- Email: [email protected]
Establishment of HPC(R2A) for regrowth control in non-chlorinated distribution systems
International Journal of Food Microbiology 92 (2004) 317 – 325 www.elsevier.com/locate/ijfoodmicro
Establishment of HPC(R2A) for regrowth control in non-chlorinated distribution systems Wolfgang Uhl a,*, Gabriela Schaule b b
a Water Technology Department, Gerhard-Mercator-University Duisburg, Bismarckstr. 90, D-47048 Duisburg, Germany Microbiology Department, IWW Rheinisch-Westfa¨lisches Institut fu¨r Wasser Beratungs-und Entwicklungsgesellschaft mbH, Moritzstr. 26, D-45476 Mu¨lheim an der Ruhr, Germany
Abstract Drinking water distributed without disinfection and without regrowth problems for many years may show bacterial regrowth when the residence time and/or temperature in the distribution system increases or when substrate and/or bacterial concentration in the treated water increases. An example of a regrowth event in a major German city is discussed. Regrowth of HPC bacteria occurred unexpectedly at the end of a very hot summer. No pathogenic or potentially pathogenic bacteria were identified. Increased residence times in the distribution system and temperatures up to 25 jC were identified as most probable causes and the regrowth event was successfully overcome by changing flow regimes and decreasing residence times. Standard plate counts of HPC bacteria using the spread plate technique on nutrient rich agar according to German Drinking Water Regulations (GDWR) had proven to be a very good indicator of hygienically safe drinking water and to demonstrate the effectiveness of water treatment. However, the method proved insensitive for early regrowth detection. Regrowth experiments in the lab and sampling of the distribution system during two summers showed that spread plate counts on nutrient-poor R2A agar after 7-day incubation yielded 100 to 200 times higher counts. Counts on R2A after 3-day incubation were three times less than after 7 days. As the precision of plate count methods is very poor for counts less than 10 cfu/plate, a method yielding higher counts is better suited to detect upcoming regrowth than a method yielding low counts. It is shown that for the identification of regrowth events HPC(R2A) gives a further margin of about 2 weeks for reaction before HPC(GDWR). D 2003 Elsevier B.V. All rights reserved. Keywords: Drinking water; Distribution; Regrowth; Residence time; Temperature; Plate counts; HPC bacteria; R2A; AOC; Internal alert limit
1. Introduction According to the German Drinking Water Regulation (GDWR; Anonymous (2001)), which is the national version of the European directive (Anonymous (1998), final disinfection of drinking water is required only if the water does not meet the hygienic * Corresponding author. Fax: +49-89-2443-36969. E-mail address: [email protected] (W. Uhl). 0168-1605/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.ijfoodmicro.2003.08.010
standards, i.e., if the water is not free from pathogens (no E. coli, coliforms or enterococci may be detected in 100 ml) and/or if pour plate counts for heterotrophic bacteria on a defined nutrient-rich agar (to be determined at 20 and 36 jC incubation for 48 h) exceed the guidance value of 100 cfu/ml. In many cases, in Germany, drinking water is distributed without disinfection as the water already meets the hygienic standards and careful management of the distribution system by the utilities and pur-
318
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
veyors is guaranteed. In fact, a survey carried out in 1991 revealed that more than 50% of responding utilities do not disinfect (Haberer (1994). Almost all of these (>99%) use protected ground- and source water as raw water. Approximately half of them meet the physicochemical and microbiological standards without any further treatment; in many of the others only simple treatment like aeration and iron/manganese removal is necessary. In general, customers in Germany, as well as in many other European countries, prefer drinking water without chlorine residual for reasons of taste and disinfection by-products. Therefore, in several cases suppliers check whether they meet the requirements to distribute their drinking water without disinfectant residual; or, if not, which measures need to be taken to fulfill the requirements. An example is described by Hambsch (1999). On the other hand, there are many utilities who have been distributing water without disinfectant residual and who have met the bacteriological GDWR standards for many decades without any disinfection. As this was working long before the bacterial regrowth debate started, there are cases where the water does not necessarily meet the requirements with respect to regrowth. However, regrowth in distribution systems may occur when the conditions in the system are subject to changes, such as when flow rates decrease and water temperatures increase as a consequence of declining consumption. It is often not recommended in these systems to establish chlorination as it may result in destabilization of water main encrustations and formation of disinfection by-products. Preference is given to monitor microbiological water quality and to take early measures to prevent regrowth.
Table 1 Criteria for biologically stable drinking water (regrowth parameters) Method
Parameter, criteria
Source
AOC BDOC
AOC < 10 Ag/l BDOC < 0.15 mg/l
BRP
l < 0.1/hF < 5
van der Kooij (1990) Laurent et al. (1993), Servais et al. (1995) Hambsch (1994)
methods have been established (with numerous modifications) to measure bacterial regrowth potential. The AOC-method (assimilable organic carbon) developed by van der Kooij et al. (1982) measures carbon, expressed in acetate-C-equivalents, which is built into biomass. The biodegradable organic carbon (BDOC) methods originally developed by Servais et al. (1987) and Joret et al. (1988) measure organic carbon mineralized by bacteria. Finally, the BRP method (bacterial regrowth potential), developed by Werner (1985) and Hambsch et al. (1992), determines the growth rate l and multiplication factor F of indigenous heterotrophic bacteria. For all three methods the generally accepted criteria listed in Table 1 have been derived, according to which drinking water may be distributed without excessive risk for bacterial regrowth. It has to be kept in mind that these criteria are based on laboratory as well as field investigations. Especially for the field investigations, no or negligible regrowth was observed when the parameter values were below those given in Table 1, which mainly are parameters for limiting nutrients. However, there are also other factors limiting regrowth. The most important are water temperature, residence time and initial bacterial concentration. Bacterial growth rates are extremely temperature-dependent.
