CHAPTER 2 Spatial and temporal variation of drinking water quality in ten small Quebec utilities

Table des matières

Overview. The first part of this study has allowed identifying two types of small municipal drinking water utilities in the province of Quebec: those that historically did not have problems with distribution water quality, i.e., nonproblematic utilities, and those that did have such problems, i.e., problematic utilities. That portrait focused on microbiological water quality and management strategies, while also attempting to disclose relationships between them and some important water quality and operational parameters. As such, the portrait gives a general overview of the situation of small Quebec utilities.

Although that portrait was a very important and necessary first step to understanding the situation of small Quebec utilities, it, nonetheless, had certain limits, since bearing essentially on data that have been gathered a number of years before the study, that is historical data. The question is whether or not the overall picture reflected through the portrait corresponds to the current situation of the portrayed utilities as for water quality all along the distribution systems, based on the opposition nonproblematic vs. problematic, and what are the potential water quality parameters explaining the observed differences between the two distinguished groups of utilities. This may allow identifying the parameters upon which it would be possible to act to achieve better water quality in each of the two utility groups, along with exploring the capacity of such utilities to simultaneously and effectively handle the acute disease risk, associated with micro-organisms, and the chronic health hazard tied to chlorination by-products. All of these raised questions made indispensable initiating fieldwork in the corresponding municipalities to find answers. That fieldwork, which represented the second part of this study, has been designed as a water sampling campaign aiming at studying the spatial and temporal variation of drinking water quality in a number of small Quebec municipal utilities.

Abstract. A comparative study relating to distributed water quality was undertaken in ten small municipal drinking water utilities in Quebec. All of these utilities apply direct chlorination to surface water or groundwater under the direct influence of surface water without any previous treatment. These utilities were divided into two groups: four utilities that had never or rarely served water infringing upon the provincial drinking water microbiological standards (relating to fecal and/or total coliform bacteria), and six utilities that very often infringed upon said standards. The objective of this study was to identify key parameters responsible for the differences between the two groups of utilities, to explore the capacity of studied utilities to simultaneously and effectively handle the acute disease risk associated with micro-organisms and the chronic health hazard linked to chlorination by-products, and to identify the parameters upon which it may be possible to act in order to achieve better water quality in each of the two utility groups. The study includes comparisons of characteristics of water quality at the source, chlorination conditions in the plant, and water quality from the entrance to the extremity of the distribution system. Results show that the differences between the two groups of utilities are associated essentially with maintained chlorine residuals and heterotrophic plate count bacteria populations in corresponding distribution systems and, to a lesser extent, to the applied chlorine doses. Subsequent multivariate analyses allowed identification of variables upon which utility managers may act in order to improve the quality of distributed water in each group of utilities. For the group of utilities that had very little or no infringement, these factors are related to disinfection levels, whereas for the group that often infringed upon quality standards, raw water natural organic matter content reduction through source water protection and raised chlorine doses and residuals appear to be the factors that may lead to better microbiological quality of distributed water.

Key words: drinking water, water quality, distribution systems, small utilities, Quebec

Résumé. Une étude comparative sur la qualité de l’eau d’adduction a été menée dans dix petits systèmes municipaux de distribution d’eau potable au Québec. Tous ces systèmes appliquent une chloration directe à de l’eau de surface ou à de l’eau souterraine sous influence directe de l’eau de surface, sans aucun autre traitement. Ces systèmes furent répartis en deux groupes : quatre systèmes qui n’ont jamais ou ont rarement distribué de l’eau dérogeant aux normes microbiologiques provinciales relatives à l’eau potable (en ce qui a trait aux coliformes fécaux et/ou aux totaux) et six systèmes qui ont très souvent dérogé auxdites normes. L’objectif de cette étude était d’identifier les paramètres clés responsables des différences entre les deux groupes, d’explorer la capacité des systèmes à l’étude à faire face simultanément et efficacement au risque de maladie aiguë associé aux micro-organismes pathogènes d’une part, et au risque de maladie chronique relié aux sous-produits de la chloration d’autre part, de même que d’identifier les paramètres sur lesquels il serait possible d’agir afin d’obtenir une meilleure qualité de l’eau distribuée par chacun des deux groupes de systèmes. L’étude comprend des comparaisons des caractéristiques de la qualité de l’eau à la source, des comparaisons des conditions d’ajout du chlore aux postes de chloration respectifs, et de la qualité de l’eau de l’entrée du système de distribution à l’extrémité de celui-ci. Les résultats montrent que les différences entre les deux groupes de systèmes de distribution d’eau potable sont principalement associées aux teneurs en chlore résiduel libre et au nombre de colonies de bactéries hétérotrophes aérobies et anaérobies facultatives (BHAA) dans les réseaux de distribution correspondants et, dans une moindre mesure, aux doses de chlore appliquées. Des analyses multivariées subséquentes ont permis l’identification de variables (ou facteurs) sur lesquels peuvent agir les gestionnaires des systèmes municipaux en vue d’améliorer la qualité de l’eau distribuée par chaque groupe de systèmes. Pour le groupe de systèmes qui n’avaient pas ou avaient peu de dérogations aux normes provinciales de qualité, ces facteurs étaient associés aux niveaux de chloration, tandis que pour le groupe qui dérogeait souvent aux normes susmentionnées, les facteurs qui pourraient mener à une meilleure qualité microbiologique de l’eau distribuée seraient la réduction de la teneur en matière organique naturelle de l’eau brute par une protection adéquate de la source, de même que le rehaussement des doses et des résiduels de chlore.

Mots-clés : eau potable, qualité de l’eau, système de distribution, petits systèmes municipaux, Québec

Small drinking water utilities have unique challenges: they have limited financial and technical resources, often lack full-time staff to manage the utility, and may be geographically isolated in rural areas where agricultural pollution is substantial. There are about 1,000 small municipal utilities (i.e., serving 10,000 people or less) in Quebec, which serve approximately 20% of the province’s population, or about one million people (Gouvernement du Québec 1997). Most of these utilities apply simple chlorination (to surface or groundwater) or no treatment at all (essentially groundwater). According to the Quebec Ministry of Environment (QME), small utilities are known to have more difficulty in ensuring distribution to their customers at all times of drinking water that complies with established standards (Gouvernement du Québec 1997). Indeed, the majority of violations of the 1984 Quebec drinking water regulations (QDWR) (Gouvernement du Québec, 1984) concerned utilities serving fewer than 5,000 people.

