensure accurate testing results is to rely on highly experienced laboratories.
It should be noted that other non-culture techniques are available for detection and quantitation of Legionella in environmental samples, including antigen-antibody based methods (such as immunofluorescence microscopy) and nucleic acid detection procedures such as polymerase chain reaction (PCR). The major limitations of these other procedures is that they may cross-react with other bacteria in the water and do not distinguish between living (infectious) and dead (non-infectious) Legionella in the sample.
LEGIONELLA CONCENTRATION IN TOWER WATER – Using the standard culture technique, many investigators have shown that Legionella is a common part of the microbial ecosystem in cooling tower water, although usually at low concentrations.
Results of a large survey of cooling towers (2,590 samples) over several years published by Miller and Koebel in 2006 15 showed that 12% of the tower samples had detectable Legionella above the limit of sensitivity of 10 CFU/ml and 2% of the samples had levels above 1,000 CFU/ml.
A similar Spanish study presented at the European Congress of Clinical Microbiology and Infectious Diseases in 2004 by Garcia-Nunez 16 found that 18 % of 554 cooling towers randomly sampled over a three-year period were culturepositive for Legionella at their increased limit of sensitivity of 10 CFU/liter.
Legionella numbers generally constitute a small percentage of the total bacterial population in tower water (usually < 1% of the total heterotrophic bacteria). However, Miller and Kenepp 17 showed that (perhaps as a result of biocide selectivity), the Legionella numbers may occasionally approach or achieve 100% of the bacterial population in the cooling tower water, often at levels exceeding 1,000 CFU/ml.
Cooling towers responsible for outbreaks of Legionnaires’ disease often have high concentrations of Legionella in their water. To the best of out knowledge, the highest concentration reported in such an outbreak investigation was in a tower which contained 10 5 CFU/ml of L. pneumophila 16. Analysis of cooling tower water with non-culture methodology tends to give higher percentages of samples positive for Legionella.
This is due to 1) the detection of both living and dead Legionella in the samples, and 2) the increased sensitivity of PCR over the standard culture method (i.e. the ability to detect very low levels of Legionella in the sample).
UNDER ESTIMATIONS OF THE STANDARD METHOD – While the standard culture technique is generally very reliable and reproducible in a qualified laboratory, this method may significantly under-estimate the actual number of Legionella in a cooling tower water sample as a result of:
1. Interference by other bacteria. Because Legionella is usually a minority of the total bacterial population in the cooling tower water, it is essential that the acid treatment and selective media successfully eliminate or inhibit the other bacteria so that the Legionella can grow without interference. While occasionally encountered in all labs, interference is a problem most common in laboratories not familiar with these critical elements of the standard procedure.
2. Viable but non-culturable (VBNC) Legionella. Examination of cooling tower water by non-culture techniques such as immunofluorescent microscopy or polymerase chain reaction (PCR) often reveal the presence of Legionella in samples that were culture-negative, or higher numbers of Legionella in samples that were culture-positive. The demonstration of a VBNC state for Legionella has been convincingly demonstrated by several investigators using nutrient limitation 18,19,20 or disinfectant exposure 21,22. These VBNC Legionella can be resuscitated by exposure to amoebae and are potentially infectious for humans.
3. Clusters of Legionella bacteria. Any clusters or clumps of Legionella introduced onto an agar plate would form a single colony and be under-counted as a single Legionella bacterium. Clusters containing more than one Legionella are readily observable when water samples are examined using immunofluorescence microscopy (IFM).
If each cluster were counted as a single Legionella, the clusters may account for as much as 5-10 % of the Legionella observed by IFM (personal observation). Additional clusters of Legionella within intact amoebae or amoebae vesicles may not be observed by IFM due to an inaccessibility to the anti-Legionella antibody used in this method, but still form single colonies when cultured onto an agar plate.
Thus, the production of vesicles (each containing potentially large numbers of viable Legionella) by amoebae or other protozoa would be of great importance in terms of disease transmission.
EPIDEMIOLOGY OF OUTBREAKS
The preceding sections have discussed a semi-quantitative methodology for evaluating the risk for a Legionella infection from a cooling tower. The fundamental question is does the epidemiology data support the methodology.