3. Factors affecting regrowth 2. Criteria for distribution without disinfectant residual At first, it is self-evident that water to be distributed without disinfectant must meet the microbiological standards of the respective drinking water regulation. In general, since the 1980s special attention is taken to conditions under which bacterial regrowth in the distribution system may occur. Overviews on methods for analysis of regrowth potential have been given by Huck (1990) and Uhl (2000). In brief, three
In Fig. 1 the results of a laboratory regrowth experiment are shown. The 1-l Erlenmyer flasks were made carbon-free according to the AOC procedures (see, e.g. Anonymous (1995a)). The water sample investigated for regrowth is a treated drinking water as it is distributed for several decades without chlorination. It is produced from a well-protected groundwater, partly enriched with surface water from a nature sanctuary. In the treatment plant, the raw water is aerated in open cascades and iron and manganese
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
319
Fig. 1. Effect of temperature on bacterial regrowth: Bacterial proliferation (HPC(R2A)) in drinking water samples incubated at 10 and 20 jC, respectively. HPC concentrations measured (closed symbols) and dynamic simulation (lines with open symbols).
are removed in deep bed filters. The finished drinking water is stored in a closed reservoir before distribution without chlorination. The microbiological quality of the treated drinking water is very good. Bacterial concentrations according to the GDWR are below detection. According to the HPC-method on R2A agar and after 7 days of incubation (Anonymous (1995b) (in the following abbreviated as HPC(R2A)), bacterial concentrations in the raw water are about 50 cfu/ml. AOC concentrations in the treated water are around 30 Ag/l. Incubation was started 1 day after sampling (until then the sample was kept at approximately 5 jC). Concentrations of heterotrophic bacteria were measured as HPC(R2A). Results of a model simulation (first-order growth kinetics, substrate limited growth, yield of 3 103 cfu/Ag AOC, temperature dependence according to Arrhenius, activation energy 63 kJ/mol) based on initial HPC concentration and AOC concentration of the water are also shown. It can be depicted clearly from Fig. 1 that it takes the bacteria, at 20 jC, only half the time to reach their maximum between 104 and 105 cfu/ml when compared to 10 jC. Or, in other words, after a reasonable residence time of 2 days the HPC concentration is more than 10 times as high at 20 jC when compared to 10 jC.
At low temperatures and short residence times even drinking water with (e.g.) AOC concentrations of more than the 10 Ag/l given by van der Kooij (1990) may be distributed without problems. However, bacterial regrowth may occur when the situation in the distribution network (temperature, residence time) changes, when changes in raw water quality are taking place or when treatment is altered. The water’s initial bacterial concentration is also of very high importance, as bacterial growth is exponential. If, e.g. two waters both meet the standards for microbiological parameters in that they both are below the guidance value, but different in heterotrophic bacteria concentration, they will show quite different regrowth behaviour even if AOC and temperature are the same. The very important influence of initial bacteria concentration and limiting substrate concentration, besides temperature, is demonstrated in Fig. 2. For the regrowth experiments and the simulations at 10 and 20 jC shown in Fig. 1, two more simulations were carried out. When the limiting substrate concentration was doubled (i.e. AOC = 60 Ag/l instead of 30 Ag/l) not only a two times higher final bacteria concentration was reached but the bacterial proliferation would also be much faster. Also, when the initial bacteria concentration was higher by a
320
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
Fig. 2. Simulation of regrowth to show, besides temperature, the importance of initial bacteria concentration and nutrient concentration on growth rate and maximum regrowth.
factor of 10 at the same substrate concentration, regrowth would take place much faster. The final concentration would be reached already after 2 days compared to approximately 4 days at lower initial concentrations. Another possible source of bacterial regrowth is the release of biodegradable substances from pipe materials into the water phase. Thus, it has to be insured that no such materials are in use. Furthermore, the unexpected intrusion of microorganisms and/or biodegradable substances into the network must be prevented. For this, pressure in the distribution system has to be high enough and technical measures have to be taken to prevent backflow from taps. Very important is also the proper training of all personnel working on the distribution system and treatment train.
4. A regrowth incident as consequence of increased residence time As already pointed out in the introduction, there may be situations where the risk for bacterial regrowth in the distribution system increases. Very often the cause is a decrease in water consumption. In some areas in Germany, consumption decreased more than 50% during the last decade as a result of
simultaneously changed consumer habits, emigration from the area and discontinuance of industrial activity. As a consequence, residence times in the distribution system increased more than twofold. During long and hot summers, this can result in water temperatures in the distribution system of more than 20 jC. As an example, consider the case of a German city in which drinking water consumption in the city had decreased about 70% over a period of 10 years since the German reunion. The drinking water distributed to the city is produced from a well-protected ground water source by aeration and iron and manganese removal. In the finished water, HPC-bacteria, determined according to the German Drinking Water Regulation (HPC(GDWR)), are usually not detectable or are at least below 5 CFU/ml. The water is distributed without disinfection. At the end of a very hot summer, a regrowth incident occurred. HPC(GDWR) approached and sometimes exceeded the guidance value of 100 cfu/ ml. At the same time no fecal indicators or other hygienically relevant bacteria were detected. In Germany, when the guidance values are violated, this has to be reported to the local health authorities immediately. They have to decide, usually in consultation with the utilities and purveyors, which measures are to be taken immediately. It is recom-
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
mended and advantageous that a thorough plan of action be developed in case of regrowth, including expected outcomes. In the case described, discussions between the health authorities, the water works, the purveyor and a consultant were begun immediately. After detailed analysis and identification of points of pronounced regrowth and after consideration of possible effects of chlorination on the distribution system (destabilization of incrustations), a unanimous decision was made to fight the regrowth problem by changing the flow regime, thus decreasing residence time. Management of flow through tanks was altered to minimize residence time. Hydrants were flushed. Furthermore, water that formerly had been pumped around the city and distributed to the backcountry was now led through the city in order to maximize flow rates in the city network. This resulted in only an insignificant increase in total residence time of the water distributed to the backcountry. These measures proved to be very successful. Colony counts started to decrease after a few days and were back to normal levels (below 10 cfu/ml GDWR) in 3 weeks.
