Contaminant Spread Forecasting and Sampling Location Identification in a Water Distribution Network
| Main Author: | |
|---|---|
| Language: | English |
| Published: |
University of Cincinnati / OhioLINK
2013
|
| Subjects: | |
| Online Access: | http://rave.ohiolink.edu/etdc/view?acc_num=ucin1383909255 |
| id |
ndltd-OhioLink-oai-etd.ohiolink.edu-ucin1383909255 |
|---|---|
| record_format |
oai_dc |
| collection |
NDLTD |
| language |
English |
| sources |
NDLTD |
| topic |
Environmental Engineering entropy information theory contamination spread forecasting sampling |
| spellingShingle |
Environmental Engineering entropy information theory contamination spread forecasting sampling Rana, SM Masud Contaminant Spread Forecasting and Sampling Location Identification in a Water Distribution Network |
| author |
Rana, SM Masud |
| author_facet |
Rana, SM Masud |
| author_sort |
Rana, SM Masud |
| title |
Contaminant Spread Forecasting and Sampling Location Identification in a Water Distribution Network |
| title_short |
Contaminant Spread Forecasting and Sampling Location Identification in a Water Distribution Network |
| title_full |
Contaminant Spread Forecasting and Sampling Location Identification in a Water Distribution Network |
| title_fullStr |
Contaminant Spread Forecasting and Sampling Location Identification in a Water Distribution Network |
| title_full_unstemmed |
Contaminant Spread Forecasting and Sampling Location Identification in a Water Distribution Network |
| title_sort |
contaminant spread forecasting and sampling location identification in a water distribution network |
| publisher |
University of Cincinnati / OhioLINK |
| publishDate |
2013 |
| url |
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1383909255 |
| work_keys_str_mv |
AT ranasmmasud contaminantspreadforecastingandsamplinglocationidentificationinawaterdistributionnetwork |
| _version_ |
1719435128355160064 |
| spelling |
ndltd-OhioLink-oai-etd.ohiolink.edu-ucin13839092552021-08-03T06:20:21Z Contaminant Spread Forecasting and Sampling Location Identification in a Water Distribution Network Rana, SM Masud Environmental Engineering entropy information theory contamination spread forecasting sampling A safe drinking water distribution system is an indispensable requirement for a developed and healthy community as they rely principally on distributed water for their everyday needs. A contamination event in the distribution system can have severe impacts on the health of the unsuspecting consumers. An efficient way of monitoring water quality is by using online sensors, however, monitoring for specific contaminants is not practical as such sensors will be useless for different contaminants and during normal operating conditions. Hence, monitoring non-specific water quality parameters (e.g. chlorine, pH, etc.) has gained more preference as these monitoring systems can play a dual role - provide non-specific water quality information during normal operations and act as sensors to detect contamination intrusion during possible contamination events.Event detection algorithms (EDA) have been developed as part of contamination warning systems (CWS) with significant success for distinguishing anomalous events from normal operating signals using non-specific water quality data. Additionally, source identification algorithms have been developed that can locate contamination sources after a CWS triggers an alarm. While these developments have been progressing, there remains a lack of tools and methods to interpret these results for the forecasting of contamination spread.In this research, a spread forecasting algorithm has been developed that is based upon the results of a probabilistic contamination source identification (PCSI) algorithm, which estimates the probability that a node maybe contaminated. The forecasted spread of contamination probability to downstream nodes was estimated by assuming the impact of the upstream nodes to be proportional to their flow contributions to the downstream nodes. Additionally a confirmatory sampling location selection algorithm has also been developed that utilizes the forecasted spread to estimate sampling locations to improve the information about the overall water distribution network, which will lead to improved forecasting and source identification.Sampling location selection was based on the expected improvement of information as quantified by concepts of entropy from Information Theory. In Information Theory, lower entropy corresponds to a greater amount of information. The developed sampling location selection algorithm determines the best sampling locations to be those that are expected to minimize the entropy of the entire distribution system when sampled at a particular time.Two distribution systems were used to evaluate the performance of the developed algorithms. The first one was a small 97 node network, while the second one was a large 12,527 node network.The accuracy of the contamination spread algorithm was dependent on the amount of available past node-time (N-T) pair information. Spread forecast accuracy increased with increasing number of sensors due to the concomitant increase of past N-T pair information and generally decreased with increasing forecasting horizon. The estimated sampling locations were shown to maximize benefit in terms of the correct identification of past N-T pairs by the PCSI algorithm. Spread forecast was also improved by the estimated sampling locations. Locations that provided more new information about unknown N-T pairs were found to be better sampling locations than the locations that strengthened existing information about known N-T pairs. 2013 English text University of Cincinnati / OhioLINK http://rave.ohiolink.edu/etdc/view?acc_num=ucin1383909255 http://rave.ohiolink.edu/etdc/view?acc_num=ucin1383909255 unrestricted This thesis or dissertation is protected by copyright: all rights reserved. It may not be copied or redistributed beyond the terms of applicable copyright laws. |