Variability of Tap Water Residual Chlorine and Microbial Counts at Spatially Resolved Points of Use

Abstract

Balancing the risk between microbial and disinfection by-product exposures in potable water through optimization of disinfectant (chlorine) dosing often proves a difficult task in aging urban drinking-water distribution systems (UDWDS). We hypothesized that geocoded pipe characteristics could improve our understanding of chlorine consumption in a city of a geographic region (Eastern Mediterranean) that has been poorly studied. The main objectives of this study were to (1) determine the UDWDS characteristics (household distance from chlorination tank and pipe leaks) that could explain the spatial variability of residual free chlorine, and (2) investigate the association between water microbial counts and residual chlorine in two areas with contrasting pipe network characteristics (low-risk and high-risk; LRA and HRA). We collected water samples from 150 households in two areas of Nicosia, Cyprus, during winter (air temperatures 0-15°C). Household addresses were linked with geocoded pipe network data, on-site residual chlorine levels, trihalomethanes (THM), and microbiological measurements. No significant difference (p > 0.05) in the average free residual chlorine levels was observed between the two areas, despite their documented differences in water age. However, the mean THM concentration in HRA (76 lg/L) was significantly (p < 0.001) higher than that in LRA (56 lg/L). Each household's distance from the chlorination tank could explain up to 29% of the variability in free residual chlorine levels for LRA, after adjusting for confounders, such as, total bacterial count (TBC) 22°C/37°C, THM, and pipe leaking incidences; this strength of association did not hold for HRA. Pooling data from both areas, a significant (p < 0.05) multivariate model of residual chlorine was constructed, with TBC 37°C being the only significant predictor. Water boards with historically low water microbial counts could apply area-specific residual chlorine monitoring and management schemes within their UDWDS to ensure minimization of total THM formation.

Spatial and seasonal variability of tap water disinfection by-products within distribution pipe networks

“…Household distances to their respective water chlorination tank varied widely in the two areas with a median distance of 5.8 km in area 1 versus 3.5 km in area 2, corroborating the estimates of water age (Pieri et al, 2014). The distance between the studied areas was not large as in other field studies, but allowed for a higher resolution of manifested trends.…”

“…The distance between the studied areas was not large as in other field studies, but allowed for a higher resolution of manifested trends. Chlorine demand in our study was high, as indicated by the absence of any residual free chlorine in all sampling points of both areas in summer, while a mean value of~0.35 mg L −1 was recorded on-site for both areas during winter time (Pieri et al, 2014). Geocoding of participating households allowed for estimating distances between sampling points (kitchen faucet) and key locations, i.e., the chlorination tank, indicated by length of main pipe, that could be used as a surrogate of water residence time (Fig.…”

“…A portion of the withinhousehold variability was attributed to diurnal variation (Singer, 2001), but this aspect was not evaluated here. Experiments with tap water flushing time up to 15 min did not show a statistically significant increase in THM levels (Pieri et al, 2014), suggesting that the sampled water was not from within premise plumbing, but it was coming from the service line. Looking into each THM analyte separately, TBM showed the smallest variance between households of the two areas.…”

“…Three households in the area 2 withdrew participation after summer sampling, hence they did not provide us with a water sample during winter collection. Residential addresses were geocoded on GIS maps with a portable geographic position system unit and then linked to digitized boundary maps supplied by the WBN as earlier described by Pieri et al (2014). Water pipe network characteristics relying upon a census-based street classification system were overlaid by the point shape files of geocoded households.…”

“…Sampling was conducted during (i) summer with high air temperatures (average values ranged between 30-45°C) (July to early September 2012) and (ii) winter with low air temperatures (average values ranged between 0-15°C) (January to March 2013) (details can also be found in Pieri et al (2014)). During the house visit conducted by trained personnel, a tap water sample was obtained from the primary faucet in the household after flushing the tap for 1 min.…”