Like some other Canadian provinces, Quebec updated its drinking water regulations shortly after the E. coli outbreak in the small community of Walkerton (Ontario), in which 7 people died and 2,300 became ill due to contaminated water. The new QDWR issued in June 2001 added new parameters (e.g., disinfection efficiency requirements for inactivation of Giardia cysts, Cryptosporidium oocysts and viruses, as well as control over heterotrophic plate count –HPC– bacteria and atypical bacteria, etc.) and strengthened control over others (e.g., turbidity, trihalomethanes –THMs–, etc.) (Gouvernement du Québec 2001). In doing so, the 2001 QDWR make the challenges facing the province’s small utilities even greater, especially considering the fact that very little is known about these utilities and the quality of the water they serve. For instance, as a direct consequence of the 2001 QDWR, practically all utilities which directly chlorinate surface water will have to either apply filtration or opt for groundwater sources. In the U.S., small water utilities using either surface or groundwater will, in the near future, have to comply with new National Primary Drinking Water Regulations (USEPA 1989; USEPA 1998a; USEPA 1998b; USEPA 2000). Additionally, as more and tighter regulations to enhance public health protection take effect, the cost of providing safe drinking water in compliance with the updated regulations will increase.

This article presents a study of spatial and temporal variation of distributed water quality in ten (10) small utilities in Quebec. All of the utilities have chlorination as the only treatment applied and use surface water or groundwater under the direct influence of surface water. Four (4) utilities that historically did not have problems with microbiological water quality (relating to total coliforms) and six (6) that did have such problems are compared through microbiological and physicochemical water quality. The objectives of this study were: 1) to identify key parameters responsible for the differences between the two groups of utilities; 2) to explore the capacity of studied utilities to simultaneously and effectively handle the acute disease risk associated with micro-organisms and the chronic health hazard tied to the presence of chlorinated disinfection by-products – DBPs – in drinking water; and 3) to identify the variables (i.e., parameters) upon which it may be possible to act upon in order to achieve better water quality in each of the two utility groups. Such information may be important for managers of small utilities and for government officials in terms of policy making.

Under the provisions of the 1984 QDWR, all utilities serving 51 or more people had to send results of their microbiological and physicochemical distribution water testing to the QME at a frequency related to their size. It is important to note that the same requirement is valid in the 2001 QDWR for utilities serving 21 or more people. In a database of small municipal utilities obtained from the QME in 1999, and containing data gathered by virtue of the 1984 QDWR follow-up, it was possible to distinguish between two types of utilities. The first type included utilities that had never recorded coliform positive samples or had recorded such samples only on rare occasions. The second type encompassed utilities that often recorded coliform positive samples. For the purpose of this research, data from three years (i.e., 1997 through 1999) were utilized. Based on data received from the QME, two concepts were defined: coliform episode and problematic utility . A coliform episode indicated one or a set of coliform positive samples occurring in a given distribution system during the three-year period (1997-1999), separated by at least 15 days from any other coliform positive sample in the same system. A problematic utility was defined as a utility that recorded one or more coliform episodes in at least two of the three reference years. Consequently, utilities that recorded no coliform episode, or had episodes in only one of the above-mentioned three years, were called nonproblematic utilities .

It is important to note that this research had been originally designed with the main goal of finding some responses and/or giving some explanations with respect to a statement of fact made by the QME in its document entitled “L’eau potable au Québec. Un second bilan de sa qualité : 1989–1994” (Gouvernement du Québec 1997) where the term “réseaux problématiques” was used to designate small or large distribution systems that frequently recorded coliform occurrences between 1989 and 1994). That statement of fact can be formulated as follows: utilities that have comparable technical, human and financial resources may be very different as for their historical microbiological water quality, and this is particularly frequent among small utilities. Moreover, in the QME small utility database originally used to determine the concept of “problematic utility”, and that contained results of about 65,000 water sample analyses for the period from January 1997 to December 1999, it was found that about 25 percent of the 927 small utilities (that is about 230 utilities) experienced repetitive coliform episodes. It was that fact that led to the division of small utilities into “nonproblematic” and “problematic” based on their historical microbiological water quality. The ten small utilities described in this paper have been chosen among the 927 mentioned earlier, with the “microbiological status” they had (i.e., having already been classified as problematic or not) in the initial QME database according to the number of coliform episodes they experienced from 1997 to 1999. It is reasonable to think that even though the historical data received from the QME may be resulting from periodic and sparse monitoring of bacteriological samples in distribution systems, drawing valid conclusions is possible when a high number of utilities (927 utilities, with 65,000 water samples analyzed) is involved and multiyear historical data are available for each utility.

In order to generate information about microbiological and physicochemical water quality of the 10 utilities under study on a spatial and temporal basis, five sampling campaigns were undertaken between May and October 2001. In the Quebec City area, this period encompasses spring, summer and fall conditions, the period during which surface water quality varies considerably with water temperatures generally higher than 5 °C (for the rest of the year, the ice cover protects surface waters naturally from runoff-related contaminants). The period from May to October corresponds relatively well to the critical period for microbial growth within distribution systems, with subsequent biofilm development, odour and taste problems and other problems.

Each utility was sampled five times, or once a month (May, June, July, August and October), at four different sampling points: raw water, chlorinated water (i.e., water from the chlorination facility outlet), water from the central part of the distribution system, and

Figure 2.1. Localization of the ten small utilities

water from the system’s extremity. In each campaign, 10 water quality parameters were measured: three microbiological (total coliform, HPC and atypical bacteria) and seven physicochemical (temperature, pH, turbidity, total organic carbon –TOC–, ultraviolet absorbance at the 254 nanometer wavelength –UV254 nm–, free chlorine residuals and THMs). Three parameters (temperature, pH and free chlorine residuals) were measured on-site, whereas the seven others were measured in the laboratory at Laval University. An important operational parameter, the chlorine dose, was also taken into consideration. The chlorine dose value was obtained either directly from the operator’s report book or calculated from utility meter readings and the quantity of chlorine utilized.

