Going Dutch! Smart Drinking Water Networks

By Ina Vertommen, Dr Karel van Laarhoven, Dr Mirjam Blokker, Dr Peter van Thienen 19 August, 2019

Making drinking water systems 'smart' is like adding a nervous system. Learn how this works from KWR experts

Many drinking water networks are far from an optimal design; however, the number of possible re-designs is so large that optimisation techniques like Gondwana are needed
Gondwana was used to revamp Helmond-Mierlo's water network which serves 105,000 inhabitants; the result: a saving of 64% (€15mn) in costs & better hydraulic performance
The optimisation tool also has other purposes, such as dividing water networks like at the Hague; ultimately, need to combine a smart 'body' and the 'nervous system' for best results

Smart networks are drinking water networks equipped with sensors, which allow a water utility to better control the network in terms of quantity, quality, or the condition of the pipes. The term may also refer to the use of algorithms to identify failures (such as leaks) or for advanced process control.

Drinking water smart networks are equipped with sensors…

…it is like adding a nervous system to the water infrastructure

All this can be considered as adding a nervous system to the body of the infrastructure. It enables water companies to operate their networks ‘smarter’, but it does not really make the networks themselves smarter. A network’s body of infrastructure could be designed to perform ‘smarter’ continuously (see [1] for more details on this concept of ‘intrinsically smart network’). In this article, we present two case studies to illustrate how numerical optimisation techniques are used to design smart network bodies and nervous systems.

Smart ‘body’ design: network blueprints

Many of the existing networks have grown organically and are quite far from an optimal design. As the aging network is being renewed, water utilities get the opportunity to improve their network’s performance, while considering future developments in water demand, regulations, or aging networks. This means that instead of replacing a pipe with a new one of the same diameter, a larger pipe or a smaller one, or perhaps even no replacement pipe needs to be installed. This requires the design of the system as a whole, to prevent suboptimal local solutions. The envisioned improved design is documented in a network blueprint, which can be used by water companies as guidelines when rehabilitating their networks.

No. of possible network designs is so large…

…thus, optimisation techniques are needed

Designing drinking water networks is aided by hydraulic network modelling software. However, the number of possible designs is so large that it is practically impossible to examine even a small number of the potential configurations.

  

Numerical optimisation techniques, as implemented in Gondwana [2, 3], are the solution to generate designs that meet all requirements and perform optimally on user-defined criteria.

Gondwana – such a technique – was tested on Helmond-Mierlo’s water distribution network…

The capacity of Gondwana in redesigning a network was tested on the water distribution network of Helmond-Mierlo, operated by the water company Brabant Water. The network supplies water to around 105,000 inhabitants, and it is represented by a network model consisting of around 15,000 nodes and 12,000 pipes. The costs for rebuilding the entire current network would be €41.1mn. At peak demand conditions (day with maximum demand of the last 10 years), the minimum required pressure (300 kPa) is not met at a few hundred nodes of the network.

…Costs for the optimised blueprint are only 64% of those of the pipe-for-pipe replacement (€26.4mn vs €41.4mn)

Figure 2 illustrates the resulting diameters for the optimal network blueprint, considering minimisation of costs as the objective function, and minimum pressure requirements as a soft constraint (i.e. when the minimum pressure at a node is not met the design solution is penalised). As can be seen the diameters for the majority of the pipes are decreased in the network blueprint, while a backbone structure is identified where pipe diameters are increased. The costs for the optimised blueprint are only 64% of those of the pipe-for-pipe replacement (€26.4mn vs €41.4mn). At the same time, the hydraulic performance of the network in terms of meeting the required pressure at all nodes is improved. Figure 3 illustrates the maximum pressure violation.

Smart ‘nervous system’ design: sensor locations to create DMAs

DMAs (district metered areas) are parts of the distribution network that are isolated from the rest by flow metres and/or closed boundary valves. For years, DMAs have been playing an essential role in pressure regulation and leakage loss reduction in many countries [4, 5]. In the Netherlands, however, DMAs have not been applied, historically, probably due to both limited height differences and small water losses (<6%). Dutch drinking water utility Dunea recently developed an interest in using DMAs, mainly to understand more about the local behaviour of water flow and demand in its distribution network through continuous monitoring.

DMAs play an essential role in pressure regulation & leakage loss reduction…

This presents Dunea with the challenging task of subdividing a large distribution network into DMAs despite the fact that it was never designed with DMAs in mind. KWR worked together with Dunea to determine DMA boundaries where sensors can be placed for Dunea’s largest and most complex distribution network: that of the city of The Hague [6]. This network is highly meshed, which complicates the search for efficient DMA designs that can be realised with a limited number of DMA boundaries. Minimising the number of boundaries is crucial, however, since installing flow meters and/or valves is an expensive and arduous task.

