Nodal Importance Concept for Computational Efficiency in Optimal Sensor Placement in Water Distribution Systems

Computational expense associated with documented methods for the optimal placement of contaminant sensors to provide maximum protection for a water distribution system (WDS) under a terrorist attack has prevented the emergence of a definitive method for sensor placement. We submit a sensor placement method that assigns "nodal importance values" in order for an optimization algorithm to efficiently maximize WDS protection as a function of detection time, contaminated water volume, and detection likelihood while obeying sensor availability constraints, accounting for terrorist attack uncertainties, and minimizing computational expense. The importance of a node is a function of contaminant presence, potential contaminated water volume, and time elapsed after a contamination event. More-important nodes are assigned to a subset, and an optimization algorithm (a simple genetic algorithm in our case) searches the subset to find the sensor node combination of best protection performance. Results from testing our method on a WDS of complex behavior indicate that this method yields desirable protection performance in a very computationally efficient manner and that protection performance resulting from this method is sensitive to the size of the subset chosen.