Samples for bacteriological testing were collected in Nalgene® polypropylene 500-mL screw capped bottles. Before sampling, 2 mL of sodium thiosulfate 5% w/v were added to the bottles, which were then sterilized in an autoclave for 15 minutes at 135 °C. All samples were collected after flushing the spigots for 3 to 5 minutes according to Standard Methods (APHA-AWWA-WEF 1998). Sterile bottles were only opened at the very moment of their filling, and were carefully handled to avoid potential extraneous contamination. Samples were then placed on ice for transport to the laboratory. Prior to any handling, the working surface was disinfected with a disinfectant soap (±4% v/v). Moreover, all handling was done near a flame to prevent extraneous contamination and maintain a sterile working zone.

The three microbiological parameters used for the purpose of this study were total coliform bacteria, HPC bacteria and atypical bacteria. As in the U.S. (USEPA 1993), the presence of coliform bacteria is used by Quebec drinking water professionals as an indicator of possible microbiological contamination. Based on World Health Organization (WHO) reports, coliform bacteria are the micro-organisms most commonly used to assess drinking water quality around the world (OMS 1994).

HPC bacteria are found both in bulk water and biofilm. HPC bacteria may be good indicators of the overall microbiological quality of distributed water. According to QME, these bacteria may be even better indicators than total coliforms (Gouvernement du Québec 1997). LeChevallier et al. (1990) mentioned that HPC bacteria could interfere with the coliform analysis. Others emphasize that abnormally high HPC counts could cause taste and odour problems in tap water (Pipes 1982; Reasoner 1990). Levels of HPC bacteria may also be used to assess microbial growth on distribution pipe surfaces and to measure bacterial after-growth in water mains (LeChevallier et al.1990; Carter et al. 2000).

Atypical bacteria may also be considered as distribution water quality indicators (Gouvernement du Québec 1997). This group of bacteria is somewhat difficult to define, because it encompasses a number of genera and species. These bacteria are able to grow on m-Endo LES medium but their colonies may or may not show the green sheen typical of true coliform bacteria; i.e., they are atypical. Atypical bacteria counts higher than 200 cfu/100 mL may hinder coliform detection in water samples, since the latter may not be able to grow under such conditions (Gouvernement du Québec 1997). So, from the strict point of view of a potential threat to public health, it appears that high HPC bacteria counts are less harmful than high atypical bacteria counts. The reason is that high HPC bacteria counts may only cause organoleptic degradations of distributed water quality, while high atypical bacteria counts may indicate the presence of harmful organisms in distributed water.

Total coliforms were enumerated by the membrane filter procedure with 0.45-µm-pore-size membrane filters and m-ENDO LES (APHA-AWWA-WEF 1998). This culture medium is the standard medium for total coliform testing in the U.S.; coliform colonies have a typical metallic green sheen. The coliform plates were incubated for 48 ± 2 hours at 35 ± 2 °C rather than 24 ± 2 hours at 35 ± 0.5 °C because the authors used method 9225-C of Standard Methods for the Examination of water and wastewater (APHA-AWWA-WEF 1998) rather than method 9225-B. Culture purification has not been done and no confirmation test has been performed. As a consequence, the total coliform results must be considered as presumptive total coliform counts. HPC bacteria were enumerated by the spread plate procedure with R2A agar incubated at 35 ± 2 °C for 48 ± 2 hours. Atypical bacteria were enumerated on the same filter media as total coliforms. Even though m-ENDO LES is considered a selective medium, some other bacterial species can grow on it. Hence, all colonies not showing the metallic green sheen were classified as atypical.

Controls were prepared from sterile demineralized water for all bacteriological analyses. The procedures for preparing the controls were the same as those used for the samples. This ensured that no extraneous contamination took place and skewed the results. Moreover, all microbiological determinations were performed in triplicate to ensure that the results were reproducible and, once again, to ensure that observed data were not distorted because of potential extraneous microbial contamination. Incubation temperatures and duration for coliform and atypical bacteria were identical to those mentioned for HPC. The enumeration was performed by counting the colonies on filter media. Since triplicate analyses were available, a mean colony number was calculated from the three obtained results. This number was then converted into cfu/100 mL, for coliform and atypical bacteria, or into cfu/mL for HPC bacteria. Please note that all microbiological sampling campaign data are shown in Appendix C.

Physicochemical parameters analyzed during the sampling campaign were temperature, pH, turbidity, TOC, UV254 nm, free chlorine, and THMs. Temperature, pH, and free residual chlorine were measured at the sampling sites. For turbidity, TOC and UV254 nm, Nalgene® polypropylene or high-density polyethylene bottles were used to collect water samples. The water temperature was measured using a standard glass alcohol column thermometer (Fisher–14–997) or, when pH and temperature were measured at the same sampling point, with the use of a thermocouple probe. Water pH was measured using an Accumet® model 25 pH/ion-meter and a Hanna HI–1332–B probe. Turbidity was measured using a Hach 2100–N turbidimeter. Thirty milliliter (30 mL) sample volumes were used to measure this parameter once the instrument was calibrated against a secondary standard consisting of a metal oxide suspension in a gel. Free residual chlorine was determined by the DPD (diethyl-para-phenylenediamine) colorimetric method using a Hach DR/890 colorimeter and Hach DPD free chlorine reagents (using 10 mL water samples) (APHA-AWWA-WEF 1998). The procedure used for water organic carbon measurement actually determined not TOC but non-purgeable organic carbon (NPOC). NPOC was measured by authors instead of TOC only because of laboratory restrictions (the analyzer available at the laboratory was not designed for samples high in inorganic carbon, therefore it becomes necessary to purge the acidified samples). Moreover, the authors assumed that the fraction of volatile organic carbon (VOC) was negligible in waters analyzed in this study, as is generally the case for natural raw waters. This is why NPOC content was considered approximately the same as TOC content and interpreted as such. TOC was determined by means of a Shimadzu TOC-5000 total organic carbon analyzer. The method consisted of 200 μL HCl-acidified water samples aerated with pure ultra zero air (Praxair Specialty Gases and Equipments) in order to remove inorganic carbon. Finally, UV254 nm was determined using a Jenway 6405—UV/Vis (ultraviolet and visible) spectrophotometer at 254-nanometer wavelength (using a 1 cm Suprasil® quartz cell).