The methodology assumes that the concentration of airborne bacteria is due primarily to four factors:
1. The concentration of Legionella bacteria in the circulating water.
2. The quantity of drift generated by the tower.
3. The quantity of air passing through the tower.
4. The dilution of the plume by ambient air.
The exposure dose and hence the risk of disease is then related to the time spent breathing in the contaminated air. Robert Breiman^5 describes a 1988 outbreak at a Los Angeles retirement home. The data from that study was used to validate the methodology used in Table 1 for calculating the concentration of Legionella bacteria in cooling tower exhaust. The cooling system consisted of a below ground, forced-draft evaporative condenser that exhausted at sidewalk level. There was an air intake to the building located near and slightly higher than the exhaust. Tests on the water in the condenser basin showed Legionella counts of 9,000 CFU/ml.
The cooling system in that study had many problems with its design and operation that are addressed with current guidelines.
1. The drift eliminators on modern counter-flow towers are 20 times better. This alone may have prevented the outbreak.
2. 9,000 CFL/ml is very high. Almost any good water treatment would lower this value by one to two orders of magnitude.
3. The location of the building inlet air was particularly bad. Cooling tower exhaust always has some buoyancy and will usually rise as it leaves the tower. Locating the building air-inlet near and above the exhaust is particularly bad.
4. Exhausting an underground condenser at sidewalk level is improper.
5. Towers sited on the ground have a tendency to draw in a broad array of organic and mineral contaminates – much more so than a tower sited on a roof. An underground tower that draws its inlet air downwards is even more likely to draw in contaminates. These contaminates make biological control of the circulating water more difficult.
6. Since this is a retirement home, it would warrant special attention An order of magnitude analysis of the condenser identified as causing the out break show that the drift rate was an order of magnitude
too high, the bacteria concentration in the recirculating water was two orders two high, and the location of the condenser exhaust and the building air intake resulted in a order of magnitude more contaminated air being drawn into the building. Had any one of the order-of-magnitude criteria conditions been properly controlled, the probability of an outbreak would have been greatly reduced. Clive Brown^23 describes a 1994 outbreak in the area around a Delaware hospital. The paper strongly suggested a relationship to time spent near the contaminated cooling tower and risk of infection.
The authors develop a variable which they call an Aerosol Exposure Unit (AEU) to describe the dose that an individual received. The AEU is proportional to the hours spent at a specific distance
from the tower with a formula of:
AEU = time (in hours) / distance (in miles).
The authors performed a detailed case-control study of 22 people who came down with the disease and matched controls of similar age and health that attend the same clinic but were disease-free. This study showed a very strong correlation with high AEU and disease.
The AEU formula assumes that:
1. The dose is linear with time spent breathing contaminated air. This is exactly the assumption that we are making.
2. The contamination falls off in a linear manner the farther the distance form the tower. The AEU formula implies that contamination-concentration is proportional to 1/distance. Since contamination-concentration is proportional to 1/ plume-dilution, the AEU formula assumes a linear increase between plume-dilution and distance. This is different from the model used in this paper.
Our model assumes a quadratic increase in plume dilution with distance. Although different, the result on this relatively small data base may not be noticeable.
Figure 6 is a plot of the plume dilution of a ground-based tower with a 12’ fan using the model proposed in this paper. Also on this plot is a linear regression of those points that is forced through the origin. This linear regression represents the relationship used to develop the AEU variable. The difference in the two plots is not likely to be significant in the disease-AEU correlation.
DISCUSSION
This description of Legionella concentration in a dispersed plume is only intended as an order-of-magnitude approximation. While there is no known safe levels of Legionella exposure, values that are less than 1 Legionella bacterium inhaled in 24 hours seem very low. This indicates that if current guidelines are followed – good drift eliminators, good mechanical repair of the tower, sound tower sitting practice, reasonable microbiological control, etc. – the risk of Legionella infection can be quite low.
The basic plume dispersion model is a very simplified case. Buildings in the area of the tower as well as other structures will dramatically affect how the plume is diluted. The simplified model is only
intended as a visualization and order-of-magnitude approximation that bacteria concentration, in general, decreases as distance from the tower increases. There are well-documented cases of Legionellosis associated with cooling tower exposure – how does this analysis help understand those cases? In general, two or more of the basic assumptions of this analysis were not followed. It is useful to review the basic assumptions of this analysis because they certainly do not apply in many situations. The assumptions of this analysis are:
1. That modern drift eliminators (0.005% for cross-flow and 0.001% for counter-flow are being used. The drift eliminators that were the standard just a few years ago were much less effective, having drift rates up to an order of magnitude higher.
