Full Description

Among the many factors contributing to ever-increasing quantities of distribution system water quality data being collected by U.S. water utilities is a growing regulatory emphasis on distribution system water quality in general. Regardless of the system-specific reasons, whether voluntary or compliance-related, few would argue with the perception that utilities will collect more distribution system water quality data in the future than in years past. The common practice of plotting trends and applying simple statistics to distribution system parameters, such as heterotrophic plate counts (HPCs), is not particularly useful for decision-making at water systems that deal with large quantities of data collected from numerous sites under dynamic environmental conditions. More complex statistical tools offer insights and provide an analytical framework that can simplify decision-making. A one-year-long monitoring program was undertaken to evaluate HPCs and other select parameters biweekly at 26 sites selected outside of the system's bacteriological compliance sites. The 26 special sites were in presumed low-flow areas or near pressure zone boundaries, or were otherwise expected to represent sub-optimal distribution system water quality conditions. A statistical approach was used to examine the overall value of the special monitoring program. For example, statistics were used to discern whether the data from the HPC values at special sites were different from similar data collected at compliance sites. Several other data relationships were also probed. The nature of HPC data makes it particularly subject to analytical results that are difficult to handle statistically. For example, a typical laboratory result is TNTC (too numerous to count). Such a result, if ignored, or if set to zero for purposes of reporting, would improperly bias the evaluation. In addition, HPC data, like many environmental parameters, may not be normally distributed, which precludes the use of the many common statistical tests. This case study applied statistical tools that allowed for non-normal data distributions, and developed techniques for appropriately analyzing censored HPC data. Use of databases also enabled consideration of vast quantities of historical data. By using these and other techniques, most of the historical data could be retained, allowing the decisions to be made on the basis of a much larger data set. Although this project focused on just a few parameters, the data management procedures and statistical techniques used in this case study can be applied to most types of distribution system water quality data to make practical operational decisions. This paper includes specific steps for maximizing the statistical value of data collected for compliance purposes. It also provides guidance for designing and implementing statistically valid voluntary monitoring programs. Finally, this paper provides advice on handling non-normal or censored data and guidelines for avoiding common errors in statistical analyses. Includes 11 references, tables, figures.

Product Details

Edition: Vol. - No. Published: 11/01/2002 Number of Pages: 20File Size: 1 file , 1.3 MB