5. Establishment of HPC(R2A) for early regrowth detection After the regrowth incident had been coped with successfully, it was discussed how such events could be prevented in the future. AOC in the finished water leaving the water works was analyzed several times a year and was found between 20 and 30 Ag/l. The measures taken to shorten water residence time in the distribution system were continued and even improved. However, as the regrowth incident had occurred very suddenly, we looked for a method to detect upcoming regrowth. The HPC-method according to the GDWR is carried out using pour plates with a nutrient rich agar (beef extract, peptone at 1% each) incubated for 2 days at 20 jC and 36 jC. The method, together with the guidance values of 100 cfu/ml, has proven to be very successful in ensuring hygienically safe drinking water. However, based upon the authors experiences, it has proven to be too insensitive for early detection of upcoming regrowth in distribution systems with
321
low bacterial concentrations. Therefore, other culture methods were investigated. Reasoner and Geldreich (1985) developed a plate count method to enumerate culturable bacteria from nutrient-poor environments. The derived HPC-method Anonymous (1995b) using a spread plate technique on nutrient-poor R2A agar and with 7-day incubation (HPC(R2A)) has generally proven to be much more sensitive than pour plate methods using nutrient-rich agar. Therefore, it was decided to compare this method and the GDWR plate count method for their suitability for early regrowth detection. During two summers, HPC(R2A) was applied routinely besides the HPC(GDWR) on samples from the non-chlorinated distribution system. The plate counts obtained with the two methods were compared statistically. Fig. 3 shows the statistical comparison of plate counts obtained from samples collected from the water works outlet and at points A and B on two different distribution paths. The regrowth-factor is calculated as the bacterial concentration at points in the distribution system divided by the bacteria concentration at the point of entry. The analysis shows clearly that a multiplication of HPC(R2A) takes place from the system entry to point A by a factor of 6 and by a factor of 12 to point B. However, with HPC(GDWR) no multiplication can be detected. This is due to the fact that the bacteria concentrations are very low and the HPC(GDWR) method is too insensitive to detect (at these low concentrations) statistically significant plate count differences between samples from different points. For the samples analyzed (from the tree points entry, A and B) the more sensitive HPC(R2A) method yielded counts up to 400 cfu/ml, whereas the HPC(GDWR) method always yielded less than 7 cfu/ml. As for all plate count methods, the standard error for counts below 10 cfu/plate is considerably high (standard error F 3 cfu/plate according to the authors’ experience), differences in plate counts for low numbers are just the result of scatter and cannot be interpreted. Therefore, with HPC(R2A) an increase in colony counts can be observed when no counts are detectable with the HPC(GDWR)-method. In an upcoming regrowth event, once the detection limit of plate counts with HPC(GDWR) is exceeded, bacteria are already in the exponential phase and grow so fast that the
322
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
Fig. 3. Statistical analysis of bacterial counts (cfu/ml) according to the German drinking water regulation method and the HPC method on paths A and B in the distribution system (regrowth factors shown).
guidance value of 100 cfu/ml could be violated in a few days. However, with the HPC method regrowth events could be detected early and preventive measures taken.
The German Drinking Water Regulation since 2001 allows the application of other plate count methods than HPC(GDWR). However, in order to be able to value the results, determinations with the new
Fig. 4. Results of regrowth experiment for comparison of HPC(R2A) (left ordinate) and HPC(GDWR) (right ordinate).
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
method have to be carried out in parallel with HPC(GDWR) over a period of 1 year. At the time we were looking for a sensitive method for regrowth control, we also wanted to establish an alarm value for HPC(R2A) at which increased awareness for possible bacterial regrowth should be given. In order to establish such an alarm value, several regrowth experiments were carried out in the lab and HPC(R2A) and HPC(GDWR) determined in parallel. Fig. 4 shows the results of such an experiment. From this graph, it can be shown that regrowth occurs early with HPC(R2A) but not with HPC(GDWR). Colony counts using the HPC(R2A) techniques are higher than HPC (GDWR) by factors of between 100 and 200. From these results the purveyor set an internal alert limit of 1.000 cfu/ml HPC(R2A). Once this value was exceeded colony counts according to HPC(GDWR) higher than 10 cfu/ml were expected and special attention was taken to keep residence time in the distribution system low. A limitation for the use of HPC(R2A) for early regrowth control may be seen in the long incubation time of 7 days. A week would have been gone until the laboratory results were available. Therefore, over 1 year and at six different sampling points in the distribution system HPC(R2A), counts were also determined after 3 days of incubation. After outlierelimination (rm-test (Grubbs (1969), a = 10%) the results were compared by linear regression as shown in Fig. 5.
323
This showed that HPC(R2A) after 7-day incubation was higher by a factor of approximately 3 than HPC(R2A) after 3-day incubation. Thus, the purveyor’s internal alarm value for HPC(R2A) after 3day incubation was set to 300 cfu/ml.
6. Methods ability to identify regrowth—a model simulation Up to this point, the investigations and data analysis showed method sensitivities and interrelations. It became evident that HPC(R2A) at 7-day as well as at 3-day incubation are better suited to detect regrowth in the system early. However, the question remained how early or at what time ongoing regrowth would be detectable. To check for this a model simulation was carried out. For the model calculations, bacterial growth on AOC with an initial concentration of 30-Ag acetateC-equivalents/l flowing through a pipe was assumed. For simplification, bacterial growth was assumed to take place in the liquid phase only. With reference to the practical situation, it was assumed that the waters residence time in the pipe was 2 days and that it increased linearly over a period of 100 days to a residence time of 5 days. Also during the same period water temperature at the end of the pipe was assumed to increase from 10 to 25 jC linearly with time. The temperature gradient was modeled
Fig. 5. Comparison of HPC(R2A) data after 3 and 7 days of incubation at six different sampling points in the distribution system.
324
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
linearly over the length of the pipe. The plot was made up for sampling intervals of 7 days. For method precision, F 3 counts/plate was used as standard error for counts below 10 cfu/plate (authors experience) and F 25% for counts higher than 10 cfu/plate (authors’ experience, n = 6, and Mikell et al. (1996)). The results of the simulation are shown in Fig. 6. For interpretation it should be noted that the ordinates on the left and right are different. HPC(R2A) (7-day incubation) are plotted to the left, HPC(GDWR) to the right. The confidence intervals shown are for 67% confidence, i.e., if a large number of measurements were made, 67% of the results would be found between the upper and lower limits of the confidence interval. For data points without a confidence interval shown, it is as broad as or narrower than the symbol size. In order to identify growth, at least three following data points (two intervals) need to show a significant increase. In order to distinguish two measured data points from each other, it is best that the confidence intervals do not overlap. Analysing Fig. 6 on this basis yields that HPC(GDWR) growth will not be detectable until day 35, as this is the first time when two increases may have been identified; whereas for HPC(R2A), regrowth will be identified with the results of the sample from day 14. Taking into account the incubation times of 2 days for HPC(GDWR) and 7 days for HPC(R2A), it turns out that the bacterial regrowth in the pipe will be identified at day 21 with HPC(R2A)
and at day 37 with HPC(GDWR). Thus, the regrowth event will be evident about 16 days earlier with HPC(R2A) than with HPC(GDWR). This gives 16 days more time to react and take further measures against regrowth.
