|
Naturally Occurring Autofluorescent Particles as Surrogate Indicator of Sub- Optimal Pathogen Removal in Drinking Water Download MS Word article with tables and figures:  Gothenborg Water Work - study on Cryptosproridium(MS Word 139 KB)
By 0. Bergstedt* and H. RydbergAbstract 
Sub-optimal pathogen reduction in conventional water treatment plants (WTPs ) could pose a danger to public health, but monitoring of drinking water for pathogens such as Cryptosporidium with as low infection doses is not feasible. Different surrogates have been studied previously and several of those that have been seeded to the water during well-controlled conditions have shown removals similar to those of Cryptosporidium. As full-scale operations can have a large variability in treatment performance, seeding is normally not advisable in full scale, and turbidity and particle counts are influenced by chemical precipitation itself. Naturally- occurring surrogates such as spores and algae have showed promising results, but require large sampling volumes and/or time consuming laboratory methods. Our aim was to explore the potential of naturally occurring surrogates using the autofluorescence (FL) of parasite-sized algae. A bench scale study showed that these FL particles were destabilized by chemical precipitation in a way similar to Cryptosporidium oocysts. They were also quite resistant to the chemicals added in full-scale treatment. Historical operational data contain many variables making evaluation of optimal conditions difficult, but factorial full-scale experiments with small changes in a couple of key variables might be useful for optimization. 192 Introduction There have been several waterborne, disease-causing outbreaks reported in Sweden, with hundreds or thousands of people infected (Hult, 1998), in spite of the abundance of water from relatively pristine watersheds. Protozoan parasites, Giardia and Cryptosporidium, were among the causing agents that were identified, but in most cases the agent remains unknown. A microbiological risk assessment (Westrell et al., 2001) showed that the multi-barrier approach of chemical precipitation and chlorination might be insufficient for pathogens that are insensitive to chlorination, e.g. Cryptosporidium. Dependence on particle removal as the only functional barrier makes the treatment even more sensitive to sub-optimal conditions. Studies have pointed out the variability of particle removal performance as a possibility of optimization (Hijnen, 2000). A study of 19 Swedish chemical precipitation WTPs (Hernebring 1980) showed that the wrong precipitation pH and too low an initial energy input in flocculation had a major impact on treatment performance. The sub-optimal conditions led to poor settling and, in spite of low surface loads in sedimentation, the filters had to take most of the load. Obvious problems were high alum contents, increased cost of filter operation, and possible interference with disinfection. Low temperature and increased surface loads worsened the problems and over-dosages of alum were not enough to compensate. The worst examples had turbidity removals from raw water to drinking water ranging from 0.2 to 0.4 log. Seeding studies in both pilot and full scale (Nieminski and Ongerth, 1995) showed that turbidity removal can work as a rough estimate of oocyst removal and that 4 -7 [-till particles could be quite a good surrogate, while heterotrophic plate count removal was an ineffective one.
The impact of sub-optimal coagulation for the removal of Cryptosporidium and surrogates by filtration was studied in pilot scale (Huck et al., 2000). Two different pilot plants with pre-chlorination and coagulation with alum and coagulation aids were investigated. Filters were seeded with jar-coagulated suspensions offormalin-inactivated oocysts and non-inactivated Bacillus subtilis spores. Experiments were performed without coagulation, with optimized coagulation and with sub-optimal coagulation. Removal of spores and 2 [-till particles was less than oocyst removal, but under sub-optimal conditions the decreases in removal were similar.
