Water distribution systems sensors are becoming increasingly more efficient and effective at discovering water quality changes in water distribution systems. We are attempting to develop an efficient and effective method to determine the source of the water quality change. We use publicly available software (EPANET) coupled with a particle backtracking model (BTX) and a conditioning method to probabilistically locate the contamination source. In prior work, we have shown that this method is effective for both steady-state and transient flow conditions in a simple water distribution system (i.e., containing one reservoir, one pump, and one tank). In this work, we demonstrate the effectiveness of this approach in more complex water distribution systems (i.e. containing multiple tanks, reservoirs, valves, and/or pumps) with more realistic sensor conditions (e.g., "fuzzy sensors", non-detect results, sampling uncertainty).
展开▼