7. Conclusions Pour plate counts according to the German Drinking Water Regulations (HPC(GDWR)) are not sensitive enough to detect regrowth in distribution systems early and to enable the purveyor to react in time. Often no increase in colony-forming units can be observed over a long period of time and then suddenly the guidance value of 100 cfu/ml is exceeded. The HPC method with nutrient-poor R2A agar HPC(R2A) is much more sensitive. Colony counts obtained after 7 days were higher than HPC(GDWR) by factors between 100 and 200. HPC(R2A) counts after 3 days were lower by a factor of 3 than counts after 7 days. Model simulations showed that the application of HPC(R2A) may allow approximately 2-week earlier reaction on an upcoming regrowth event. For a German purveyor who was aware of possible regrowth during summer months, HPC(R2A) was established successfully for early regrowth detection. An internal alarm limit was set to 1.000 cfu/ml for HPC(R2A) 7-day incubation and special attention to keep residence times low was taken once this value was exceeded.
Fig. 6. Simulation of bacterial concentrations to be measured with HPC(R2A) and HPC(GDWR) at the end of the model pipe.
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
Acknowledgements The work of Stephany Hoff and Annika Eichelhardt at IWW who thoroughly carried out the AOC determinations and regrowth experiments as well as the contribution of our former colleague Dr. Markus Gerlach are greatly acknowledged. We furthermore would like to express our deep gratitude to our clients and partners who supplied data from full-scale distribution systems for this study.
References Anonymous, 1995a. Heterotrophic plate count. In: Am. Public Health Org./Am. Water Works Ass./Water Env. Fed. (Ed.), Standard Methods for the examination of water and wastewater. Chap. 9215. Washington, 9-32-9-39. Anonymous, 1995b. Assimilable organic carbon (proposed). In: Am. Public Health Org./Am. Water Works Ass./Water Env. Fed. (Ed.), Standard Methods for the examination of water and wastewater. Chap. 9217. Washington, 9-39-9-43. Anonymous, 1998. Richtlinie 98/83/EG des Rates vom 3. November 1998 u¨ber die Qualita¨t von Wasser fu¨r den menschlichen Gebrauch. Amtsblatt der Europa¨ischen Gemeinschaften DE-5.12.98, L330/32 – L330/54. Anonymous, 2001. Verordnung zur Novellierung der Trinkwasserverordnung vom 21. Mai 2001. Bundesgesetzblatt Teil 1 24, 959 – 980. Grubbs, F.E., 1969. Procedures for detecting outlying observations in samples. Technometrics 11 (1), 1 – 21. Haberer, K., 1994. Der Einsatz von Desinfektionsmitteln und das Auftreten von Desinfektionsprodukten in deutschen Wasserwerken. GWF. Wasser, Abwasser 135 (7), 409 – 417. Hambsch, B., 1994. Wiederverkeimung von Trinkwasser in Abha¨ngigkeit von Wasserqualita¨t, Kontaktfla¨chen und Verweilzeit. DVGW-Schriftenreihe. Wasser 79, 203 – 212.
325
Hambsch, B., 1999. Distributing groundwater without a disinfectant residual. Journal-American Water Works Association 91 (1), 81 – 85. Hambsch, B., Werner, P., Frimmel, F.H., 1992. Bakterienvermehrung in aufbereiteten Wa¨ssern verschiedener Herkunft. Acta Hydrochimica et Hydrobiologica 20 (1), 9 – 14. Huck, P.M., 1990. Measurement of biodegradable organic matter and bacterial growth potential in drinking water. Journal-American Water Works Association 82 (7), 78 – 86. Joret, J.C., Levi, Y., Dupin, T., Gilbert, M., 1988. Rapid method for estimating bioeliminable organic carbon in water. Proc. 1988 Am. Water Works Assoc. Annual Conf. pp. 1715 – 1725. Laurent, P., Servais, P., Randon, G., 1993. Bacterial development in distribution networks—study and modelling. Water Supply 11 (3/4), 387 – 398. Mikell, A.T., Smith, C.L., Richardson, J.C., 1996. Evaluation of media and techniques to enumerate heterotrophic microbes from karst and sand aquifer springs. Microbial Ecology 31, 115 – 124. Reasoner, D.J., Geldreich, E.E., 1985. A new medium for the enumeration and subculture of bacteria from potable water. Applied and Environmental Microbiology 49, 1 – 7. Servais, P., Billen, G., Hascoet, M.C., 1987. Determination of the biodegradable fraction of dissolved organic matter in waters. Water Research 21 (4), 445 – 450. Servais, P., Laurent, P., Randon, G., 1995. Comparison of the bacterial dynamics in various French distribution systems. Journal Water Supply: Research and Technology. AQUA 44 (1), 10 – 17. Uhl, W., 2000. Wiederverkeimung von Trinkwasser, Teil 1-Grundlagen und Methoden zur Bestimmung der Verkeimungsneigung. Bbr. Wasser, Kanal- und Rohrleitungsbau 51 (12), 30 – 35. van der Kooij, D., 1990. Assimilable organic carbon (AOC) in drinking water. In: McFeters, G.A. (Ed.), Drinking Water Microbiology. Progress and Recent Developments. Springer, New York, pp. 57 – 87. van der Kooij, D., Visser, A., Hijnen, W.A.M., 1982. Determining the concentration of easily assimilable organic carbon in drinking water. Journal-American Water Works Association 74, 540 – 545. Werner, P., 1985. Eine Methode zur Bestimmung der Verkeimungsneigung von Trinkwasser. Vom Wasser 65, 257 – 270.