Removal of sulphite-reducing clostridia spores has been suggested as a simple and cheap potential sunogate for protozoan (oo)cyst-reilloval (Hijnen et al., 2000). Removal was studied in 8 full scale WTPs during two summer and two winter weeks. Samples of up to 500 L each were collected and analysed by standard microbiological methods. Removals were found to be similar to actual (oo)cYst- removals in other studies. 193 Algae have also been studied as sunogates for removal of Cryptosporidium oocysts (Akiba et al., 2001). Algae found in raw water sources such as lakes and rivers are known to have physical properties, such as zeta potential, that are similar to those of oocysts. In laboratory experiments comparing coagulation and filtration characteristics of algae and oocysts, the authors found algal removal to be similar to oocyst removal. Our aim was to explore the potential of even simpler naturally occuning sunogates and their usefulness as indicators of sub-optimal treatment of low turbidity raw waters. Materials and Methods Bench-scale flocculation Bench-scale flocculation trials were made on three different surface waters that supply the Gijteborg area (Table 1). Raw waters were seeded with labeled C. parvum oocysts and flocculated under optimal and sub-optimal conditions with regard to flocculant dose. Flow cytometry and turbidity were used to monitor removal of parasites and sunogates. General setup. A Flocculator 2000 system (Kemira AB), consisting of 2 x 6 cylindricall L glass beakers with microprocessor controlled stiners, was used to conduct jar tests. Raw water pH was adjusted with 0.1M NaOH prior to addition of aluminium sulphate (AI2(SO4)3.H2O). Both solutions were prefiltered. Flocculator settings (time, speed and G-value) were chosen to give precipitation results similar to those in the full-scale process: alum mixing phase (0.3 min, 400 rpm, 292 S-I) followed by two flocculation steps (20 min, 46 rpm, 11 S-I and 20 min, 28 rpm, 5 S-I) and sedimentation (45 min). Experiments were made at approximately lo°C and all jars were kept in flow-through water baths, to prevent arise in temperature. Optimal doses of aluminium sulphate (mg/L) were calculated for each raw water as (colour Pt mg/L .0.7) + 15. Adjustments of flocculation pH were made individually to give a final pH of 6.5 independent of raw water and alum dose. Table 1. Raw water quality data for Feb. 2001 to Ian. 2002 194 Cryptosporidium. Preparations (1.250108/ml) of formalinized (5% formalin with 0.01 % Tween 20) C,parvum (Waterborne Inc) were used to seed raw waters. Oocysts were labeled with Cy-5 (Nycomed Amersham, Ltd) conjugated antibodies (Crypt-a-glo, Waterborne Inc), in solution and incubated for 40 minutes at 35 °C. Miniaturized control experiments (30 ml raw water) were made to check fading/ dissociation, due to light exposure and precipitation chemicals used. The potential of non-specific binding of free antibodies to existing raw water particles was evaluated. This also included the risk of microfloc formation, incorporating enough free antibodies to generate new fluorescent particles. Control experiment samples were analysed by flow cytometry and visually verified by fluorescence microscopy (Carl Zeiss Axiovert SIOO; filter set GLF Visual Cy-5, Chroma Tech). Experimental. The three raw waters were examined individually in duplicate trials. Each trial consisted of three jars seeded with C. parvum oocysts and three control jars. The alum doses used were optimal dose, 40 % less than optimal and 60 % less than optimal. A control experiment was also made to evaluate the effect of formalin/Tween 20 on precipitation results. Parasite addition ( -30107 oocysts/ 1 raw water) and pH adjustment was made prior to the mixing phase. Alum (10 g/l) was then added during rapid mixing.
Sampling/analyses. Temperature, pH and turbidity were analysed for raw water samples, before each trial, and after 45 min sedimentation. Flow cytometry samples ( Iml in particle free microcentrifuge tubes) were taken throughout the experiments: from the raw water, after parasite addition, during alum mixing, and during and after sedimentation (15,30 and 45 minutes). At the end of each trial the floc was resuspended by rapid mixing and analysed by flow cytometry.
C. parvum oocysts were analysed directly by flow cytometry (Microcyte, Optoflow AS). The instrument, using a 635 nm diode laser, measured forward scatter (particle size -0.4-20 f!m) and particle-associated fluorescence (650-900 nm). Water samples were also analysed for two surrogate parameters: total particles I -15 f!rn/ml, defined by I and 15 f!m polystyrene standards, and autofluorescent particles 1-15 f!rn/ml (FL). Particles showing autofluorescence were mainly algae, probably due to 635 nm excitation of photosynthetic pigments.
