Scientific and Technical papers:
Supporting a drinking water contaminant warning system using the adenosine triphosphate test
ABSTRACT: A contaminant warning system (CWS) with the capability to detect aberrations in drinking water in real-time or near real-time represents significant value for protection of consumers from accidental or intentional contamination of drinking water. The capabilities of a two-tier CWS are examined, the first including nonpurgeable organic carbon (NPOC), free chlorine, turbidity, pH, and conductivity, followed by a confirmatory adenosine triphosphate (ATP) analysis, to control false positives. The utility of the confirmatory analysis is improved by use of a continuous ultrafiltration system, which improves the detection and the correlation between the concentration of Escherichia coli in the sample and measured ATP. The sample was concentrated one hundredfold in 22 min increasing the ATP value of the sample from 980 to 9.8 × 104 CFU/mL as microbial equivalents. The two-tier system is shown to be a successful method for confirmation of biological contamination in near real-time, while controlling false positives and (or) false negatives.
Fuzzy-Logic Modeling of Risk Assessment for a Small Drinking-Water Supply SystemReal-Time Water Quality
ABSTRACT: A set of risk factors and a fault tree methodology in conjunction with fuzzy logic is described to assess potential contributing factors to failures in water supply systems. The fault tree methodology establishes the structure of the risks, and the fuzzy logic analysis translates qualitative risk data into probabilities. The methodology affords guidance to the decision-making process indicating when the risk is severe and what constitutes a contributory factor that warrants specific attention. The model is applied to a case study in North Battleford, Saskatchewan and sensitivity tests are used to demonstrate the model’s utility to assess the implications of individual factors.
Monitoring: Assessment of Multisensor Data Using Bayesian Belief Networks
ABSTRACT: Real-time sensing in water distribution systems provides a potentially powerful analytical tool for providing water security. Through monitoring surrogate parameters (e.g., pH, turbidity, and residual chlorine) over time, the natural variations of a distribution system’s parameters are established, allowing rapid detection of changes in water quality. However, the level of performance that water quality event detection algorithms have exhibited to date is insufficient for real-world utilization. Bayesian belief networks (BBNs) offer a formalized method of reasoning under uncertainty and are well suited to the analysis of multiple sources of information. The application of a BBN to water quality event detection is described. Surrogate parameters (pH, conductivity, and turbidity) were monitored during an experimental E. coli contamination. Difference filtration using a 60-s moving window of observations identified rapid rates of change present in the surrogate parameter signals, demonstrated as responsive to contamination as simulated in bench-scale studies. A BBN was constructed to assimilate the surrogate parameter variation and compute temporal probability distributions about the contamination of an experimental system. The BBN topology, probability distributions and data transformation techniques applied were validated through successful identification of contaminant injections.
Comparative Evaluation of Two Algorithms for Locating Contaminant Ingress Points
ABSTRACT: A procedure involving Data Mining based on Flow Direction and shortest flow time (DMFD) is described, which consists of two components: (i) possible ingress nodes (PINs) identification and (ii) probability quantification. PINs identification is completed based on the flow information, i.e. the flow direction and time in each pipe. Through shortest time which is calculated by Dijksta algorithm, from one node to a specific sensor, ingress time in the node is obtained. A distance metric is described to quantify the probability of one node as PIN. A multi-stage response is described to decrease the elapsed time before a contaminant ingress event is identified and responded to, which is essential to minimize the risk from consumption of the hazard. The roles of two algorithms, namely a Data Mining method based on Injection and Detection information (DMID) and DMFD, are examined. A case study is employed in a network with 285 nodes and 5 sensors. The five sensors all alarm to the injection event. With DMID, the number of PINs is decreased from 44 in the 1st stage to 18, 11, 11 and 11 subsequently; For DMFD, the number is reduced from 44 in the 1st stage to 21, 21, 21 and 21 in the following stages; DMID identifies the true intrusion node 44 with the highest probability in the five stages, while DMFD identifies it with the highest probability in the 3rd and 4th stage; the run time of both DMFD and DMID is less than 2 min, which suggests the two are effective in guiding emergency response.