Moreover, Dunea is reluctant to close pipes on DMA boundaries to save costs, lest this degrades the reliability of supply and intrinsic resilience of the network, making the minimisation of boundaries all the more important. The evolutionary algorithms implemented in KWR’s optimisation tool Gondwana proved well suited to solve this classic network partition problem [7] for the utility.

…Again Gondwana proved well suited to integrate DMAs into Dunea’s system; despite not being part of the original network

 

 

Through an iterative cycle of design sessions, experts from Dunea and KWR worked closely together to execute a series of numerical optimisation steps. In each session, it was evaluated how well the latest results matched with practical boundary conditions such as: costs, digging activities in the historical city centre, security of supply and sensor detection limits.

Based on this, the optimisation could be adjusted further each cycle, in order to arrive at an optimal design that fitted the water utility practice as well as possible (figure 4). This design creates 15 DMAs with 92 boundaries. Dunea and KWR are currently investigating which of these boundaries can be closed and which must be outfitted with a flow metre. The next step is organising a gradual realisation of this design; the first ‘cut’ illustrated in the figure will be executed in 2020.

Closing remarks

The concept of smart drinking water networks has gained ground over the past ten years. The Dutch water utilities embrace these concepts and have started to implement them in pilot projects such as those discussed in this article. Evolutionary algorithms prove to be a powerful tool to support these smart designs. Moreover, they will prove indispensable to automate and structure the enormously complex task of redesigning hundreds of thousands of kilometres of existing infrastructure.

Evolutionary algorithms prove to be a powerful tool to support smart water network designs

The application of such algorithms, however, requires a clear definition of goals that do justice to all aspects of a smart design, including both a smart ‘body’ and ‘nervous system’. Such goals can only be defined in close collaboration with experts from the drinking water utilities. This will then lead to a network with optimal diameters, valve locations, sensor locations, and so on, which will in turn lead to an optimal balance between water quality, continuity of supply and cost.


References

[1] van Thienen, P. and E.J.M. Blokker, Towards intrinsically smart drinking water networks, in Water Matters 2019. Available at: https://library.kwrwater.nl/publication/59485681/
[2] van Thienen, P. and I. Vertommen. Gondwana: A generic optimization tool for drinking water distribution systems design and operation, in Procedia Engineering 13th Computing and Control for the Water Industry Conference, CCWI 2015. Available at: https://www.sciencedirect.com/science/article/pii/S187770581502648X
[3] Vertommen, I., K.A. van Laarhoven, P. van Thienen, C.M. Agudelo-Vera, T. Haaijer, and R. Diemel. Optimal Design of and Transition towards Water Distribution Network Blueprints, in Proceedings – The 3rd EWaS International Conference 2018. Lefkada, Greece. Available at: https://www.mdpi.com/2504-3900/2/11/584/pdf
[4] UK Water Authorities Association (1980). Report 26 Leakage Control Policy & Practice.
[5] Farley, M. (2001). Leakage management and control: a best practice training manual. W. H. Organization. Geneva, Switzerland.
[6] van Laarhoven, K.A. and D. Gardien, How to detect as many leaks as possible with as few flow meters as possible, in Water Matters 2019. Available at: https://library.kwrwater.nl/publication/59485407/
[7] Kim, J., I. Hwang, Y. Kim, B. Moon (2011) Genetic approaches for graph partitioning: A survey. Proceedings of GECCO’11, Dublin, Ireland.
[8] Rossman, L. A., EPANET 2 Users Manual, U. S. Environmental Protection Agency, Cincinnati, OH, USA, 2000.
[9] Garrett, A. L., Inspyred 1.0 Documentation. Available online at: http://pythonhosted.org/inspyred/overview/html (last access on January 2015)

Further Reading

  • 8 Asia Water Risks: Here Today & Here To Stay In Asia – Damaging typhoons, life & business disrupting water outages and threatening sea level rise… China Water Risk review’s 8 water threats too great to miss in Asia from just the past 3 years
  • Hot, Thirsty, Sweaty & Wet: HK’s Future Down The Drain? – China Water Risk’s Woody Chan & Debra Tan look beyond current tensions and see very real threats to Hong Kong’s future from climate change. Get ready for a hot, thirsty, sweaty & wet future
  • Climate Change, Groundwater & Agriculture In India – The hidden risks of groundwater are clear in India as it is key for the country’s food security and already is largely over extracted. What can India do? Dr Aditi Mukherji from the ICIMOD, shares ways forward
  • 3 Takeaways From CEWP’s 2019 Groundwater Policy Dialogue – With China and Europe joining forces to tackle groundwater over-exploitation, China Water Risk’s Yuanchao Xu was on hand to bring us the latest policy and tech ideas from the Jinan forum
  • Environmental Watering In The Murray-Darling Basin – Megan McLeod from the Alliance for Water Stewardship explores how the Renmark Irrigation Trust benefits the Murray-Darling Basin by providing environmental watering, enhancing biodiversity and promoting tourism
  • The Power of Pipe Management – Mark Nicol from Echologics tells us on how acoustic technologies can non-evasively detect underground leaks as well as save water. Globally, 35% of water supplied is lost through leaking pipes; managing this is is key given rising urbanisation
  • Water As Leverage For Resilient Cities – Water represents man’s most challenging & complex risk but it can be leveraged for catalytic change. China Water Risk asks Henk Ovink, the first Special Envoy for Water in the world, how this can be achieved
  • Hong Kong’s Pricey Water Deal With China – Much is made of the DongShen Agreement’s price tag but discussions need to move onto more complex issues such as the city’s rampant overuse & leakage. Hear from Civic Exchange on HK’s ‘illusion of plenty’
  • Rural Drinking Water Solutions – 783 million people in rural areas still lack safe drinking water due to diseases coursing through waterways. Ling Li on why a traditional water distribution system is not necessarily the best answer & shares cheaper alternatives