Calculations. Log-removals of Cryptosporidium and surrogates were calculated as -log (C/Co)' where C = concentration at the end of sedimentation (45 min) and Co = raw water concentration. For Cryptosporidium, initial concentration was calculated from the alum mixing phase, and final concentration was not compensated for with the controls. Thus, removal was conservatively expressed as minimum reduction. Relationship between variables was tested by rank correlation (Spearmans rs ). 195 Full scale Only 12 of the more than 2000 WTPs in Sweden regularly monitor raw water for Cryptosporidium and Giardia (Bergstedt and Andersson, 2001). One of those, the Lackareb1ick WTP was chosen for a full-scale study. This WTP treats low turbidity raw water (Lake Delsjon, Table 1) using the most common process in Sweden, coagulation with alum, sedimentation and rapid filters. Process monitoring. Throughout 2001 flow cytometry samples were analysed for surrogate parameters (total particleslautofluorescent particles 0.4-15 J.lm). Detailed sampling profiles, through the entire process, were also checked at periods with and without pre-chlorination, 0.3 glm.3. Pre-rinsed centrifuge tubes (FALCON 50 ml) were used for all samples. A control experiment was set up to test the influence of treatment chemicals (chlorine, CaCO3 and alum) on autofluorescent particles.
Factors that are supposed to influence the particle removal efficiency of chemical precipitation and the response in on-line particle counts of particles > 1 [.lm were checked for the full 12-month period and for 3-month periods. The mean values before and after changes in the process were also compared using nearby periods as external reference distributions.
Experimental. A simple 22-factorial design for the variables coagulation dosage and pH was used in a preliminary experiment. Changes in random order were made at a set time on four consecutive days with sampling on two occasions per day. Particle removals from raw water to filtered water were used as response factors. Other factors such as surface loads were kept as constant as possible by avoiding changes in raw water flow and the regular cleaning of sedimentation tanks. Samples were taken in the same part of the filter cycle, after ripening well before expected breakthrough, since backwash occurred on a set time each day. The changes were limited to +1- Iglm3 of aluminium sulphate and +1- 0.1 pH-units compared to the current target values. Results Bench'"scale flocculation The general flocculation setup gave consistent results. The temperature change throughout the experiments was less than 1°C and final pH varied between 6.3 -6.7. Minimum reductions of Cryptosporidium were very similar for each raw water, at optimal and 40 % less than optimal dose. Removals typically ranged between 1-3 log. Reducing the dose further (60 % less than optimal) could generate flocculation, but on most occasions it did not. In both cases Cryptosporidium 196 removal was close to zero. The strongest association was seen between removal of Cryptosporidium and autofluorescent particles 1-15 [!ill (Figure I) . Both experimental and control jars were in good agreement concerning turbidity after sedimentation, and formalin/Tween 20 had no general effect on precipitation results in the control experiment. Miniaturized control experiments showed no fading/dissociation of labeled Cryptosporidium during 5 h exposure to low light and the chemicals used. Too high a concentration of free antibodies initially generated new fluorescent particles, mainly in the < 1 [!ill region, as soon as alum was added. Thus, the amount of free antibodies was decreased in the jar tests. Fullscale Process monitoring, Removal of surrogates differed significantly over the year (Figure 2). Removal of particles 0.4-1[!ill and 1-15 [!ill was 1.5 and 1.6 log, respectively. In the summer, when pre-chlorination was initiated, removal of autofluorescent particles 1-15 [!ill ranged between 2.2 and 3.8 log compared to 1.2 and 3.0 log in periods with no pre-chlorination. 197 Full-scale variations. Even though there were substantial variations in the key parameters (Table 2), no strong relationship between any of the parameters and the total number of particles > 1 !-lm according to the on-line counter was found for either the 12-month or the 3-month period. The mean values before and after changes in the process were significantly different only between pre-chlorination and no pre-chlorination. Pre-chlorination typically reduced turbidity and on-line particle counts after filtration. Table 2 Factorial experiment. Even though the changes were very small, the responses were quite clear. The worst particle removal was achieved when coagulation pH was ncreased and the alum dosage was decreased (table 3). The turbidity of filtered water varied between 0.02 and 0.07 FNU (Formazin Nephelometric Units). Log-removals were generally lowest for turbidity and highest for fluorescent particles 1-15 !