Multi-stage response to contaminant ingress into water distribution systems and probability quantificationCanadian Journal of Civil Engineering, 2009, 36(11): 1764-1772, 10.1139/L09-100
ABSTRACT: A multi-stage response procedure for identifying possible ingress nodes (PINs) and quantifying the likelihood that a PIN in a given water distribution system is the actual point of ingress is described. The procedure uses data mining to successively decrease the number of PINs based on a pre-constructed database. In each stage, query sentences are executed to locate the PINs and a Euclidean distance is proposed to estimate the probability, to allow the identification of locations with the highest probabilities of being the true ingress location. As demonstrated in a case study, the ranges of PINs are reduced in the 1st, 2nd, and 3rd stages; except the first sensor alarm, the Euclidean distance metric can identify the true ingress node with the program run-time of less than 2 min; the multi-stage procedure saves roughly 3 h in identifying the true ingress node after the second sensor alarm, instead of waiting for a third sensor alarm to provide the location information. The multi-stage response procedure is shown to be an effective and efficient way for identification and probability quantification of PINs.
ABSTRACT: A contaminant warning system (CWS) with the capability to detect aberrations in drinking water in real-time or near real-time represents significant value for protection of consumers from accidental or intentional contamination of drinking water. The capabilities of a two-tier CWS are examined, the first including nonpurgeable organic carbon (NPOC), free chlorine, turbidity, pH, and conductivity, followed by a confirmatory adenosine triphosphate (ATP) analysis, to control false positives. The utility of the confirmatory analysis is improved by use of a continuous ultrafiltration system, which improves the detection and the correlation between the concentration of Escherichia coli in the sample and measured ATP. The sample was concentrated one hundredfold in 22 min increasing the ATP value of the sample from 980 to 9.8 × 104 CFU/mL as microbial equivalents. The two-tier system is shown to be a successful method for confirmation of biological contamination in near real-time, while controlling false positives and (or) false negatives.
Fuzzy-Logic Modeling of Risk Assessment for a Small Drinking-Water Supply SystemReal-Time Water Quality
ABSTRACT: A set of risk factors and a fault tree methodology in conjunction with fuzzy logic is described to assess potential contributing factors to failures in water supply systems. The fault tree methodology establishes the structure of the risks, and the fuzzy logic analysis translates qualitative risk data into probabilities. The methodology affords guidance to the decision-making process indicating when the risk is severe and what constitutes a contributory factor that warrants specific attention. The model is applied to a case study in North Battleford, Saskatchewan and sensitivity tests are used to demonstrate the model’s utility to assess the implications of individual factors.
Monitoring: Assessment of Multisensor Data Using Bayesian Belief Networks
ABSTRACT: Real-time sensing in water distribution systems provides a potentially powerful analytical tool for providing water security. Through monitoring surrogate parameters (e.g., pH, turbidity, and residual chlorine) over time, the natural variations of a distribution system’s parameters are established, allowing rapid detection of changes in water quality. However, the level of performance that water quality event detection algorithms have exhibited to date is insufficient for real-world utilization. Bayesian belief networks (BBNs) offer a formalized method of reasoning under uncertainty and are well suited to the analysis of multiple sources of information. The application of a BBN to water quality event detection is described. Surrogate parameters (pH, conductivity, and turbidity) were monitored during an experimental E. coli contamination. Difference filtration using a 60-s moving window of observations identified rapid rates of change present in the surrogate parameter signals, demonstrated as responsive to contamination as simulated in bench-scale studies. A BBN was constructed to assimilate the surrogate parameter variation and compute temporal probability distributions about the contamination of an experimental system. The BBN topology, probability distributions and data transformation techniques applied were validated through successful identification of contaminant injections.
Comparative Evaluation of Two Algorithms for Locating Contaminant Ingress Points
ABSTRACT: A procedure involving Data Mining based on Flow Direction and shortest flow time (DMFD) is described, which consists of two components: (i) possible ingress nodes (PINs) identification and (ii) probability quantification. PINs identification is completed based on the flow information, i.e. the flow direction and time in each pipe. Through shortest time which is calculated by Dijksta algorithm, from one node to a specific sensor, ingress time in the node is obtained. A distance metric is described to quantify the probability of one node as PIN. A multi-stage response is described to decrease the elapsed time before a contaminant ingress event is identified and responded to, which is essential to minimize the risk from consumption of the hazard. The roles of two algorithms, namely a Data Mining method based on Injection and Detection information (DMID) and DMFD, are examined. A case study is employed in a network with 285 nodes and 5 sensors. The five sensors all alarm to the injection event. With DMID, the number of PINs is decreased from 44 in the 1st stage to 18, 11, 11 and 11 subsequently; For DMFD, the number is reduced from 44 in the 1st stage to 21, 21, 21 and 21 in the following stages; DMID identifies the true intrusion node 44 with the highest probability in the five stages, while DMFD identifies it with the highest probability in the 3rd and 4th stage; the run time of both DMFD and DMID is less than 2 min, which suggests the two are effective in guiding emergency response.
