CONFIRMATION OF STATISTICAL ASSOCIATIONS OF EXTERNAL PARAMETERS WITH
COLIFORM FAILURES IN DISTRIBUTION SYSTEMS USING DATA OBTAlNED FOR 1992
Report No FR0420
The customer will benefit from using fluctuations in external parameters to predict an increased probability of coliforms in the distribution.
Identifying associations of the presence of coliforms with external parameters may provide information on the root causes and the origin such that steps may be taken to eliminate the problem.
To confirm (using distribution supply data collected for 1992) the associations between intermittent coliform failures and external factors identified using data previously supplied by certain water companies for 1990/91.
Coliform failures in samples of water collected from distribution systems may occur for a variety of reasons such as problems at the treatment works, infiltration, release from biofilms and sediments and regrowth. Statistical studies performed previously for this contract have identified associations between the presence of coliform organisms in the distribution and fluctuations in certain external physical, chemical and biological parameters. In particular, the presence of coliforms was associated with higher temperatures, higher plate counts and lower total chlorine concentrations. These associations, performed with data supplied by certain water companies for the years 1990 and/or 1991, suggested that coliform failures reflected real changes in the local environment within the distribution system and were not merely due to the chance 'capture' of a coliform from a homogeneously distributed population during sampling. Identifying such associations may facilitate implementation of operational processes to minimise coliform fail ures. It is therefore important to confirm these associations using data collected in 1992.
Results are presented from analyses for associations of fluctuations in external parameters with the presence of coliforms using 1992 statutory sampling data for Water Companies A, C and D and January 1990 to March 1993 data for Water Company B. Also, results from analyses of 1991 data for additional zones in Water Company A and from 1990/1991 data for all zones with failures in Water Company H are presented.
The associations observed in the previous report (FR 0361) are generally confirmed. Thus, in Water Company A statistically significant associations of higher temperatures with the presence of coliforms were found. Mean temperatures for samples with coliforms were higher than for samples without coliforms, for all water companies (except Water Company H, 1991), although these differences were only statistically significant for operational hydrant and fixed tap samples from Water Company A. Strong associations of higher plate count concentrations (one, two, three and seven day) with the presence of coliforms were also confirmed. There appears to be evidence with data from Water Companies A, B, C and D that associations with coliforms were stronger for three day plate counts than for one or two day plate counts. Similarly for Water Company H in 1991 a statistically significant association between seven day plate counts, but not one day plate counts, and the presence of coliforms was detected. Statistically significant associations between lower total chlorine concentrations and the presence of coliforms were confirmed for random statutory tap data from Water Companies A and D. Mean total chlorine concentrations were lower for samples with coliforms compared to those without for Water Companies B, C and H although differences were not significant. In no cases were statistically significant associations between higher total chlorine concentrations and the presence of coliforms found. Associations between lower nitrate ion concentrations and the presence of coliforms in fixed statutory taps (1990 and 1992) were found for Water Company A. In the same fixed tap samples for 1992 a statistically significant association between higher nitrite ion concentrations and the presence of coliforms was found.
Higher aluminium ion concentrations were associated with the presence of coliforms in data from Water Company B (1990 - 1993). Similarly for Water Company C (1992), an association of higher iron (III) concentrations with the presence of coliforms was detected. In general associations of the presence of coliforms with fluctuations in pH, turbidity, colour and conductivity were not observed, although there was one exception for turbidity (random taps from all zones with failures in Water Company C, 1992) and conductivity (fixed tap samples from 11 zones in Water Company A, 1990). THM data were not sufficient for powerful statistical analyses.
The differences found previously (FR0361) between operational hydrants and consumers' taps (both random and fixed) for Water Company A appear to have been unique to the data sets from the 11 zones in 1991. For those zones in 1991, statistically significant associations between the presence of coliforms and higher three day plate counts and lower total chlorine concentrations were found only for operational hydrant samples and not in random or fixed statutory samples. Here, however, analyses of the additional data, show that associations of the presence of coliforms with higher three plate counts occurred in random and fixed statutory samples as well as in operational hydrants. Furthermore, statistically significant associations of the presence of coliforms with lower total chlorine concentrations and higher two day plate count concentrations appear in the random statutory tap samples but not in the operational hydrant or fixed statutory samples. It is also shown that associations between the presence of coliforms and two day plate counts were relatively weaker compared to three day plate counts in operational hydrant samples than in random statutory tap samples. Thus, differences between operational and consumer taps do exist in Water Company A.
Recommendations on possible operational implementations identified from the associations of coliform failures with fluctuations in external parameters will be made in the final report.
Further work is to be performed. Data have been obtained from water supply distributions in Scotland and are currently being processed. In addition, a water company not represented on the steering group is preparing data from statutory monitoring of random taps during 1990 - 1992. Water Companies G and H are also compiling 1992 data for analysis. Data have been obtained from Water Companies E and F. Considerable data for water samples with coliforms are now available in computer format. Linear regression analyses will be performed and it is intended to use a multivariate approach.
VI RESUME OF CONTENTS
This Progress Report details statistical analyses performed between April and September 1993 to confirm associations, found previously in this contract (FR 036 1), between the presence of coliforms and fluctuations in external parameters within the distribution supply.
Section 2 describes the graphical presentation and statistical treatment of data, discussing the use of probability plots in determining the most appropriate and best estimates for the statistical parameters, namely the mean and standard deviation. In S ection 3, results of analyses are presented for a variety of parameters including temperature, pH, conductivity, plate counts (one, two, three and seven), concentrations of metal ions, total chlorine, nitrate and nitrite, turbidity and colour. Statutory sampling data for 1992 from Water Companies C and D and for January 1990 to March 1993 for zones in Water Company B are analysed to confirm associations identified in the previous report (FR 0361). In addition, data from other zones for 1991 in Water Company A and all zones with failures in Water Company H (1990/1991) are analysed. The conclusions are presented in Section 4.
Copies of the Report are available from FWR, price £25.00 less 20% to FWR Members