Assessing biological stability of drinking water without disinfectant residuals in a full-scale water supply system
Frederik Hammes, Cordula Berger, Oliver Ko
̈
ster and Thomas Egli
ABSTRACT
Frederik Hammes
(corresponding author)
Cordula Berger
Thomas Egli
Swiss Federal Institute of Aquatic Science and
Technology (Eawag),
U
̈ berlandstr. 133,
CH-8600, Du
̈
bendorf,
Switzerland
Tel.: +41 44 823 5350;
Fax: +41 44 823 5547
E-mail:
frederik.hammes@eawag.ch
;
www.eawag.ch
Cordula Berger
Oliver Ko
̈
ster
Zu
̈ rich Water Supply (WVZ),
Hardhof 9, P.O. Box 1179,
CH-8021, Zu
̈
rich,
Switzerland
Thomas Egli
Institute of Biogeochemistry and Pollutant
Dynamics, ETH Zu
̈
rich,
CH-8092, Zu
̈
rich,
Switzerland
Biological stability refers to the inability of drinking water to support microbial growth.
This phenomenon was studied in a full-scale drinking water treatment and distribution system of the city of Zu
̈
rich (Switzerland). The system treats lake water with successive ozonation and biological filtration steps and distributes the water without any disinfectant residuals. Chemical and microbiological parameters, notably dissolved organic carbon (DOC), assimilable organic carbon (AOC), heterotrophic plate counts (HPC) and flow-cytometric total cell concentration (TCC), were measured over an 18-month period. We observed a direct correlation between changes in the TCC, DOC and AOC concentrations during treatment; an increase in cell concentration was always associated with a decrease in organic carbon. This pattern was, however, not discerned with the conventional HPC method. The treated water contained on average a TCC of
8.97
£
10
4 cells ml
2
1
, a DOC concentration of 0.78 mg l
2
1 and an AOC concentration of 32 m gl
2
1
,
and these parameters hardly changed in the distribution network, suggesting that the treated water had a high level of biological stability. This study highlights the descriptive value of alternative parameters such as flow-cytometric TCC for drinking water analysis, and pinpoints some of the key aspects regarding biological stability in drinking water without disinfectant residuals.
Key words
|
assimilable organic carbon (AOC), biological stability, drinking water, flow cytometry, total cell concentration (TCC)
INTRODUCTION
Uncontrolled and excessive growth of bacteria in drinking water can lead to a deterioration of the aesthetic water quality, such as the development of undesirable tastes and odours or visual turbidity ( van der Kooij 2000
;
Hammes et al.
2008
). It can also lead to process malfunctioning: for example, the clogging of point-of-use filters, bio-fouling of distribution pipes and bio-corrosion (
Lee
et al.
1980
).
In a worst-case scenario, regrowth can allow the prolifer- ation of pathogenic bacteria (
Vital
et al.
2007
,
2008
), resulting in a hygienic risk to the consumer. One common approach to limit potential regrowth in drinking water is the addition of disinfectants such as chlorine, chlorine dioxide or monochloramine after the treatment train (
LeChevallier
1999
;
van der Kooij 2000
). While this has proven effective- ness, it is also known that some bacteria are resistant to chlorine (
Barbeau
et al.
2005
), that there is a health risk associated with disinfection by-products, and that a negative consumer perception associated with the chlorinous taste exists (
Hambsch 1999
;
Uhl & Schaule
2004
). Some European countries—notably the Netherlands,
Germany, Austria and Switzerland—have taken the approach of distributing high quality drinking water with- out the use of additional residual disinfectants (
Hambsch
1999
;
van der Kooij et al.
1999
). Drinking water treatment in such countries aims to limit microbial regrowth through limitation of the nutrients essential for growth, which is doi: 10.2166/aqua.2010.052
31
Q
IWA Publishing 2010
Journal of Water Supply: Research and Technology—AQUA
|
59.1
|
2010 usually (but not necessarily) organic carbon (
Miettinen
et al.