Samples for THM determinations were collected in 300 mL glass BOD (biochemical oxygen demand) bottles, which are airtight thus avoiding THM volatilization. Before collecting samples, a standardized dose (approximately 500 mg/L) of a dechlorinating agent (sodium thiosulfate or ammonium chloride) was added to each bottle, which was then placed in a Thelco® Model 18 PS (Precision Scientific) incubator for 12 hours at 110 oC to evaporate the water. THMs were then measured using a Perkin Elmer (Autosystem XL) gas chromatograph (GC) equipped with an electron-capture detector and a ZB–624 column (30 m X 0.32 mm ID X 1.8 μm FT). Analytical criteria for this determination were injector, oven and detector temperatures (175 oC, 80 oC and 375 oC, respectively), carrier gas (helium: 8.5 mL/min during 7 minutes, followed by a flow ramp of 3.5 mL/min until 15 minutes, then by a flow of ramp of 15 mL/min during 11 minutes), make-up gas argon/methane P5 Mix at 30 mL/min (Praxair Specialty Gases and Equipments) , analysis duration (17 minutes) and injection volume (1 µL of the sample). The THMs in the water sample are concentrated by liquid-liquid extraction with pentane. GC analysis was conducted based on USEPA method 551.1 described by Rodriguez and Sérodes (2001). Please note that all microbiological sampling campaign data are shown in Appendix C.

Table 2.1 presents general characteristics of the 10 studied utilities. As shown, only two utilities obtain their raw water from lakes; the others obtain it from surface wells (in the form of springs with a single basin or with horizontal drainpipes). All of them use chlorination as the only treatment, using a 12% sodium hypochlorite solution. Only two utilities are located in municipalities under very high agricultural pressure.

* This factor is measured by the annual balance of phosphorus in terms of kilograms of phosphorous (P2O5) per hectare. It considers the total manure production within the municipality, the nutrient requirements of crops and the cultivated area. When the annual balance is more than 20 kg P2O5/ha/year or when the municipality is located in watersheds with already significant phosphorus excess in the soils, the Quebec provincial government considers the municipality as being in manure surplus. Even if such an annual balance is not calculated based on watershed limits but rather on municipal limits, it can be used as an indicator of the susceptibility of surface waters to be contaminated by surface or subsurface runoff.

Raw water characteristics of nonproblematic utilities were compared to those of problematic utilities in order to depict potential differences between them at the source water stage. To carry out this comparison, a number of key raw water parameters were chosen: turbidity, TOC, UV254 nm, total coliform, HPC and atypical bacteria. Abnormally high counts of coliforms in source water would indicate poor quality requiring steady disinfectant (i.e., chlorine) residuals to prevent breakthrough or regrowth of these organisms in the distribution system. The absence of a treatment (e.g., coagulation, flocculation, settling, filtration) to remove colour (due to natural organic matter – NOM –) and suspended matter means that TOC, UV254 nm (used as indicators of organic matter in drinking water) (Krasner 1999), and turbidity will enter the distribution system in levels comparable to those encountered in the source water. This situation may reduce disinfection efficiency (McCoy and Olson 1986), increase chlorine demand and favour bacterial breakthrough, regrowth or recovery (LeChevallier et al. 1996). It is important to mention that assimilable organic carbon – AOC – and biodegradable organic carbon – BDOC – are better indicators of the availability of microbial nutrients in drinking water (van der Kooij et al. 1999), but these indicators were not defined in this study.

Water temperature, an important parameter for microbial growth, remained relatively low and showed little variation. This is understandable, since most of the utilities obtain their water from surface wells, with the water remaining relatively cool, even during summer months, because of natural protection from the rays of the sun. Overall, the average raw water temperature recorded was 8.9 oC (with the 10th percentile = 6.5 oC and the 90th = 15.3 oC).

In general terms, the raw water appeared to be of good quality in both groups of utilities. Indeed, only two of them draw their raw water from strictly surface water sources, —i.e., lakes—; the others supplied from surface wells. Consequently, during the study period, the average raw water turbidity, UV254 nm and TOC levels were low. Average values for those parameters were 0.5 ntu, 0.046 cm-1 and 1.7 mg/L, respectively (for the two lakes, the average values were 1.16 ntu, 0.115 cm-1 and 3.68 mg/L, respectively). Moreover, previous studies showed that these values could be much higher in surface raw waters in southern Quebec (Milot et al. 2000; Vinette 2001). Raw water total coliform counts also appeared very low in comparison with other southern raw waters in Quebec (Payment et al. 2000).

Figure 2.2-a to Figure 2.2-f feature some differences with regard to source water physicochemical quality between the two groups of utilities. These figures illustrate group differences as well as monthly differences. In both groups, the highest mean values for these three parameters (corresponding to lesser water quality) were observed in June and, in nonproblematic utilities, to a lesser degree in October. For turbidity, measured value variations appeared higher in nonproblematic utilities, but mean values are relatively comparable for all months between the two groups of utilities, with the exception of June. For UV254 nm, mean values were generally higher in nonproblematic utilities, but maximum values were much higher in problematic utilities. UV254 nm appeared subject to important value fluctuations in the problematic utilities: minima were very low, maxima relatively high. As for TOC, June and October exhibited relatively important differences in terms of mean values when the two groups of utilities were compared.

Differences with respect to raw water microbiological quality are shown in Figure 2.3-a toFigure 2.3-f. At first glance, monthly differences appear much greater between the two groups for microbiological raw water quality than they are for physicochemical quality. Hence, for total coliforms, June and October showed the highest counts in nonproblematic utilities, whereas July was the month with the highest count in problematic utilities. For all other months, even maximum values rarely reached 30 cfu/100 mL. It is important to note however, that for all months other than June and October, problematic utilities recorded higher counts in terms both of mean values and maxima. This may be an indication that problematic utilities are at more frequent risk of coliform contamination. Another important indication is that the highest counts for total coliforms correspond to the highest measured values of turbidity, UV254 nm, and TOC in both nonproblematic (June and October) and problematic utilities (June, July, August); however, this appears particularly clear in nonproblematic utilities. As far as HPC bacteria are concerned, May and June are the two

Figure 2.2. Comparison of raw water quality between nonproblematic (NP) and problematic (P) utilities: a and b, turbidity; c and d, TOC; e and f, UV254 nm . Bar, mean value; upper bar, maximum; lower bar, minimum

Figure 2.3. Comparison of raw water quality between nonproblematic (NP) and problematic (P) utilities: a and b, total coliforms; c and d, HPC bacteria; e and f, atypical bacteria. Bar, mean value; upper bar, maximum; lower bar, minimum. Atypical bacteria quantification limit was 400 cfu/100 mL.

