The current environmental monitoring of the city’s waters highly depends on manual field sampling, and laboratory testing. This does not give a sufficiently high spatial and temporal resolution to be able to execute measures in a modern and effective way.
Recreation is one of the uses of the city’s waters by citizens. From a health and safety perspective, the City needs to be able to provide citizens with as updated information regarding the quality of the water as possible. Measuring and informing about the bath water quality in real time is of utter importance to achieve this. Currently, it is difficult for citizens to obtain information regarding bathwater quality. This is mainly due to data being hard to interpret, but also due to sampling and analysis not being carried out often enough to be usable.
Bacteria in the drinking water can be a potential risk for spreading disease, meaning continuous measurements and follow-up is very important. Even if samples are taken regularly, a more effective method for detection can save large amounts of money whilst assuring our drinking water has the highest possible quality.
By producing a cloud-based sensor technology identifying changes in the water and analyzing the results automatically, large amounts of data can be collected without burdening personnel resources. Near real time update also gives the temporal resolution the conventional environmental monitoring is not able to provide currently. This creates a possibility to allocate resources where the most effect is obtained and to get a better picture of the dynamics of the city’s waters.
The use of an IoT-platform and algorithms with AI-analysis creates the possibility for multisensory analysis, which means a network of sensors connected to each other becomes a possibility. This could be used to track spills and leakages in pipelines, and identify bathing sites at risk for contamination of bacteria. Other kinds of data could be compared to the results regarding water quality, such as storm water dynamics and car traffic to mention some. By combining multiple datasets, the knowledge regarding how and when different sources affect the city’s waters would increase.
By creating a user friendly presentation of the collected data, the information can be used by both citizens as well as operators and managements in the City. This creates an opportunity for citizens to understand the dynamics of the water as well as sources of pollution, and to be able to judge the water quality themselves. This could lighten the burden on officials answering these questions e.g. during the summer high season for bathing.
By using this technology in the drinking water network, contamination can be identified faster, and early warn for poor-quality water. Sensors are already at use in drinking water production plants, but not with the same wireless and automatic configuration as tested in iWater.
The project has entered its second phase, and has a partial grant from VINNOVA. During the summer of 2019 the IoT data pipeline platform was finalized by Ericsson, as well as the algorithms created by KTH. During the fall of 2019 the data collection will be started, and verification and evaluation of algorithims and data catchment will follow. The sensor from the Linköping University is also ready for deployment, and will be tested at Tekniska Verken in Linköping during the fall of 2019.
Ericsson Research has provided an IoT based data solution called “iWater Data Pipe” which orchestrate various tasks such as data collection, data analysis, and data visualization. The process of collecting, analyzing, and visualizing is automated in a click-run manner, to take in raw data and produce an easily interpretable output without practical effort.
KTH has been working on fundamentally new machine learning theory over networks, and has developed practical machine learning algorithms using recurrent neural networks for the iWater IoT network. KTH has tested the machine learning algorithms with data collected in 2017, and is working on incorporating the code with the IoT-platform provided by Ericsson. KTH is also working with with University of Linköping to develop new machine learning algorithms for the “electronic tongue” for distribution network monitoring.
Stockholm University has been contributing to the geographical set-up of the sensor network, the choice of key chemical substances to monitor, and the interpretation of historical and real time data and main chemical changes occurring over time. Through different meetings with project partners the place for sensors in Lake Mälaren was defined.
Linköping University has developed its sensor “electronic tongue” to measure bacterial detection, which has not before been used in drinking water production. The sensor does not measure bacteria directly, but indicator substances suggesting contamination. The sensor is currently adapted to drinking water application.
The City of Stockholm (Environment and Health Administration), Stockholm Vatten och Avfall AB, Ericsson, Telia, Royal University of Technology (KTH), Stockholm University (SU) and Linköping University.
The project receives partial funding from Vinnova.
The project is run and coordinated by the Environment and Health Administration at the City of Stockholm.
Maya Militell – Stockholm stad
Dr. Bin Xiao – Ericsson, technical coordinator
Prof. Carlo Fischione and Prof. Viktoria Fodor – KTH
Prof. Gia Destouni and assoc. prof. Zahra Kalantari – SU
Assoc. prof. Mats Eriksson – LiU
Marie J. Karlsson, Joakim Strandh and Henrik Werner – Telia
Tommy Giertz – SVOA
Good status of the water is an important and highly prioritized aspect for the City of Stockholm, as well as for the EU. To ensure the city’s waters maintain good quality and follows the EUs framework directive, sensors or other technologies can be used to measure the waters condition in real time. The data collected can later be made available in real time as well. In this way, the water quality of the city becomes easy to track. Data produced can also be used as a base for simulations of e.g. effects from exploitation near waters.
The potential of continuous measurements of pollutants in recipients and in the water distribution network is to make identification and action against pollution early on possible, and to minimize eventual negative effects. If the water can be measured at multiple sites, it is easier to identify the spatial extent of a pollution, and focused action and implementation is possible, which minimizes the effect on drinking and bathing waters, so that a high quality can be ensured as resources are used effectively.