2. That the tower is in good mechanical shape. Damaged or missing drift eliminators will greatly increase the level of drift. While much of the drift from such a tower will be in very large, rain-drop sized drops that fall quickly to the ground, some will be en-trained in the air and contribute to
the respirable Legionella loading of the plume.
3. The tower is assumed sited on a clear area of ground with people below the level of the exhaust. A subterranean installation with a ground-level exhaust would be more dangerous as would locating the cooling tower such that the exhaust can be drawn into a building’s air inlet.
4. Reasonable water treatment is in effect. Although not specifically aimed for Legionella eradication as is required by some non-US regulations, water treatment that keeps good general biological control is assumed.
Another underlining assumption is that infection is caused by the inhalation of individual Legionella bacteria. As previously described in this paper, there may be vesicles in cooling towers which may contain tens if not hundreds of Legionella bacteria. The cooling towers in the outbreaks discussed in this paper all had old-style drift eliminators, high levels of Legionella in the basin, and were improperly sited. The contamination level was sufficiently high to allow an individual to inhale tens if not hundreds of Legionella bacteria by spending a relatively short time in the tower plume.
These outbreaks can be explained by inhalation of sufficient individual bacteria to cause a disease. However, these outbreaks could also be explained by the inhalation of individual vesicles emitted in the drift. There are several studies 24,25,26 that implicate a cooling tower with infection that occurred at a considerable distance from the cooling tower. A vesicle-vector would help explain how an infectious dose could be inhaled at a significant distance form the tower.
If vesicles are the disease vector, and if a single vesicle contained sufficient Legionella bacteria to provide an infectious dose, then the analysis would change. A very simplified example illustrates the difference. If we had an aerosol contamination such that there was 1 chance in 10 that a person would inhale 1 bacteria in a minute and 1000 people spent 1 minute breathing the air then 100 people would breath in 1 bacteria. A single bacteria is probably too low a dose to cause disease so nobody gets infected.
If we had an aerosol contamination such that there was 1 chance in 100 that a person would inhale 1 vesicle in a minute and 1000 people spent 1 minute breathing the air then 10 people would inhale a
vesicle. If every vesicle contained a large quantity of bacteria then some of the 10 could become infected, depending on the health of the individual.
Current epidemiological data has not been able to distinguish between the alternative vectors and it is possible that both vectors are involved in infections. Reduction or elimination of vesicles in
tower water could require a different water treatment approach than reduction of Legionella bacteria.
CONCLUSION
There are many excellent guidelines for minimizing the risk of Legionella infection from cooling towers. These guidelines all recommend proper maintenance, good drift eliminators, proper siting, and good biological control among many other recommendations; however, none of these guidelines attempts to describe the relative importance of these recommendations. Because of the quantitative nature of biological control, that aspect is often emphasized over other aspects of control. A 1990- genre cooling tower with drift eliminators that reduce drift to 0.02% and a Legionella count of 50 CFU/ml might be thought of being a low risk of infection while a 2000-genre cooling tower with drift
eliminators that reduce drift to 0.001% and a Legionella count of 1000 CFU/ml would require immediate disinfection.
Using the approach in this paper, the 2000-genre cooling tower presents a similar risk for a community-acquired infection than the less contaminated, older tower. One way to reduce the risk of Legionella exposure from older equipment is to upgrade the drift eliminators to the modern, high-efficiency designs.
Further studies of the transmission vector for cooling-tower infection (bacteria, vesicles, or both) are needed. The control of amoebae and vesicles in the cooling water may require different water treatment than the control of planktonic Legionella bacteria.
The factor that vesicles may play in disease transmission needs to be better understood. It has been well known that the level of Legionella bacteria in the cooling tower plays some role in the risk of infection; however, the mere presence of the bacteria is insufficient to predict the potential for disease transmission. The semi-quantitative approach to Legionella aerosol exposure level, as outlined in this paper, can be very beneficial in the management of several of the differing factors that contribute to risk of Legionnaires’ disease transmission.
The importance of drift eliminator design, physical maintenance practices, and siting as well as bacteria counts can all be roughly weighed as to their contribution to the overall risk-of-infection. The authors
feel that the semi-quantitative approach taken in this paper can be beneficial in evaluating the risks associated with a particular cooling tower, evaluating the benefits of proposed changes in cooling tower guidelines, evaluating the importance of equipment design modifications, and aiding in the investigation of outbreaks.
REFERENCES
CTI Journal, Vol. 31, No. 1
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