months that showed significant differences between the two groups. This parameter appears less influenced than the three above-mentioned physicochemical ones, but the fact that it exhibits its highest values in July and August may indicate a greater dependence on temperature. Atypical bacteria counts show their highest mean values in June and October in nonproblematic utilities. It is interesting to note that the same trend is observed for total coliform counts in these utilities. In problematic utilities, May, July, and August recorded the highest mean values of atypical bacteria counts. These counts show a monthly trend that is different from the one exhibited by coliform counts in problematic utilities; rather, it indicates a much greater similarity with HPC bacteria counts in the same group (as shown by comparing monthly trends for May, July, and August). This fact is surprising, since the same type of bacteria seemed to behave differently depending on the group of utilities. In fact, for problematic utilities, atypical bacteria appear to behave in a way similar to HPC bacteria, having their numbers boosted by warm water temperatures (that favour bacterial growth and multiplication), especially in July and August. In nonproblematic utilities, the similarity in monthly trends with total coliform counts, as well as turbidity, UV254 nm, and TOC, may suggest a possible impact of these three physicochemical parameters on total coliform and atypical bacteria counts. It is important to note that in Figure 2.3-e and Figure 2.3-f maximum atypical bacteria counts are assumed equal to 400 cfu/100 mL because the colony counting method utilized did not allow counting of plates over 400 cfu/100 mL. Such value has been considered as maximum although some monthly atypical bacteria counts may in reality be higher.

As a conclusion to these monthly water quality profiles, it must be noted that the differences in raw water quality between the two groups of utilities were not great. In order to determine if the observed differences are statistically significant, a test of means (independent samples t test) was performed. It indicates, as shown in Table 2.2, that the observed mean differences between nonproblematic and problematic utilities were not significant (the difference being statistically significant when P<0.10). The results of this statistical analysis suggest that other factors (related for instance to disinfection practices, distribution system management or properties, etc.) could have a greater impact than the raw water on the microbiological water quality in the distribution system.

* Means test significance level (Student’s t-test)

A comparative study of the water quality at the chlorination facility outlet and in the distribution system was also carried out. Emphasis was placed on the applied chlorine dose, the HPC and atypical bacteria, as well as on residual chlorine and chlorination by-products (THMs).

All the utilities in this study use chlorination as the unique treatment. The chlorination dose is an important operational parameter that affects micro-organism inactivation, available residual chlorine and DBP occurrence in distributed water (Connell 1996; Rodriguez et al. 1999). Figure 2.4-a and Figure 2.4-b feature this parameter. The mean values were relatively close, although generally slightly higher in nonproblematic utilities (Figure 2.4-a). However, monthly standard deviations (as much as 2.90 mg/L in July) and maxima in the latter are almost twice as high as those of corresponding sampling months for problematic utilities. Nonetheless, the subsequent means test performed for the dose confirms that the statistical difference between the mean values is not significant ( P = 0.352, mean value for nonproblematic utilities = 2.63 mg/L and the mean value for problematic utilities = 2.19 mg/L). Because of a lack of adequate data (for example, no information was available about the hydraulic efficiency factor), the disinfection effectiveness (i.e., the CT values) could not be estimated. Thus, the emphasis was placed on the impact of the dose on microbiological water quality, as well as on the free residual chlorine concentration and THMs.

Figure 2.4. Comparison of applied chlorine doses between a nonproblematic (NP) utilities and b problematic (P) utilities. Bar, mean value; upper bar, maximum; lower bar, minimum

During the five-month sampling campaign, seven total coliform positive samples for total coliforms were recorded. Three were found in samples from nonproblematic utilities (one at a chlorination facility outlet, two in the central part of distribution system) and four in samples from problematic utilities (two at a chlorination facility outlet, two in the central part of distribution system). It was surprising that three of the total coliform positive samples came from a chlorination facility outlet. This may have been the result of either poor chlorination effectiveness (bad mixing or insufficient contact time of chlorine with water) or extraneous contamination (for example, coming from operators). Likewise, it was surprising that none of the positive coliform samples came from a distribution system extremity which is known to be a location where free residual chlorine and water flowrate are generally low. However, the detected coliform cases could not be classified as violations of the QDWR in force, since none of them recorded more than 10 cfu/100 mL. As for HPC bacteria, three violations (i.e., > 500 cfu/mL) of the QDWR were recorded in distribution systems. All of them were from problematic utilities: one from the central part of the distribution system and two from the extremity of systems. It is interesting to note that problematic utilities appeared to exhibit a greater predisposition to HPC bacterial growth, particularly when considering the potential links that may exist between HPC and coliform bacteria (LeChevallier 1990). The fact that practically all violations linked to HPC bacteria came from a distribution system extremity also supports the widespread opinion that control samples for these organisms must be taken precisely at that location (Gouvernement du Québec 2001). Only one violation (i.e., > 200 cfu/100 mL) on the QDWR was recorded for atypical bacteria counts in a distribution system. It came from the chlorination facility outlet at a problematic utility. The fact that atypical bacteria were detected in abnormal numbers at the same location where coliforms were detected three times is interesting, considering the relationship that may exist between atypical and coliform bacteria (Gouvernement du Québec 1997). However, the chlorination effectiveness may also be a factor.

Average HPC counts in nonproblematic utilities were compared to those in problematic utilities (Figure 2.5-a, Figure 2.5-b). It appears that regrowth takes place in both cases, but the phenomenon is of greater magnitude in problematic utilities. Moreover, the means difference between the two groups of utilities is statistically significant ( P<0.10 ) for the chlorination facility outlet (Table 2.3). This indicates a possible deficiency in the disinfection procedures of problematic utilities (which exhibited a much higher mean count value). However, this difference is also important for the other sampling points, although not statistically significant. Since levels for indicators of organic matter were found to be relatively comparable between the two groups, the observed result seems to be related to: 1) insufficient mixing of recently- added chlorine with bulk water, or 2) very low residual chlorine in problematic utilities.

The results obtained for atypical bacteria showed no significant difference between the two groups of utilities in terms of growth (Figure 2.5-c, Figure 2.5-d). Likewise, statistical results of means tests proved not to be significant between the groups, despite obvious monthly fluctuations in average counts. It appears that HPC bacteria grew more actively in both utility groups than the atypical bacteria did, and monthly growth patterns were very different from those observed from compared raw water values.

