(58d) Optimal Design of a Real-Time Monitoring System for Water Distribution Networks

Concerns about accidental or intentional contamination in drinking water distribution networks have lead to increased interest in the design of real-time systems to help prevent and/or minimize the effect of a contamination event. One proposed method of protection is the installation of an early warning detection system. Sensors installed at various locations throughout the drinking water network could provide warning in the event of a contamination. However, an early warning detection system provides only a coarse measure of the time and location of the contamination event. In order to determine the time and location of the contamination event a source inversion problem must be solved using data from the monitoring system. Due to the cost of installing and maintaining a significant fixed sensor grid within a water distribution system, the number of sensors used must be limited. Several researchers (Osfeld A. and Salomons E. 2004, Berry, J., et al. 2005) have investigated the problem of optimal placement of the fixed sensors within drinking water networks. For source inversion to be effective, measurements from the fixed sensor grid must be combined with information obtained from manual grab samples taken during the event. Here, we present an MILP formulation that determines the optional placement of fixed sensors for improved effectiveness in source inversion. This approach is further extended to determine optimal locations for taking manual grab samples in a real-time context. Results are presented on large-scale networks with over 10,000 nodes.