Establishment of HPC(R2A) for regrowth control in non-chlorinated distribution systems Wolfgang Uhl a,*, Gabriela Schaule b b
a Water Technology Department, Gerhard-Mercator-University Duisburg, Bismarckstr. 90, D-47048 Duisburg, Germany Microbiology Department, IWW Rheinisch-Westfa¨lisches Institut fu¨r Wasser Beratungs-und Entwicklungsgesellschaft mbH, Moritzstr. 26, D-45476 Mu¨lheim an der Ruhr, Germany
Abstract Drinking water distributed without disinfection and without regrowth problems for many years may show bacterial regrowth when the residence time and/or temperature in the distribution system increases or when substrate and/or bacterial concentration in the treated water increases. An example of a regrowth event in a major German city is discussed. Regrowth of HPC bacteria occurred unexpectedly at the end of a very hot summer. No pathogenic or potentially pathogenic bacteria were identified. Increased residence times in the distribution system and temperatures up to 25 jC were identified as most probable causes and the regrowth event was successfully overcome by changing flow regimes and decreasing residence times. Standard plate counts of HPC bacteria using the spread plate technique on nutrient rich agar according to German Drinking Water Regulations (GDWR) had proven to be a very good indicator of hygienically safe drinking water and to demonstrate the effectiveness of water treatment. However, the method proved insensitive for early regrowth detection. Regrowth experiments in the lab and sampling of the distribution system during two summers showed that spread plate counts on nutrient-poor R2A agar after 7-day incubation yielded 100 to 200 times higher counts. Counts on R2A after 3-day incubation were three times less than after 7 days. As the precision of plate count methods is very poor for counts less than 10 cfu/plate, a method yielding higher counts is better suited to detect upcoming regrowth than a method yielding low counts. It is shown that for the identification of regrowth events HPC(R2A) gives a further margin of about 2 weeks for reaction before HPC(GDWR). D 2003 Elsevier B.V. All rights reserved. Keywords: Drinking water; Distribution; Regrowth; Residence time; Temperature; Plate counts; HPC bacteria; R2A; AOC; Internal alert limit
1. Introduction According to the German Drinking Water Regulation (GDWR; Anonymous (2001)), which is the national version of the European directive (Anonymous (1998), final disinfection of drinking water is required only if the water does not meet the hygienic * Corresponding author. Fax: +49-89-2443-36969. E-mail address: [email protected] (W. Uhl). 0168-1605/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.ijfoodmicro.2003.08.010
standards, i.e., if the water is not free from pathogens (no E. coli, coliforms or enterococci may be detected in 100 ml) and/or if pour plate counts for heterotrophic bacteria on a defined nutrient-rich agar (to be determined at 20 and 36 jC incubation for 48 h) exceed the guidance value of 100 cfu/ml. In many cases, in Germany, drinking water is distributed without disinfection as the water already meets the hygienic standards and careful management of the distribution system by the utilities and pur-
318
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
veyors is guaranteed. In fact, a survey carried out in 1991 revealed that more than 50% of responding utilities do not disinfect (Haberer (1994). Almost all of these (>99%) use protected ground- and source water as raw water. Approximately half of them meet the physicochemical and microbiological standards without any further treatment; in many of the others only simple treatment like aeration and iron/manganese removal is necessary. In general, customers in Germany, as well as in many other European countries, prefer drinking water without chlorine residual for reasons of taste and disinfection by-products. Therefore, in several cases suppliers check whether they meet the requirements to distribute their drinking water without disinfectant residual; or, if not, which measures need to be taken to fulfill the requirements. An example is described by Hambsch (1999). On the other hand, there are many utilities who have been distributing water without disinfectant residual and who have met the bacteriological GDWR standards for many decades without any disinfection. As this was working long before the bacterial regrowth debate started, there are cases where the water does not necessarily meet the requirements with respect to regrowth. However, regrowth in distribution systems may occur when the conditions in the system are subject to changes, such as when flow rates decrease and water temperatures increase as a consequence of declining consumption. It is often not recommended in these systems to establish chlorination as it may result in destabilization of water main encrustations and formation of disinfection by-products. Preference is given to monitor microbiological water quality and to take early measures to prevent regrowth.
Table 1 Criteria for biologically stable drinking water (regrowth parameters) Method
Parameter, criteria
Source
AOC BDOC
AOC < 10 Ag/l BDOC < 0.15 mg/l
BRP
l < 0.1/hF < 5
van der Kooij (1990) Laurent et al. (1993), Servais et al. (1995) Hambsch (1994)
methods have been established (with numerous modifications) to measure bacterial regrowth potential. The AOC-method (assimilable organic carbon) developed by van der Kooij et al. (1982) measures carbon, expressed in acetate-C-equivalents, which is built into biomass. The biodegradable organic carbon (BDOC) methods originally developed by Servais et al. (1987) and Joret et al. (1988) measure organic carbon mineralized by bacteria. Finally, the BRP method (bacterial regrowth potential), developed by Werner (1985) and Hambsch et al. (1992), determines the growth rate l and multiplication factor F of indigenous heterotrophic bacteria. For all three methods the generally accepted criteria listed in Table 1 have been derived, according to which drinking water may be distributed without excessive risk for bacterial regrowth. It has to be kept in mind that these criteria are based on laboratory as well as field investigations. Especially for the field investigations, no or negligible regrowth was observed when the parameter values were below those given in Table 1, which mainly are parameters for limiting nutrients. However, there are also other factors limiting regrowth. The most important are water temperature, residence time and initial bacterial concentration. Bacterial growth rates are extremely temperature-dependent.