Maintaining an adequate level of residual chlorine is of great importance in terms of distribution water quality management (Sérodes et al. 1998; Haas 1999). Figure 2.6-a and Figure 2.6-b indicate a considerable difference between free chlorine levels of nonproblematic and problematic utilities. Average concentrations of measured residual chlorine of nonproblematic utilities were higher than 0.2 mg/L during the period under study at the three sampling points: the chlorination facility outlet water, water from the central part of the distribution system, and water from the system’s extremity. Conversely, in problematic utilities, levels of free residual chlorine in both the central part and the distribution system extremity were on average lower than 0.1 mg/L. This level of residual chlorine may appear as a minimum in order to prevent microbiological deterioration of

Figure 2.5. Comparison of HPC bacteria and Atypical bacteria between a and c nonproblematic (NP) utilities; and b and d problematic (P) utilities. Bars represent monthly means (from left: May, June, July, August, October)

* Means test significance level (Student’s t-test) ** Calculated with assumed maximum value of 400 cfu/100 mL

Figure 2.6. Comparison of free chlorine and total THMs between a and c nonproblematic (NP) utilities; and b and d problematic (P) utilities. Bars represent monthly means (from left: May, June, July, August, October)

water quality within the distribution system since, according to Haas (1999), water distribution systems in the United States usually carry residuals more than 0.1 mg/L. The difference in average residual chlorine between the two types of utilities was found to be statistically significant ( P<0.10 ) for each of the three sampling locations (Table 2.3). Such results give a good indication of the benefits of sufficient levels of residual chlorine. Considering that indicators of chlorine demand related to raw water (TOC, UV254 nm, turbidity) were relatively close between the two groups of utilities, observed differences of residual chlorine levels are most probably associated with chlorination practices at the facility (applied chlorine doses, contact time, etc.) or to the presence of oxidizable material attached to the pipe surfaces in the distribution system. However, it is important to mention that there are many other factors that likely affect chlorine decay, including the type of TOC, ammonia, iron, etc.

When added to water, chlorine reacts with NOM, resulting in the formation of DBPs (Bellar et al. 1974; Rook 1974), which have a carcinogenic potential. Four THMs are the most commonly known group of DBPs: chloroform, bromoform, chlorodibromomethane and bromodichloromethane (Levallois 1997). The sum of the values of these four species is called total THMs. Because one of the authors’ goals is to explore the capacity of the studied utilities to simultaneously handle the acute disease risk associated with microorganisms and the chronic health hazard linked to the presence of DBPs in drinking water (Fowle and Kopfler 1986; Putnam and Graham 1993), total THMs were analyzed concurrently with microorganisms. THM levels observed during the sampling campaign are low (the average for nonproblematic utilities was: at the facility outlet, 14.4 μg/L, at the system extremity, 22.3 μg/L; for problematic utilities: at the facility outlet, 8.49 μg/L, at the extremity, 12.2 μg/L). This is because 8 out of 10 utilities extract their raw water from surface wells, such water sources containing relatively little NOM, the principal THM precursor. By comparison, the two utilities that draw their raw water from lakes had an average THM level of 33.0 μg/L at the facility outlet, and 49.6 μg/L at distribution system extremity. Mean differences for total THM appear significant at first glance (Figure 2.6-c and Figure 2.6-d). In fact, statistical tests show that mean total THM differences between the two groups are almost significant for water sampled at the chlorination facility outlet, and significant ( P<0.10 ) for water from the central part of the distribution system and for water at the system extremity (Table 2.3). In all three cases, average THM levels in nonproblematic utilities were almost twice the levels found in problematic ones. This result means that utilities that are nonproblematic from a microbial point of view may experience some difficulties with DBP occurrence (measured values being, however, notably below the maximum contaminant levels of the QDWR in force). This result is very probably related to the higher chlorine doses applied in nonproblematic utilities in comparison to problematic ones. Such doses ensure higher levels of residual chlorine in the distribution system in nonproblematic utilities, but also generate higher levels of THMs. In addition, one may also suspect the reactivity of the NOM to be a supplementary factor explaining such important differences (mean values of measured UV254 nm being slightly higher in nonproblematic utilities). These results underscore the difficulty small utilities experience by not having treatment before chlorination to efficiently handle microbial and chronic health risks simultaneously.

Previous analyses showed that factors explaining differences between nonproblematic and problematic utilities are the applied chlorine dose, free residual chlorine and HPC bacteria counts in distribution systems, with the last two factors showing significant differences. To better understand factors that significantly influence the distributed water quality, multivariate analyses were performed. The purpose of these analyses is to explain the water quality within each group of utilities after chlorination, i.e., from the chlorination facility outlet to the distribution system extremity.

Because of the nature of the variables to explain (called dependent variables), a linear multivariate regression analysis was performed. This analysis provides an estimate of the linear relationship between a dependent variable and one or more explanatory variables (called independent variables). In fact, the linear regression estimates the coefficients of the linear equation, involving one or more independent variables that best predict the value of the dependent variable (Norušis 2000). The statistical software package SPSS® 10.0 (SPSS for Windows 1999) was used to perform these multivariate analyses (based on a stepwise method for variable selection).

In these analyses, the variables to be explained were HPC and atypical bacteria counts and total THM levels. The development of multivariate regression models was carried out for each of these dependent variables for data from the following locations: the chlorination facility outlet, the distribution system (average of central part and system extremity values), and the distribution system extremity. Raw water quality and/or operational parameters (i.e., variables) were used to explain the chlorination facility outlet water quality. Similarly, chlorination facility outlet variables were used to explain distribution system and extremity water quality. The successful models (i.e., those that gave significant results; P<0.10 ) are shown in Table 2.4.

Multivariate analyses yielded interesting descriptive models. First, it must be noted that models related to THMs provided better results (higher R2 giving higher explained variance ratios) than those related to HPC bacteria. A possible explanation involves the very discrete nature of microbial dissemination within water distribution lines (presence in bulk water, attached to pipe wall, presence within corrosion tubercles). No significant multivariate model was found to explain atypical bacteria presence in the studied utilities, even though some variables (e.g., temperature, UV254 nm) showed significant correlations on a bivariate basis. This is possibly linked to the fact that this group of micro-organisms is a complex mixture of species (as mentioned earlier) for which the determining factors may be various, so that no one factor materializes as vital for all species.

