| Despite the production of high quality drinking water at the treatment plant, many opportunities exist in the distribution system for water quality to deteriorate. The quality of water in the distribution system must be assessed by extracting samples of the drinking water from various locations in the system. The research presented here examines sampling for water quality in distribution systems from several perspectives. First, a summary of case studies is presented to illustrate the variable nature of disinfection by-product occurrence in distribution systems and to highlight the need for intensive sampling. Comprehensive sampling in distribution systems is then addressed through the development of a randomized, stratified sampling methodology. The sampling methodology presents a set of tools for water utilities to design a sampling program that addresses a specific sampling goal. Through the use of stratification, water quality estimates can be compared between sub-groups of the system. Furthermore, the statistical framework of the methodology helps to eliminate bias in sampling results and provides a level of confidence in sample data that is lacking under current ad-hoc sampling protocols. Finally, several important components of sampling design are examined, based on the application of the sampling methodology to a case study water utility. The number of samples and time of sample collection were found to affect the estimates of water quality made from sample data. |