1997
; van der Kooij 2000
). Under these conditions, the need to understand and accurately monitor the general quality and microbial stability of drinking water has a high priority.
The inability of drinking water to support microbial proliferation is expressed as the biological stability
(intermittently termed ‘microbial stability’, ‘microbiological stability’, ‘biostability’ or ‘regrowth potential’) of the water
(
Rittmann & Snoeyink 1984
;
Miettinen et al.
1997
; van der
Kooij 2000
;
Laurent et al.
2005
). Simply viewed, biological stability is a function of biologically available organic carbon substrate, and ‘instability’ is measured as an increase of biomass and a concomitant decrease of substrate.
However, different interpretations, coupled to preferences for different parameters and methods, are often applied to the concept of biological stability. For example,
Rittmann &
Snoeyink (1984) defined ‘biostability’ as the lack of microbial growth specifically in the absence of disinfectant residuals, while
Srinivasan & Harrington (2007) modelled biological stability taking into account the presence of disinfectant residuals.
Rittmann & Snoeyink (1984) and van der Kooij (2000) regarded both the quality of water, as well as the growth-supporting nature of the materials used for distribution, as important factors when considering biologi- cal stability. In this regard, van der Kooij (2000) highlighted assimilable organic carbon (AOC) and biofilm formation rate (BFR) as the key parameters of biological stability.
Other groups have disregarded BFR and favoured biode- gradable organic carbon (BDOC) as the most important water quality parameter for biological stability (
Escobar &
Randall 2001
;
Laurent et al.
2005
). Moreover, depending on the composition of the water, inorganic nutrients
(e.g. phosphate), rather than organic carbon, can also be growth limiting (
Kerneı
̈ s et al.
1995
;
Miettinen
et al.
1997
).
One of the more contentious aspects of biological stability is which microbiological parameter should be used to describe and monitor this phenomenon. The most used parameter is conventional heterotrophic plate counts
(HPC) (
Hambsch 1999
;
Carter et al.
2000
;
Mu
̈ ller et al.
2003
;
Srinivasan & Harrington 2007
), even though it is well known that the HPC method does not accurately reflect total microbial abundance in drinking water (
Yokomaku
et al.
2000
;
Hoefel
et al.
2003
;
Hammes
et al.
2008
).
Although the concept of biological stability is well known among drinking water microbiologists, little exper- imental and/or field data exist that underpin the main factors involved, and that adequately demonstrate the presence/absence of biological stability. One reason for this is the complexity of the problem: different countries often have completely different water qualities, different treatment technologies and employ different analytical tools for monitoring (
Miettinen
et al.
1997
; van der Kooij 2000
;
Laurent et al.
2005
). Another reason is a shortage of methods specifically describing the two main parameters: namely biologically available substrate (organic carbon) and microbial biomass. With regard to the latter, we have previously shown that flow-cytometric total cell concen- tration (TCC) is an important parameter for drinking water treatment and distribution systems, and that it holds more descriptive value for the treatment process than conventional HPC measurements (
Hammes
et al.
2008
;
Siebel
et al.
2008
).
Here we present a case study of a full-scale drinking water treatment system that treats surface water through successive ozonation and filtration steps and where the treated water is distributed without the addition of disin- fectant residuals. Conventional drinking water parameters
(HPC and DOC) were complemented with AOC and flow-cytometric TCC measurements during an 18-month sampling campaign. This study highlights the descriptive and complementary value of these two additional parameters for monitoring the microbial quality of drin- king water, and it contributes to a better understanding of the fundamental principles of biological stability of drinking water.
MATERIALS AND METHODS
Preparation of AOC-free glassware
Borosilicate glass sampling bottles (250ml) with glass caps were used for sampling, while 20ml borosilicate glass vials were used for the AOC assays. Sterile, carbon-free glassware was prepared by heat-treatment (500
8
C, 6h) as described previously (
Greenberg
et al.