On the whole, HPC bacteria presence at chlorination facility outlets was relatively well explained by variables bearing on the natural logarithm of HPC bacteria counts in raw water, and occasionally, NOM-related variables (UV254 nm, TOC) and pH. It is surprising

* Significance level for the variable § Significance level for the model

† Number of cases Not applicable

Pearson determination coefficient

to note that raw water temperature did not appear among the significant variables. This may be explained by the relatively low variability of water temperature during the period under study. It is interesting to point out that disinfectant-related variables (chlorine dose, free residual chlorine) appeared in most models relating to nonproblematic utilities, but in none of those bearing on problematic utilities. The chlorine dose appeared in the initial model, explaining the logarithm of HPC bacteria counts in raw water for nonproblematic utilities, but when TOC was introduced, the dose was removed. The model improved with TOC replacing the chlorine dose, but the latter showed a significant bivariate correlation with the logarithm of HPC bacteria. So, it is obvious that the dose plays a major role, even if it does not appear in the final model. It is also understandable that the chlorine dose and the TOC are antagonistic, since the former kills micro-organisms, whereas the latter contributes to their nutrition. UV254 nm appeared in all models relating to problematic utilities, which suggests that this parameter, reflecting NOM reactivity, may be much more critical for this group of utilities although its mean value appears lower than for nonproblematic utilities. For average total THMs within a distribution system and at its extremity, the prevailing explanatory factors in nonproblematic utilities are total THMs, free residual chlorine, and UV254 nm (all at the facility outlet). In fact, a study by Vinette (2001) with three large utilities demonstrated that up to 50 percent of total THMs present in distribution system could already be formed in water leaving the treatment plant or chlorination facility. However, in problematic utilities, only total THMs and NOM-related variables appeared. The absence of residual chlorine in these models may be explained by low variability of this variable in problematic utilities. As for total THMs at the facility outlet, they are well explained by raw water NOM-related variables.

The implications of these models in terms of distribution system management are the following: the nonproblematic utilities should place more emphasis on chlorine doses and residuals by applying appropriate doses and maintaining adequate residuals. In fact, utilities of this group tend to apply too much chlorine. Although this allows effective control of distribution system microbial flora, it may generate bad tap water taste (chlorine taste) and result in a relatively high potential for DBP formation. Conversely, in problematic utilities better management of chlorination doses and residuals would probably improve microbiological water quality by achieving better control of micro-organisms within the distribution mains. Better protection of source water would also allow these utilities increased control over NOM-related water quality parameters (especially UV254 nm), which appear critical to them, even though average values for these parameters are below those of nonproblematic utilities. The issue about controlling NOM is not easy and may appear really impractical compared to simply installing some sort treatment for example, but it does make sense to protect water sources, to the extent possible, from any kind of pollution where conditions for that exist and appear adequate. To end, controlling man-made wastes and wastes originating from human-directed activities (like those related to agricultural land use) remains an effective preventive measure that both utility groups should implement.

The results of this investigation effectively demonstrate that problematic utilities (defined as having recurrent occurrences of coliforms based on historical regulatory information) have lower overall microbiological water quality from the plant to the distribution extremity. Indeed, for all of the studied parameters that characterize treated and distributed water quality, except for THMs, the situation is better in nonproblematic utilities. During the time period of this study, all observed violations of the 2001 QDWR occurred in problematic utilities (three times for HPC bacteria, once for atypical bacteria counts).

The average applied chlorine doses appeared only slightly different between the two groups of utilities. However, significant differences in free residual chlorine levels were found between them all along the distribution system. The study results suggest that maintaining average residual chlorine in problematic utilities comparable to those used in nonproblematic utilities (i.e., about 0.7-0.8 mg/L at the facility outlet, and 0.3-0.5 mg/L at distribution system extremity) would likely bring about significant changes in microbiological water quality. It is noteworthy that significantly higher concentrations of THMs were observed in nonproblematic utilities.

Because of the characteristics of the raw waters used by the ten investigated utilities, nonproblematic utilities appear to be able to successfully deal with the challenge of efficient and simultaneous control of the acute disease risk (represented by pathogenic micro-organisms) and the chronic health hazard linked to DBPs, even if measured THMs were higher than those in problematic systems.

Univariate (means tests) analyses indicated that differences in distributed water quality between problematic and nonproblematic utilities are related to applied chlorine doses and, to a much greater extent, to HPC bacteria counts and free residual chlorine. As for multivariate models, they indicated that in terms of distributed water quality management priorities, nonproblematic utilities should devote more attention to appropriate, balanced disinfection practices and avoid continually overestimating the microbial risk. This would allow them to serve their customers with safe and palatable drinking water with a much lower incidence of chronic health hazards. Problematic utilities need to achieve better control of UV254 nm and TOC through adequate water source protection, combined with an important increase of chlorine doses and residuals. If these conditions were fulfilled, they could attain high, sustainable water quality standards.

The 2001 QDWR require that all surface water utilities conduct at least a filtration prior to chlorination. Most of the ten utilities chosen for this study are not typical surface water utilities. Eight of them obtain raw water halfway between surface and groundwater, i.e., from surface wells. Thus, these utilities do not fall directly into the category for which filtration is required. Therefore, they will have to scientifically demonstrate that they possess the technical and operational capabilities to produce water that consistently meets the new provincial standards without filtration. Understanding the parameters identified herein as explaining differences between nonproblematic and problematic utilities, as well as those identified as explanatory variables in multivariate analyses, may help these utilities, especially problematic ones, to find ways to comply with the 2001 QDWR standards. Nevertheless, the situation obviously will require significant readjustments in disinfection practices, and may necessitate hiring qualified operators. The utilities that will undoubtedly have to install filtration (e.g., the two utilities that draw their raw water from lakes), should devote significant attention to CT requirements and post-chlorination THM levels.

From a strict public health standpoint, it must be mentioned that water distributed by problematic utilities selected in this study does not necessarily present more of a threat than the water served by nonproblematic utilities. In fact, microbiological parameters considered in this study (i.e., total coliform, HPC and atypical bacteria) are essentially hygienically relevant and do not represent a real or direct disease risk, since it is recognized that these organisms are not normally pathogenic. Thus, more attention must be given to the potential relationships that may exist between the presence of these bacteria and the potential presence of real pathogens such as parasites, viruses and others.

