project

INTEREST – Investment in, and returns from, sensor networks

Expert(s):
Peter van Thienen PhD

  • Start date
    01 Mar 2015
  • End date
    30 Jun 2016
  • collaborating partners
    Dunea, Evides, KWR Watercycle Research Institute, Optisense B.V., Vitens en WMD

The current monitoring of our drinking water quality is done on the pure water as it leaves the production location and as it flows from the customer’s tap. This is done through random sampling and lab analysis of a limited water volume. However, more and more water quality sensors are being put on the market, which in principle could significantly improve monitoring coverage, both in terms of time and space.

Technology

Several Dutch drinking water companies are experimenting with water-quality sensors in their distribution networks, but many have yet to install them in the networks. A key explanation for this concerns the lack of clarity about what the sensors actually provide. A chicken-and-egg situation has arisen: the real returns will only become clear following large-scale implementation, but without a very clear sense of these returns the water companies won’t undertake the investment. Recent BTO research in the area has produced a method which permits the quantification of a sensor network’s performance, and connect it to the investment required to achieve said performance (see Figure 1). This provides the basis for a solid business case for sensoring water quality in the distribution network.

Challenge

The performance of water-quality sensor networks can be effectively predicted through numerical simulations, without having to actually install the necessary network. In this project such simulations are conducted for a representative selection of characteristic network forms (mainly meshed, mainly branched, etc.), and whenever possible for several networks per type. This is done for varying numbers of sensors, placed in optimal locations (as determined by simulations). Different types of pre-selected sensors are also considered. In this way, a relationship is quantified between the number of sensors in a network and the performance of the sensor network, for individual networks and types of sensors. The application of a cost model for sensors (based on acquisition, maintenance, operations, etc.) produces a relationship between investment and performance. Subsequently, the results for individual networks and/or sensor types are collected in an attempt to distil investment rules of thumb for different types of networks/sensors and, whenever possible, for general application.

Solution

The overall result is the quantified performance of a sensor network as a function of the investment in its installation and maintenance. The water companies thus dispose of a welcome quantitative basis for decisions regarding business cases for sensoring water quality during distribution. Moreover, sensor suppliers have a quantitative and scientifically-grounded marketing instrument to create realistic and well-founded expectations on the part of their clients.

Interest graph

Figure 1: Required relative investment as a function of the performance achieved (coverage ratio) and the associated sensor density.