3. Factors affecting regrowth 2. Criteria for distribution without disinfectant residual At first, it is self-evident that water to be distributed without disinfectant must meet the microbiological standards of the respective drinking water regulation. In general, since the 1980s special attention is taken to conditions under which bacterial regrowth in the distribution system may occur. Overviews on methods for analysis of regrowth potential have been given by Huck (1990) and Uhl (2000). In brief, three
In Fig. 1 the results of a laboratory regrowth experiment are shown. The 1-l Erlenmyer flasks were made carbon-free according to the AOC procedures (see, e.g. Anonymous (1995a)). The water sample investigated for regrowth is a treated drinking water as it is distributed for several decades without chlorination. It is produced from a well-protected groundwater, partly enriched with surface water from a nature sanctuary. In the treatment plant, the raw water is aerated in open cascades and iron and manganese
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
319
Fig. 1. Effect of temperature on bacterial regrowth: Bacterial proliferation (HPC(R2A)) in drinking water samples incubated at 10 and 20 jC, respectively. HPC concentrations measured (closed symbols) and dynamic simulation (lines with open symbols).
are removed in deep bed filters. The finished drinking water is stored in a closed reservoir before distribution without chlorination. The microbiological quality of the treated drinking water is very good. Bacterial concentrations according to the GDWR are below detection. According to the HPC-method on R2A agar and after 7 days of incubation (Anonymous (1995b) (in the following abbreviated as HPC(R2A)), bacterial concentrations in the raw water are about 50 cfu/ml. AOC concentrations in the treated water are around 30 Ag/l. Incubation was started 1 day after sampling (until then the sample was kept at approximately 5 jC). Concentrations of heterotrophic bacteria were measured as HPC(R2A). Results of a model simulation (first-order growth kinetics, substrate limited growth, yield of 3 103 cfu/Ag AOC, temperature dependence according to Arrhenius, activation energy 63 kJ/mol) based on initial HPC concentration and AOC concentration of the water are also shown. It can be depicted clearly from Fig. 1 that it takes the bacteria, at 20 jC, only half the time to reach their maximum between 104 and 105 cfu/ml when compared to 10 jC. Or, in other words, after a reasonable residence time of 2 days the HPC concentration is more than 10 times as high at 20 jC when compared to 10 jC.
At low temperatures and short residence times even drinking water with (e.g.) AOC concentrations of more than the 10 Ag/l given by van der Kooij (1990) may be distributed without problems. However, bacterial regrowth may occur when the situation in the distribution network (temperature, residence time) changes, when changes in raw water quality are taking place or when treatment is altered. The water’s initial bacterial concentration is also of very high importance, as bacterial growth is exponential. If, e.g. two waters both meet the standards for microbiological parameters in that they both are below the guidance value, but different in heterotrophic bacteria concentration, they will show quite different regrowth behaviour even if AOC and temperature are the same. The very important influence of initial bacteria concentration and limiting substrate concentration, besides temperature, is demonstrated in Fig. 2. For the regrowth experiments and the simulations at 10 and 20 jC shown in Fig. 1, two more simulations were carried out. When the limiting substrate concentration was doubled (i.e. AOC = 60 Ag/l instead of 30 Ag/l) not only a two times higher final bacteria concentration was reached but the bacterial proliferation would also be much faster. Also, when the initial bacteria concentration was higher by a
320
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
Fig. 2. Simulation of regrowth to show, besides temperature, the importance of initial bacteria concentration and nutrient concentration on growth rate and maximum regrowth.
factor of 10 at the same substrate concentration, regrowth would take place much faster. The final concentration would be reached already after 2 days compared to approximately 4 days at lower initial concentrations. Another possible source of bacterial regrowth is the release of biodegradable substances from pipe materials into the water phase. Thus, it has to be insured that no such materials are in use. Furthermore, the unexpected intrusion of microorganisms and/or biodegradable substances into the network must be prevented. For this, pressure in the distribution system has to be high enough and technical measures have to be taken to prevent backflow from taps. Very important is also the proper training of all personnel working on the distribution system and treatment train.
4. A regrowth incident as consequence of increased residence time As already pointed out in the introduction, there may be situations where the risk for bacterial regrowth in the distribution system increases. Very often the cause is a decrease in water consumption. In some areas in Germany, consumption decreased more than 50% during the last decade as a result of
simultaneously changed consumer habits, emigration from the area and discontinuance of industrial activity. As a consequence, residence times in the distribution system increased more than twofold. During long and hot summers, this can result in water temperatures in the distribution system of more than 20 jC. As an example, consider the case of a German city in which drinking water consumption in the city had decreased about 70% over a period of 10 years since the German reunion. The drinking water distributed to the city is produced from a well-protected ground water source by aeration and iron and manganese removal. In the finished water, HPC-bacteria, determined according to the German Drinking Water Regulation (HPC(GDWR)), are usually not detectable or are at least below 5 CFU/ml. The water is distributed without disinfection. At the end of a very hot summer, a regrowth incident occurred. HPC(GDWR) approached and sometimes exceeded the guidance value of 100 cfu/ ml. At the same time no fecal indicators or other hygienically relevant bacteria were detected. In Germany, when the guidance values are violated, this has to be reported to the local health authorities immediately. They have to decide, usually in consultation with the utilities and purveyors, which measures are to be taken immediately. It is recom-
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
mended and advantageous that a thorough plan of action be developed in case of regrowth, including expected outcomes. In the case described, discussions between the health authorities, the water works, the purveyor and a consultant were begun immediately. After detailed analysis and identification of points of pronounced regrowth and after consideration of possible effects of chlorination on the distribution system (destabilization of incrustations), a unanimous decision was made to fight the regrowth problem by changing the flow regime, thus decreasing residence time. Management of flow through tanks was altered to minimize residence time. Hydrants were flushed. Furthermore, water that formerly had been pumped around the city and distributed to the backcountry was now led through the city in order to maximize flow rates in the city network. This resulted in only an insignificant increase in total residence time of the water distributed to the backcountry. These measures proved to be very successful. Colony counts started to decrease after a few days and were back to normal levels (below 10 cfu/ml GDWR) in 3 weeks.