In the future, the task of maintaining water quality that corresponds at all times to increasingly stringent standards will continue to be very challenging for small utilities, since they are handicapped by technical, managerial and financial deficiencies.

APHA-AWWA-WEF. 1998. Standard methods for the examination of water and wastewater. 20th Edition. American Public Health association, 1015, 15th Street NW, Washington D.C. 20005-2605.

Bellar, T.A., Lichtenberg, J.J., and Kroner, R.C. 1974. The Occurrence of organohalides in chlorinated drinking water. J. Am. Water Works Assoc. 66: 703.

Carter, J.T., Rice, E.W., Buchberger, S.G., and Lee, Y. 2000. Relationship between levels of heterotrophic bacteria and water quality parameters in a drinking water distribution system. Water Research 34: 1495–1502.

Connell, G. F. 1996. The chlorination/chloramination handbook. Water Disinfection Series. American Water Works Association, Denver, CO.

Fowle III, J.R., and Kopfler, F.C. 1986. Water disinfection : microbes versus molecules – an introduction of issues. Environmental Health Perspectives 69: 3–6.

Gouvernement du Québec. 1984. Règlement sur l’eau potable. Éditeur officiel du Québec, Québec. 7 p.

Gouvernement du Québec. 1997. L’eau potable au Québec. Un second bilan de sa qualité : 1989–1994. Ministère de l’Environnement et de la Faune, Québec. 36 p.

Gouvernement du Québec. 2001. Règlement sur la qualité de l’eau potable. Ministère de l’Environnement, Québec. 19 p.

Haas, C.N. 1999. Benefits of employing a disinfection residual. Journal of Water SRT—Aqua 48: 11–15.

Krasner, S.W. 1999. Chemistry of disinfection by-product formation. In: Formation and control of disinfection by-products in drinking water. P.C. Singer (ed.), AWWA, pp. 27-52.

LeChevallier, M.W., Schulz, W.H., and Lee, R.G. 1990. Bacterial nutrients in drinking water . In: Assessing and controlling bacterial regrowth in distribution systems. AWWARF (ed.), pp. 143–201. American Water Works Association Research Foundation, Denver, CO.

LeChevallier, M.W., Welch, N.J., and Smith, D.B. 1996. Full-scale studies of factors related to coliform regrowth in drinking water. Applied Environmental Microbiology 62: 2201–2211.

Levallois, P. 1997. Qualité de l’eau potable et trihalométhanes. Bulletin d’Information en Santé Environnementale (BISE) 8: 1–4.

McCoy, W.F., and Olson, B. H. 1986. Relationship among turbidity, particle counts and bacteriological quality within water distribution lines. Water Research 20: 1023–1029.

Milot, J., Rodriguez, M.J., and Sérodes, J. 2000. Modeling the susceptibility of drinking water utilities to form high concentrations of trihalomethanes.  Journal of Environmental Management 60: 155–171.

Norušis, M.J. 2000. SPSS® 10.0 Guide to data analysis. Prentice-Hall, Inc. Upper Saddle River, New Jersey 07458.

OMS 1994. Directives de qualité pour l’eau de boisson : recommandations. Deuxième édition, volume 1. Genève, Suisse.

Payment, P., Berte, A., Prévost, M., Ménard, B., and Barbeau, B. 2000. Occurrence of pathogenic microorganisms in the St. Lawrence River (Canada) and comparison of health risks for populations using it as their source of drinking water. Canadian Journal of Microbiology 46: 565-576.

Pipes, W.O. 1982. Introduction. In: Bacterial indicators of pollution , pp. 1-9. CRC Press.

Putnam, W.S., and Graham, J.D. 1993. Chemicals versus microbials in drinking water: a decision sciences perspective. J. Am. Water Works Assoc. 85: 57–61.

Reasoner, D.J. 1990. Monitoring heterotrophic bacteria in potable water. In: Drinking Water Microbiology: Progress and Recent Developments. G. A. McFeters (ed.), New York, Springer-Verlag, pp.453-473.

Rodriguez, M.J., Milot, J., Sérodes, J.-B., and Montixi, M.-D. 1999. A New modelling approach to simulate chlorine demand and trihalomethane formation in drinking water. Proceedings of the American Water Works Association Annual Conference, Chicago, Ill. 20-24 June.

Rodriguez, M.J., and Sérodes, J.-B. 2001. Spatial and temporal evolution of trihalomethanes in three water distribution systems. Water Research 35: 1572–1586.

Rodriguez, M.J., Coulibaly, H.D., and Banville, J. 2002. Strategies for ensuring a safe drinking water in small utilities of Quebec (Canada). Proceedings of the American Water Works Association Annual Conference and Exposition (ACE), New Orleans, La. 16-20 June.

Rook, J.J. 1974. Formation of haloforms during chlorination of natural waters. Water Treatment and Examination 23: 234-243.

Sérodes J.B., Rodriguez, M.J., and Ponton, A. 1998. Development and on-site evaluation of a decision-making tool for chlorine disinfection dose and residual control. Presented at the 8th National Conference on Drinking Water, Canadian Water and Wastewater Association (CWWA), Quebec City, Quebec, Canada. 28-30 October.

SPSS for Windows 1999. Release 10.0.0 standard version. Copyright® SPSS Inc.

USEPA. 1989. National primary drinking water regulations: filtration, disinfection, turbidity, Giardia lamblia, viruses, Legionella, and heterotrophic bacteria . Final Rule. Fed. Reg., 54:124:27486.

USEPA. 1993. Preventing waterborne disease: a focus on EPA’s research. Office of Research and Development (EPA/640/K–93/001), Washington, DC 20460.

USEPA. 1998a. National primary drinking water regulations: disinfectants and disinfection by-products. Final Rule. Fed. Reg., 63:241:69389.

USEPA. 1998b. National primary drinking water regulations: interim enhanced surface water treatment rule. Fed. Reg., 63:241:69477.

USEPA. 2000. National primary drinking water regulations: groundwater rule. Proposed Rules. Fed. Reg., 65:91:30194.

van der Kooij, D., van Lieverloo, J.H.M., Schellart, J., and Hiemstra, P. 1999. Maintaining quality without a disinfectant residual. J. Am. Water Works Assoc. 91: 55-64.

Vinette Y. 2001. Évolution des trihalométhanes dans divers réseaux de distribution d'eau potable municipaux. Mémoire de maîtrise en Génie Civil, Université Laval.