Ina Vertommen
Author: Ina Vertommen
Ina Vertommen is a scientific researcher in the Water Infrastructure (WIS) team. She works with her colleagues on the development of the Gondwana optimisation platform and translates water-practice problems, such as the design of network blueprints and DMAs, into mathematical optimisation problems. Ina also researches the impact of the weather, holidays and vacation periods on water consumption, and the use of customer complaints as alternative indicators of water-quality problems. Moreover, Ina is experienced in the detection of leakages and changes in consumption patterns based on the CFPD method, contamination source identification for water-quality incidents, and the registration and analysis of failures in USTORE. As a basis, Ina holds a Master’s in Environmental Engineering from the University of Coimbra in Portugal. Her Master’s graduation research, at Sapienza University in Rome, focused on scale effects in water consumption. Her doctoral research (at both Coimbra and Sapienza) deals with the uncertainty in water consumption and the optimal design of distribution networks. During her course of studies, Ina has built up a knowledge base about the uncertainty in water consumption, statistics and optimisation problems and techniques. Thanks to her international background, Ina also speaks Portuguese, English and Italian, besides Dutch.
Read more from Ina Vertommen →
Dr Karel van Laarhoven
Author: Dr Karel van Laarhoven
Karel van Laarhoven works as a scientific researcher in the Water Infrastructure team. In 2016, he received his doctorate in physics from Eindhoven University of Technology. His doctoral work dealt with the interface of thermodynamics, microbiology and fluid transport in permeable materials. Multidisciplinary collaboration and practice-orientation played a central role in this. On the same basis, Karel will be busy at KWR with various subjects in the field of water distribution. Specifically, he will be focusing on the role of pipe materials in problems associated for instance with water quality or infrastructure failure.
Read more from Dr Karel van Laarhoven →
Dr Mirjam Blokker
Author: Dr Mirjam Blokker
Mirjam Blokker is a principal scientist on the drinking water infrastructure team. She is an expert in drinking water demand and developed the SIMDEUM model, which can be used to predict the demand for shower water, toilet flushing water, water consumption etc. With this model, she also did research on the impact of flow speeds and residence times on water quality in the pipeline network and completed her PhD on that topic in 2010. As an extension to this, Mirjam carried out the research and implementation in designing the water network. In recent years, Mirjam researched the temperature in the pipeline network and the microbiological regrowth in the network. In collaboration with her colleagues on the microbiology team, Mirjam developed models for quantitative microbiological risk analysis (QMRA) for collection, purification and in the distribution network. Mirjam’s knowledge of statistics came in rather handy in her research on pipeline, valve and fire hydrant failure. Mirjam also stood at the forefront of the introduction of the performance indicator for inadequate supply minutes (OLM). Mirjam owns a few Watershare tools. She is a guest editor for Water, special issue on “Water Quality in Drinking Water Distribution Systems”
Read more from Dr Mirjam Blokker →
Dr Peter van Thienen
Author: Dr Peter van Thienen
Peter van Thienen is Chief Information Officer at KWR. He is responsible for coordinating Hydroinformatics, a combination of water research and information technology. From his background as a quantitative physicist, Peter van Thienen and a number of colleagues focus on the drinking water distribution network from a quantitative, model-based perspective. Key words in this are mathematical-physical models and optimisation. The starting point is to transform a simple concept into a powerful tool. A few examples include the analysis of volume flow data to understand the network and the identification of leaks (including the development of the VLPV method), the numerical optimisation of the design of networks (development of the Gondwana platform), and the initiation of the development of an inspection robot for drinking water pipelines (Ariel). These have applications in Joint Research Programme, TKI and advisory projects as well as internationally in a Watershare(r) context.
Read more from Dr Peter van Thienen →