The protection against contamination events in water distribution systems involves two distinct phases: detection of the presence of a contaminant and implementation of actions to isolate and/or expel it rapidly. The problem of detection is confronted by installing a series of monitoring stations, strategically placed across the distribution system and consisting of sensors to detect the presence of contaminants. The actions to be implemented may include operations on distribution system devices (valves and hydrants), injection of reagents to eliminate the contaminant or simply alert users. The procedure proposed here attempts to address the problems related to the two phases by means of two consecutive optimisation processes, both of them performed off-line and assuming a specific 24-hour water demand sequence in each network node, whereas the accidental/intentional injection of contaminant can occur in any node and at any hour of the day. With reference to this vast range of possible injection scenarios, the first multi-objective optimisation process defines the position of a pre-selected number ns of sensors across the distribution system in order to minimize the expected percentage of undetected contamination events and the expected volume of contaminated water consumed up to the beginning of the response operations following detection. A single configuration of stations is then selected from the Pareto front produced by this optimisation process (“knee point” of the Pareto front). At the end of this first optimisation process and with reference to the selected set of sensors, a potentially contaminated area in the network is associated to each sensor for each sub-period of the day. The second multi-objective optimisation process is then aimed to identify, with reference to each station and sub-period, and thus inside the corresponding potentially contaminated area, the hydrant opening and valve closing operations to be carried out in order to minimize both the number of operations and the expected volume of contaminated water consumed between the beginning of the response operations and the disappearance of the contaminant, assuming the availability of an unlimited number of response teams. Once these devices have been identified (“knee point” of the Pareto front relevant to the second optimisation process), an a posteriori analysis is performed to determine the sequence in which they should be activated based on the number of response teams actually available. In these optimisation processes, a hydraulic and quality simulator (EPANET) is linked to a multi-objective genetic algorithm (NSGA-II) in order to compute the value of the objective functions of the problem across different contamination scenarios. The results obtained applying the procedure to a real and complex water distribution system have shown it to be a robust and effective method for reducing the impact on the population.

A multi-objective approach for detecting and responding to accidental and intentional contamination events in water distribution systems

GUIDORZI, Marco;FRANCHINI, Marco;ALVISI, Stefano
2009

Abstract

The protection against contamination events in water distribution systems involves two distinct phases: detection of the presence of a contaminant and implementation of actions to isolate and/or expel it rapidly. The problem of detection is confronted by installing a series of monitoring stations, strategically placed across the distribution system and consisting of sensors to detect the presence of contaminants. The actions to be implemented may include operations on distribution system devices (valves and hydrants), injection of reagents to eliminate the contaminant or simply alert users. The procedure proposed here attempts to address the problems related to the two phases by means of two consecutive optimisation processes, both of them performed off-line and assuming a specific 24-hour water demand sequence in each network node, whereas the accidental/intentional injection of contaminant can occur in any node and at any hour of the day. With reference to this vast range of possible injection scenarios, the first multi-objective optimisation process defines the position of a pre-selected number ns of sensors across the distribution system in order to minimize the expected percentage of undetected contamination events and the expected volume of contaminated water consumed up to the beginning of the response operations following detection. A single configuration of stations is then selected from the Pareto front produced by this optimisation process (“knee point” of the Pareto front). At the end of this first optimisation process and with reference to the selected set of sensors, a potentially contaminated area in the network is associated to each sensor for each sub-period of the day. The second multi-objective optimisation process is then aimed to identify, with reference to each station and sub-period, and thus inside the corresponding potentially contaminated area, the hydrant opening and valve closing operations to be carried out in order to minimize both the number of operations and the expected volume of contaminated water consumed between the beginning of the response operations and the disappearance of the contaminant, assuming the availability of an unlimited number of response teams. Once these devices have been identified (“knee point” of the Pareto front relevant to the second optimisation process), an a posteriori analysis is performed to determine the sequence in which they should be activated based on the number of response teams actually available. In these optimisation processes, a hydraulic and quality simulator (EPANET) is linked to a multi-objective genetic algorithm (NSGA-II) in order to compute the value of the objective functions of the problem across different contamination scenarios. The results obtained applying the procedure to a real and complex water distribution system have shown it to be a robust and effective method for reducing the impact on the population.
Guidorzi, Marco; Franchini, Marco; Alvisi, Stefano
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11392/530287
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 39
  • ???jsp.display-item.citation.isi??? 38
social impact