Multi-stage response to contaminant ingress into water distribution systems and probability quantificationCanadian Journal of Civil Engineering, 2009, 36(11): 1764-1772, 10.1139/L09-100
ABSTRACT: A multi-stage response procedure for identifying possible ingress nodes (PINs) and quantifying the likelihood that a PIN in a given water distribution system is the actual point of ingress is described. The procedure uses data mining to successively decrease the number of PINs based on a pre-constructed database. In each stage, query sentences are executed to locate the PINs and a Euclidean distance is proposed to estimate the probability, to allow the identification of locations with the highest probabilities of being the true ingress location. As demonstrated in a case study, the ranges of PINs are reduced in the 1st, 2nd, and 3rd stages; except the first sensor alarm, the Euclidean distance metric can identify the true ingress node with the program run-time of less than 2 min; the multi-stage procedure saves roughly 3 h in identifying the true ingress node after the second sensor alarm, instead of waiting for a third sensor alarm to provide the location information. The multi-stage response procedure is shown to be an effective and efficient way for identification and probability quantification of PINs.
Conference papers:
"Capabilities of Alternative Sensors in the Detection of Water Contamination"
OWWA/OMWA Joint Annual Conference and Trade show, Blue Mountains, Ontario, May 2011.
"Response to an alarm within a drinking water distribution network"
OWWA/OMWA Joint Annual Conference and Trade show, Blue Mountains, Ontario, May 2011.
"Real-time water quality assessment with Bayesian Belief Networks"
WDSA Meeting, Tucson, Arizona, September 2010.
"Multi-stage strategy and probability utilization for identifying ingress to a water distribution system"
World City, Water Forum, Songdo International City, August 2009.
"Role of warehousing in data mining algorithms for locating contaminant sources in water distribution systems"
OWWA-OMWA Joint Annual Conference & Trade Show, Toronto, Ontario, May 2009.
"Comparative Evaluation of Two Algorithms for Locating Contaminant Ingress Points"
Environmental & Water Resources Institute of ASCE Meeting, Kansas City, Missouri, May 2009.
"An efficient tracking procedure to identify locations of contaminant ingress into water distribution systems"
Computational Hydraulics Inc., Meeting, Toronto, ON, February 2009.
"Use of Bayesian-Belief Networks to improve real-time sensor information"
24th CAWQ Eastern Canada Symposium, Montreal, QC, November 2008.
"Multi-stage response and probability metric selection in data mining for identifying contaminant source ingress to a water distribution system"
24th CAWQ Eastern Canada Symposium, Montreal, QC, November 2008.
"The detection of water quality aberrations after the injection of Escherichia coli into Tap Water"
24th CAWQ Eastern Canada Symposium, Montreal, QC, November 2008.
"Current technologies for on-line monitoring of quality of water in distribution systems"
Stormwater and Urban Water Systems Modeling, Toronto, ON, February 2008.
OWWA/OMWA Joint Annual Conference and Trade show, Blue Mountains, Ontario, May 2011.
"Response to an alarm within a drinking water distribution network"
OWWA/OMWA Joint Annual Conference and Trade show, Blue Mountains, Ontario, May 2011.
"Real-time water quality assessment with Bayesian Belief Networks"
WDSA Meeting, Tucson, Arizona, September 2010.
"Multi-stage strategy and probability utilization for identifying ingress to a water distribution system"
World City, Water Forum, Songdo International City, August 2009.
"Role of warehousing in data mining algorithms for locating contaminant sources in water distribution systems"
OWWA-OMWA Joint Annual Conference & Trade Show, Toronto, Ontario, May 2009.
"Comparative Evaluation of Two Algorithms for Locating Contaminant Ingress Points"
Environmental & Water Resources Institute of ASCE Meeting, Kansas City, Missouri, May 2009.
"An efficient tracking procedure to identify locations of contaminant ingress into water distribution systems"
Computational Hydraulics Inc., Meeting, Toronto, ON, February 2009.
"Use of Bayesian-Belief Networks to improve real-time sensor information"
24th CAWQ Eastern Canada Symposium, Montreal, QC, November 2008.
"Multi-stage response and probability metric selection in data mining for identifying contaminant source ingress to a water distribution system"
24th CAWQ Eastern Canada Symposium, Montreal, QC, November 2008.
"The detection of water quality aberrations after the injection of Escherichia coli into Tap Water"
24th CAWQ Eastern Canada Symposium, Montreal, QC, November 2008.
"Current technologies for on-line monitoring of quality of water in distribution systems"
Stormwater and Urban Water Systems Modeling, Toronto, ON, February 2008.
| google48a8d96c2b07d1dc.html |