5. Establishment of HPC(R2A) for early regrowth detection After the regrowth incident had been coped with successfully, it was discussed how such events could be prevented in the future. AOC in the finished water leaving the water works was analyzed several times a year and was found between 20 and 30 Ag/l. The measures taken to shorten water residence time in the distribution system were continued and even improved. However, as the regrowth incident had occurred very suddenly, we looked for a method to detect upcoming regrowth. The HPC-method according to the GDWR is carried out using pour plates with a nutrient rich agar (beef extract, peptone at 1% each) incubated for 2 days at 20 jC and 36 jC. The method, together with the guidance values of 100 cfu/ml, has proven to be very successful in ensuring hygienically safe drinking water. However, based upon the authors experiences, it has proven to be too insensitive for early detection of upcoming regrowth in distribution systems with
321
low bacterial concentrations. Therefore, other culture methods were investigated. Reasoner and Geldreich (1985) developed a plate count method to enumerate culturable bacteria from nutrient-poor environments. The derived HPC-method Anonymous (1995b) using a spread plate technique on nutrient-poor R2A agar and with 7-day incubation (HPC(R2A)) has generally proven to be much more sensitive than pour plate methods using nutrient-rich agar. Therefore, it was decided to compare this method and the GDWR plate count method for their suitability for early regrowth detection. During two summers, HPC(R2A) was applied routinely besides the HPC(GDWR) on samples from the non-chlorinated distribution system. The plate counts obtained with the two methods were compared statistically. Fig. 3 shows the statistical comparison of plate counts obtained from samples collected from the water works outlet and at points A and B on two different distribution paths. The regrowth-factor is calculated as the bacterial concentration at points in the distribution system divided by the bacteria concentration at the point of entry. The analysis shows clearly that a multiplication of HPC(R2A) takes place from the system entry to point A by a factor of 6 and by a factor of 12 to point B. However, with HPC(GDWR) no multiplication can be detected. This is due to the fact that the bacteria concentrations are very low and the HPC(GDWR) method is too insensitive to detect (at these low concentrations) statistically significant plate count differences between samples from different points. For the samples analyzed (from the tree points entry, A and B) the more sensitive HPC(R2A) method yielded counts up to 400 cfu/ml, whereas the HPC(GDWR) method always yielded less than 7 cfu/ml. As for all plate count methods, the standard error for counts below 10 cfu/plate is considerably high (standard error F 3 cfu/plate according to the authors’ experience), differences in plate counts for low numbers are just the result of scatter and cannot be interpreted. Therefore, with HPC(R2A) an increase in colony counts can be observed when no counts are detectable with the HPC(GDWR)-method. In an upcoming regrowth event, once the detection limit of plate counts with HPC(GDWR) is exceeded, bacteria are already in the exponential phase and grow so fast that the
322
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
Fig. 3. Statistical analysis of bacterial counts (cfu/ml) according to the German drinking water regulation method and the HPC method on paths A and B in the distribution system (regrowth factors shown).
guidance value of 100 cfu/ml could be violated in a few days. However, with the HPC method regrowth events could be detected early and preventive measures taken.
The German Drinking Water Regulation since 2001 allows the application of other plate count methods than HPC(GDWR). However, in order to be able to value the results, determinations with the new
Fig. 4. Results of regrowth experiment for comparison of HPC(R2A) (left ordinate) and HPC(GDWR) (right ordinate).
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
method have to be carried out in parallel with HPC(GDWR) over a period of 1 year. At the time we were looking for a sensitive method for regrowth control, we also wanted to establish an alarm value for HPC(R2A) at which increased awareness for possible bacterial regrowth should be given. In order to establish such an alarm value, several regrowth experiments were carried out in the lab and HPC(R2A) and HPC(GDWR) determined in parallel. Fig. 4 shows the results of such an experiment. From this graph, it can be shown that regrowth occurs early with HPC(R2A) but not with HPC(GDWR). Colony counts using the HPC(R2A) techniques are higher than HPC (GDWR) by factors of between 100 and 200. From these results the purveyor set an internal alert limit of 1.000 cfu/ml HPC(R2A). Once this value was exceeded colony counts according to HPC(GDWR) higher than 10 cfu/ml were expected and special attention was taken to keep residence time in the distribution system low. A limitation for the use of HPC(R2A) for early regrowth control may be seen in the long incubation time of 7 days. A week would have been gone until the laboratory results were available. Therefore, over 1 year and at six different sampling points in the distribution system HPC(R2A), counts were also determined after 3 days of incubation. After outlierelimination (rm-test (Grubbs (1969), a = 10%) the results were compared by linear regression as shown in Fig. 5.
323
This showed that HPC(R2A) after 7-day incubation was higher by a factor of approximately 3 than HPC(R2A) after 3-day incubation. Thus, the purveyor’s internal alarm value for HPC(R2A) after 3day incubation was set to 300 cfu/ml.
6. Methods ability to identify regrowth—a model simulation Up to this point, the investigations and data analysis showed method sensitivities and interrelations. It became evident that HPC(R2A) at 7-day as well as at 3-day incubation are better suited to detect regrowth in the system early. However, the question remained how early or at what time ongoing regrowth would be detectable. To check for this a model simulation was carried out. For the model calculations, bacterial growth on AOC with an initial concentration of 30-Ag acetateC-equivalents/l flowing through a pipe was assumed. For simplification, bacterial growth was assumed to take place in the liquid phase only. With reference to the practical situation, it was assumed that the waters residence time in the pipe was 2 days and that it increased linearly over a period of 100 days to a residence time of 5 days. Also during the same period water temperature at the end of the pipe was assumed to increase from 10 to 25 jC linearly with time. The temperature gradient was modeled
Fig. 5. Comparison of HPC(R2A) data after 3 and 7 days of incubation at six different sampling points in the distribution system.
324
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
linearly over the length of the pipe. The plot was made up for sampling intervals of 7 days. For method precision, F 3 counts/plate was used as standard error for counts below 10 cfu/plate (authors experience) and F 25% for counts higher than 10 cfu/plate (authors’ experience, n = 6, and Mikell et al. (1996)). The results of the simulation are shown in Fig. 6. For interpretation it should be noted that the ordinates on the left and right are different. HPC(R2A) (7-day incubation) are plotted to the left, HPC(GDWR) to the right. The confidence intervals shown are for 67% confidence, i.e., if a large number of measurements were made, 67% of the results would be found between the upper and lower limits of the confidence interval. For data points without a confidence interval shown, it is as broad as or narrower than the symbol size. In order to identify growth, at least three following data points (two intervals) need to show a significant increase. In order to distinguish two measured data points from each other, it is best that the confidence intervals do not overlap. Analysing Fig. 6 on this basis yields that HPC(GDWR) growth will not be detectable until day 35, as this is the first time when two increases may have been identified; whereas for HPC(R2A), regrowth will be identified with the results of the sample from day 14. Taking into account the incubation times of 2 days for HPC(GDWR) and 7 days for HPC(R2A), it turns out that the bacterial regrowth in the pipe will be identified at day 21 with HPC(R2A)
and at day 37 with HPC(GDWR). Thus, the regrowth event will be evident about 16 days earlier with HPC(R2A) than with HPC(GDWR). This gives 16 days more time to react and take further measures against regrowth.