1993
;
Hammes & Egli 2005
).
Teflon-coated caps for the AOC vials were cleaned of
32
F. Hammes et al.
|
Biological stability of drinking water
Journal of Water Supply: Research and Technology—AQUA
|
59.1
|
2010 residual AOC by soaking in warm persulfate (60
8
C, 1h) as described previously (
Greenberg
et al.
1993
).
Layout of the full-scale drinking water treatment plant and sampling sites
The full-scale plant that was monitored in this study produces roughly 50% of the drinking water for the city of
Zu
̈ rich (Switzerland) by treating surface water (Lake
Zu
̈ rich) through sequential ozonation and filtration steps.
The treatment train consists of the following specific steps
(
Figure 1
): (1) pre-ozonation (ozone dose 1.1mgl
2
1
^
25%, hydraulic contact time 50min, ozone residual after contact time c
. 0.15mgl
2
1
); (2) rapid sand filtration (RSF) (double layer filter with 50cm of pumice stone and 80cm of quartz sand; filter flow rate 1.4–4mh
2
1
); (3) intermediate ozon- ation (ozone dose 0.5mgl
2
1
, hydraulic contact time 26min, ozone residual after contact time 0.28mgl
2
1
); (4) granular active carbon (GAC) filtration (double layer filter with
130cm of GAC (Norit ROW 0.8 supra) and 40cm of quartz sand, filter flow rate 4.6–13.2mh
2
1
); (5) slow sand filtration
(SSF) (quartz sand
,
65cm, filter flow rate
0.18–0.52mh
2
1
); and (6) reservoir in the plant. Samples
(250ml) were taken roughly every two weeks over an
18-month period. Samples were taken before and after each step and also of the raw water and of the water coming from the reservoir (Reservoir 1) in the plant after treatment.
Two points in the distribution network (DN) were sampled.
Both points received their water from the same reservoir
(Reservoir 2) in the distribution network which is located
2.4km from the treatment plant (
Figure 1
). Point DN-1 was located a further 1km from this reservoir, and point DN-2 was located 6.1km from the reservoir. Assuming a retention time of one day in the reservoir, the water collected at points
DN-1 and DN-2 have hydraulic retention times in the network of 29h and 43.5h, respectively. All samples were collected in AOC-free glassware and transported in cold- storage containers to the laboratory, where they were processed within four hours of sampling.
Total cell counts with fluorescence staining and flow cytometry (FCM)
Bacterial cells were stained with 10 m lml
2
1
SYBR
w
Green I
(1:100 dilution in dimethylsulfoxide (DMSO); Invitrogen) and incubated in the dark for at least 15min before measurement. Where necessary, samples were diluted in filtered (0.22 m m; Millex w -GP, Millipore) bottled mineral water (Evian, France) just before analysis, so that the con- centration measured with the flow cytometer was always less than 2
£
10
5
cellsml
2
1
. Flow cytometry was per- formed using a PASIII flow cytometer (Partec, Hamburg,
Germany) equipped with a 25mW solid state laser (488nm) and volumetric counting hardware. Green fluorescence was collected in the FL1 channel (520
^
20nm), red fluorescence (also resulting from SYBR w Green I) was collected in the FL3 channel (
.
615nm) and all data were processed with the Flowmax software (Partec). Electronic gating with the software was used to separate positive signals from noise. All samples were collected as logarith- mic (3 decades) signals and were triggered on the green
Figure 1
|
Schematic presentation of the full-scale drinking water treatment plant
(Lengg, Zu
̈
rich, CH) monitored in this study. Water samples were taken during 18 months before and after every treatment step and, additionally, at two points in the distribution network (DN-1 and DN-2).
33
F. Hammes et al.
|
Biological stability of drinking water
Journal of Water Supply: Research and Technology—AQUA
|
59.1
|
2010 fluorescence channel (FL1). The standard instrument error on the FCM measurements was experimentally determined to be below 5%.