-lm. The difference between the two daily samples was smaller for 0.4- 15 FL than for the other parameters. A high particle reduction after sedimentation was not connected to a high total reduction, especially not for turbidity. 198 Table 3 Particle profiles. Detailed sampling profiles in the course of the treatment process revealed a significant increase in turbidity from 1 FNU in raw water to 5 FNU in the alum mixer. This was due to a large increase in particle content mainly in the 1 !-tm region (Figure 3). Autofluorescent particles 0.4 -15 !-tm were not sensitive to the addition of treatment chemicals. In a 9 h control experiment, when autofluorescent particles 0.4-15 !-tm were exposed to chlorine, chlorine + CaCO3 and chlorine + CaCO3 + alum, respectively, the mode of fluorescence intensity decreased -13.5 % from approximately 14800 to 12800 molecular equivalents of allophycocyanine. This did not cause a significant loss of autofluorescent particle content due to detector sensitivity. Fig 3 199 Discussion The bench-scale results confirmed the work of previous investigators that turbidity under estimates oocyst removal and that particles of similar size may be a better surrogate. The use of Cryptosporidium-sized autofluorescent particles looks promising as a tool for optimization. The particles can be easily sampled/ analysed and are naturally occurring tracers of raw water origin. Any association between removal of Cryptosporidium and surrogates should be made carefully. In the jar tests oocyst concentration did not reflect occurrence in natural surface waters. However, sufficient flocculation was required for oocysts to settle. Under suboptimal conditions (60% less than the optimal dose) Cryptosporidium did not settle although flocs were formed. Jar tests were made with inactivated oocysts, for practical reasons, which may have influenced flocculation properties. The use of formalinized Cryptosporidium was investigated by Emelko (2001). An apparent effect of scale was the loss of seeded oocysts, probably due to binding on walls and stirrers, resulting in a lower initial concentration than expected. Despite this, reduction of Cryptosporidium seemed to model removal, since resuspension of floc recovered the initial concentration. Criteria for making a rank correlation may not be fulfilled completely. Removal of autofluorescent particles should be used as an operational optimization tool rather than for estimating true Cryptosporidium removals. However, further work on filtering is needed. More than a year of analyses on raw water and filtered water from full scale operation showed that there were enough fluorescent particles to avoid repeated zero counts that would make evaluation of removals less reliable. Throughout the year at least 3 log removal could be monitored and excitation at 635 nill gave very low levels of background fluorescence due to non-algal material. Profiles with particle counts throughout the treatment process showed that numbers of autofluorescent (FL) particles were quite constant during pH changes, low dosage pre-chlorination, lime addition, alum dosage, coagulation and flocculation. Another advantage was the insensitivity to air bubble formation, which can be a problem with conventional particle counters. This is promising since studies can be made independent of the changes in turbidity and normal particle counts created by the process itself. It can be especially useful when comparisons are made between different parts of the process. Even though autofluorescent particles seem to withstand disinfectants at moderate doses, annual variation in algal composition may well affect sensitivity. At the moment, the most stable count includes both FL particles smaller than 1 [till and larger FL particles. This means that even if FL particles are a good surrogate for pathogen removal, they are far from specific to Cryptosporidium-sized particles, since smaller particles may be removed in part by different mechanisms. Further work will include pre-treatment of samples to distinguish between the algae that were of oocyst size from the beginning and smaller algae that have grown in the course of treatment. 200 Conclusion Naturally occurring autofluorescent particles may be surrogate indicators of sub-optimal pathogen removal and an optimization tool in drinking water treatment. References 
|
||||||
|
©2004 BioDETECT AS - Contact Webmaster - Design: Silicom Web Design - Please View in MS IExplorer 6, Opera, or Netscape 6 | |||||||
applications
sales
FAQs
about biodetect