7. Conclusions Pour plate counts according to the German Drinking Water Regulations (HPC(GDWR)) are not sensitive enough to detect regrowth in distribution systems early and to enable the purveyor to react in time. Often no increase in colony-forming units can be observed over a long period of time and then suddenly the guidance value of 100 cfu/ml is exceeded. The HPC method with nutrient-poor R2A agar HPC(R2A) is much more sensitive. Colony counts obtained after 7 days were higher than HPC(GDWR) by factors between 100 and 200. HPC(R2A) counts after 3 days were lower by a factor of 3 than counts after 7 days. Model simulations showed that the application of HPC(R2A) may allow approximately 2-week earlier reaction on an upcoming regrowth event. For a German purveyor who was aware of possible regrowth during summer months, HPC(R2A) was established successfully for early regrowth detection. An internal alarm limit was set to 1.000 cfu/ml for HPC(R2A) 7-day incubation and special attention to keep residence times low was taken once this value was exceeded.
Fig. 6. Simulation of bacterial concentrations to be measured with HPC(R2A) and HPC(GDWR) at the end of the model pipe.
W. Uhl, G. Schaule / International Journal of Food Microbiology 92 (2004) 317–325
Acknowledgements The work of Stephany Hoff and Annika Eichelhardt at IWW who thoroughly carried out the AOC determinations and regrowth experiments as well as the contribution of our former colleague Dr. Markus Gerlach are greatly acknowledged. We furthermore would like to express our deep gratitude to our clients and partners who supplied data from full-scale distribution systems for this study.
References Anonymous, 1995a. Heterotrophic plate count. In: Am. Public Health Org./Am. Water Works Ass./Water Env. Fed. (Ed.), Standard Methods for the examination of water and wastewater. Chap. 9215. Washington, 9-32-9-39. Anonymous, 1995b. Assimilable organic carbon (proposed). In: Am. Public Health Org./Am. Water Works Ass./Water Env. Fed. (Ed.), Standard Methods for the examination of water and wastewater. Chap. 9217. Washington, 9-39-9-43. Anonymous, 1998. Richtlinie 98/83/EG des Rates vom 3. November 1998 u¨ber die Qualita¨t von Wasser fu¨r den menschlichen Gebrauch. Amtsblatt der Europa¨ischen Gemeinschaften DE-5.12.98, L330/32 – L330/54. Anonymous, 2001. Verordnung zur Novellierung der Trinkwasserverordnung vom 21. Mai 2001. Bundesgesetzblatt Teil 1 24, 959 – 980. Grubbs, F.E., 1969. Procedures for detecting outlying observations in samples. Technometrics 11 (1), 1 – 21. Haberer, K., 1994. Der Einsatz von Desinfektionsmitteln und das Auftreten von Desinfektionsprodukten in deutschen Wasserwerken. GWF. Wasser, Abwasser 135 (7), 409 – 417. Hambsch, B., 1994. Wiederverkeimung von Trinkwasser in Abha¨ngigkeit von Wasserqualita¨t, Kontaktfla¨chen und Verweilzeit. DVGW-Schriftenreihe. Wasser 79, 203 – 212.
325
Hambsch, B., 1999. Distributing groundwater without a disinfectant residual. Journal-American Water Works Association 91 (1), 81 – 85. Hambsch, B., Werner, P., Frimmel, F.H., 1992. Bakterienvermehrung in aufbereiteten Wa¨ssern verschiedener Herkunft. Acta Hydrochimica et Hydrobiologica 20 (1), 9 – 14. Huck, P.M., 1990. Measurement of biodegradable organic matter and bacterial growth potential in drinking water. Journal-American Water Works Association 82 (7), 78 – 86. Joret, J.C., Levi, Y., Dupin, T., Gilbert, M., 1988. Rapid method for estimating bioeliminable organic carbon in water. Proc. 1988 Am. Water Works Assoc. Annual Conf. pp. 1715 – 1725. Laurent, P., Servais, P., Randon, G., 1993. Bacterial development in distribution networks—study and modelling. Water Supply 11 (3/4), 387 – 398. Mikell, A.T., Smith, C.L., Richardson, J.C., 1996. Evaluation of media and techniques to enumerate heterotrophic microbes from karst and sand aquifer springs. Microbial Ecology 31, 115 – 124. Reasoner, D.J., Geldreich, E.E., 1985. A new medium for the enumeration and subculture of bacteria from potable water. Applied and Environmental Microbiology 49, 1 – 7. Servais, P., Billen, G., Hascoet, M.C., 1987. Determination of the biodegradable fraction of dissolved organic matter in waters. Water Research 21 (4), 445 – 450. Servais, P., Laurent, P., Randon, G., 1995. Comparison of the bacterial dynamics in various French distribution systems. Journal Water Supply: Research and Technology. AQUA 44 (1), 10 – 17. Uhl, W., 2000. Wiederverkeimung von Trinkwasser, Teil 1-Grundlagen und Methoden zur Bestimmung der Verkeimungsneigung. Bbr. Wasser, Kanal- und Rohrleitungsbau 51 (12), 30 – 35. van der Kooij, D., 1990. Assimilable organic carbon (AOC) in drinking water. In: McFeters, G.A. (Ed.), Drinking Water Microbiology. Progress and Recent Developments. Springer, New York, pp. 57 – 87. van der Kooij, D., Visser, A., Hijnen, W.A.M., 1982. Determining the concentration of easily assimilable organic carbon in drinking water. Journal-American Water Works Association 74, 540 – 545. Werner, P., 1985. Eine Methode zur Bestimmung der Verkeimungsneigung von Trinkwasser. Vom Wasser 65, 257 – 270.