Conventional water parameters
The heterotrophic plate count (HPC) method was performed according to the Swiss guidelines for drinking water (
SLMB 2000
). In short: 1ml of the water sample was transferred to a sterile Petri dish and mixed with about
15ml plate count agar (PCA, Oxoid). The agar was kept at
46
8
C before plating. The samples were incubated at 30
8
C for 3 days. Phytoplankton in the lake water was measured as described in
Mu
̈ ller et al.
(2003)
. Dissolved organic carbon (DOC) (after 0.45 m m filtration) and particulate organic carbon (POC) (concentrated on a fibreglass filter
(Whatman GF/F)) were measured on a Dimatoc 2000
TOC analyser equipped with a Dima 1000 Universal detector. Water temperature was measured on site during sampling. Assimilable organic carbon (AOC)
AOC was determined with a batch growth assay as described previously (
Hammes & Egli 2005
;
Vital et al.
2007
). In short: the pasteurised and filtered water samples
(15ml) were inoculated with 10 m l(1
£
10
4
cellsml
2
1 initial concentration in the assay) of a bacterial AOC test- inoculum. These suspensions were then incubated at 30
8
C for three days (until stationary phase was reached) and the resulting growth was measured with flow cytometry (see above). The AOC test-inoculum comprised autochthonous bacteria from the treatment plant that was studied, and was prepared as described previously (
Vital
et al.
2007
).
The same bacterial community was used for all AOC determinations throughout the present study. As standard quality control prior to use, the performance of this inoculum was compared with bacterial AOC test-inocula used in previous studies in our group (
Hammes
et al.
2006
;
Vital
et al.
2007
), using different types of natural surface water as media. A difference of less than 10% in the average
AOC values was deemed acceptable for use. AOC ( m gl
2
1
)is
estimated from cell concentrations (cellsml
2
1
) using a theoretical conversion factor (
Hammes
et al.
2006
)
(Equation 1). All assays were performed in triplicate.
The detection limit of the method was 10 m gl
2
1 and the average standard deviation was
^
10%.
AOC
ð m gl
2
1
Þ¼
netgrowncells ð cellsl
2
1
Þ
conversionfactor ð 1
£
10
7
cells m g
2
1
Þ
ð
1
Þ
Data presentation and calculations
Given the large set of data collected over an 18-month period, we have opted to present the data as box plots to illustrate the spread of the data, and we have used the geometrical mean values of all data for the calculations.
RESULTS AND DISCUSSION
Water supply without the addition of disinfection residuals The configuration of the Zu
̈
rich drinking water treatment plant targets specifically the production of biologically stable, high quality drinking water that can be distributed without the need for additional disinfectants (
Figure 1
).
The raw water (Lake Zu
̈
rich) has a low organic carbon content (POC
¼
0.22 (
^
0.1) mgl
2
1
; DOC
¼
1.3 (
^
0.1) mgl 2
1
; AOC
¼
0.023 (
^
0.017) mgl
2
1
), which means that relatively low ozone dosages are required during treatment.
Two ozonation steps are meant to serve as double disinfective barriers against any malignant microorganisms that may enter the system through the raw water, while also oxidising possible micropollutants in the water (
Mu
̈ ller et al.
2003
;
Von Gunten 2003
). The ozonation also transforms stable, dissolved natural organic matter (NOM) molecules and organic particles (e.g. phytoplankton) into typical AOC molecules (
Von Gunten 2003
;
Hammes et al.
2006
,
2007
).
Three separate biological filtration processes (rapid sand filtration, granular activated carbon filtration and slow sand filtration) are the basis for the removal of organic carbon from the water. We have demonstrated previously that these filters are biologically active with high AOC removal capacity (
Hammes
et al.
2006
;
Velten
et al.
2007
). As a result, regrowth of microorganisms occurs in the biological filters together with the removal of biologically available
34
F. Hammes et al.
|
Biological stability of drinking water
Journal of Water Supply: Research and Technology—AQUA
|
59.1
|
2010