project

DMA Fingerprinting – Fase II

Expert(s):
Claudia Quintiliani PhD, Mirjam Blokker PhD

  • Start date
    01 Jan 2020
  • End date
    31 Dec 2020
  • Principal
    DPWE
  • collaborating partners
    PWN, Evides, Dunea, Waternet, University of Sheffield

The objective of this project is to investigate how the use of transients measurement (pressure waves) with spectral analysis in an exploratory study of closed-valve detection can be scaled up to detect anomalies on a larger scale.

Supplementary experiments planned in this project will provide a clearer picture of which information about the operation of a DMA can be obtained using high-frequency pressure loggers. The project will also generate important insights relating to the definition of the potential applications and limitations of the method (such as the frequency (Hz) that can be used for transient measurements, how to translate/interpret spectral analysis quickly to detect anomalies and, once the spectrum changes, how to determine which disturbance could have generated the changes).

Measurements will be carried out in the pilot area(s) of one of the drinking water companies involved to ensure that all the findings will also be available for the other water companies.

How can we investigate the occurrence of anomalies in a DMA structure quickly and more efficiently?

Incorrect valve statuses can have various adverse effects on the operations of a water company. Examples are (1) pressure problems, (2) a sudden increase in OLM (the interruption of water supply) due to the fact that a neighbouring section has been inadvertently depressurised during work, (3) water quality degradation associated with water stagnation in the pipe when a valve has accidentally been left closed after repairs and (4) the incorrect interpretation of the measurement data and/or incorrect validation of hydraulic models.

With the increasing use of sensors and the application of hydroinformatics in drinking water distribution networks, the importance of the correct position of valves is also increasing. Unfortunately, there are hardly any methods at present that do not require large amounts of manual work and that can be used for investigations from a single location, on a larger scale (in other words, the entire DMA), and to determine to what extent there are anomalies in the network structure. A previous study conducted by the University of Sheffield suggested that measuring transients generated in a network with high-frequency (100 Hz) loggers could provide useful information for the identification of anomalies in normal operations. At the more detailed level, every time someone opens a tap (or takes a shower, uses the washing machine, etc.) a small pressure wave is transmitted through the DMA that is not detectable at the entrance. The geometry of the DMA determines the path of these pressure waves and, in combination with day-to-day demand variations, this leads to a “fingerprint” of the DMA in the spectral domain. Changes in geometry (associated with, for example, the closure of a valve or a leak) result in different paths and therefore to a different fingerprint.

The 2019 BTO Exploratory Study included the first field test which showed that the spectral analysis of transients measurements in a DMA does establish a fingerprint and that anomalies can be detected. In order to take these results further, the present project will conduct more extensive research, including additional experiments in the field that are expected to provide an insight into the practical applicability and limitations of the method and to establish the basis for its future development.

Use of high-frequency loggers for transients measurements

The project will use pilot area(s) of one of the drinking water utilities involved in the joint research programme. Project development will draw on international expertise (particularly from the UK).

High-frequency loggers will be installed in various locations in the DMA(s) and data will be recorded for several days in order to obtain the fingerprint of the system in normal circumstances, and then in various anomalous situations. The latter will include operations such as the total/partial closure of valves, the opening of hydrants for a limited time and artificially-generated leaks in various locations. Additional anomalies and measurements will be implemented after increasing the size of the DMA and changing the number and locations of the supply.

The measurements collected (the signals registered by the loggers) will be decoded and then elaborated through the Fast Furrier Transformation to obtain the spectrum of frequencies. The interpretation of the spectrograms will tell us if a change has occurred in the DMA and how each disturbance has affected the fingerprint.

Transients spectral analysis as a tool to obtain more information about DMA operation

The main expected outcomes of the project will be:

  • the supplementary measurements planned will provide a clearer picture of which information about the operation of a DMA can be obtained with high-frequency pressure loggers: whether various anomalies (such as a fully or a partially closed valve or a leak) can be detected clearly using spectrum analysis.
  • to determine whether spectrum analysis could lead to similar outcomes when it is also applied to larger DMAs;
  • insight into the frequency (Hz) that should be used for transients measurements in order to identify anomalies;
  • insight into how to translate/interpret spectral analysis quickly to identify anomalies and into whether, once the spectrum changes, it is possible to determine which anomaly could have generated the changes (a valve closure, a leak, a hydrant, etc.).

Findings for the pilot area(s) are expected to generate additional research questions and they will constitute a basis for planning future experiments and developments. The latter may give drinking water companies the opportunity to extract additional information about DMAs that will be useful in the future to improve design or for decision-making about operational interventions. Furthermore, other applications may result, such as the possibility of testing whether a valve closes properly, or a way to improve leak localisation in the network when flow and pressure are not accurate enough for that purpose.