EGU General Assembly 2020
© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

DASH of Water – water distribution system modelling in the age of smart water meters

David Steffelbauer1, Mirjam Blokker2, Arno Knobbe3, and Edo Abraham1
David Steffelbauer et al.
  • 1Water Management Department, Delft University of Technology, Delft, Netherlands (
  • 2KWR Water Cycle Research Institute, Nieuwegein, Netherlands (
  • 3LIACS Leiden Institute of Advanced Computer Science, Leiden University, Leiden, Netherlands (

Worldwide, water utilities face exceptional challenges as communities are running out of water and new resources are ill-equipped to meet rising water demands. Furthermore, in many cities, years of stringent financial constraints on water utilities, unoptimized operations and the unaffordability for utilities to maintain and replace their aging infrastructure has resulted in dramatically growing leakage levels, especially in places already under high water stress. Even in Europe, as a matter of fact, nearly one quarter of treated water is lost in the distribution systems before reaching the customers. As a result, the aging water infrastructure is challenged to become more efficient.

Nowadays, an increasing number of water utilities use hydraulic simulation software to design and operate water systems in a more efficient way. However, measurements in water distribution are scarce, which results in inaccurate computer models of real systems. Recently, smart meters have become available as a promising remedy. These smart meters measure water usage of households and transmit information to water utilities in real-time. Now is the time to make water distribution simulation software fit for the future, by exploiting this new Big-data source and start a new era in hydraulic modeling, aiming to increase the operational efficiency of our drinking water systems and save our precious water resources.

This work proposes an innovative new way of combining hydraulic models, data from smart meters and stochastic demand modelling to develop beyond state-of-the-art methods to simulate water distribution systems. It is shown how data science algorithms (e.g. dynamic time warping, clustering, demand disaggregation, household activity identification, …) can be used to extract high-level information from smart meter data (e.g. daily water use routines, work schedules, socio-economic characteristics). Such information is crucial for simulating water demand accurately. Hence, data science algorithms can be used to automatically parametrize stochastic demand models (e.g. SIMDEUM) based on smart meter data, and improve their accuracy. The improved demand models are coupled with hydraulic simulations, leading to a more realistic way of simulating real water systems. Examples on a wide range of real-world applications show how these novel modelling approaches can be used to increase the operational efficiency of drinking water systems. For instance, more accurate models enable faster detection and localization of leaks in water pipes and, thus, minimize distribution losses. This work is part of the project “DASH of Water”, which aims to develop advanced data-driven stochastic hydraulic (DASH) models of drinking water distribution systems.

How to cite: Steffelbauer, D., Blokker, M., Knobbe, A., and Abraham, E.: DASH of Water – water distribution system modelling in the age of smart water meters, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-13439,, 2020.


Display file

Comments on the display

AC: Author Comment | CC: Community Comment | Report abuse

displays version 1 – uploaded on 03 May 2020
  • CC1: Comment on EGU2020-13439, Jiada Li, 05 May 2020

    Thank you for presenting your work. I got some questions.

    1. What are the parameters for SIMDUEM model?

    2. A couple of algorithms are introduced to determine these parameters. What criteria did you use to screen these algorithms?

    3. Is this model open somewhere?



    • AC1: Reply to CC1, David Steffelbauer, 05 May 2020

      Dear Jiada,

      Thank you very much for your interest. Here are the answers to your questions:

      1. The SIMDEUM model consists of a lot of statistical parameters, ranging from different household types, users per household, their age, gender, employment status, … to their average wake-up, being at home, sleep times; and also statistics about the usage duration and intensity of different end-use devices (shower, kitchen tap, dishwasher).
        You can find and overview of these parameters in this article “Blokker, E. J. M., Vreeburg, J. H. G., & van Dijk, J. C. (2010). Simulating Residential Water Demand with a Stochastic End-Use Model. Journal of Water Resources Planning and Management, 136(1), 19–26.”
      2. When I was searching for algorithms that are capable of retrieving the SIMDEUM parameters from SWM measurements, I always had a certain (set) of parameters in mind, since no single algorithm is capable of getting all the parameters at once.
        For the indoor-outdoor disaggregation, I made use of the fact that outdoor uses have in general higher magnitudes, longer run-times and might have repeated patterns. So I went for a rule-based algorithm, once I transformed the SWM signal in single single-rectangular pulses. The rules were applied on each pulse.
        For the pattern recognition, I wanted to overcome the obstacle that similar patterns (e.g. when people leave the house) might be shifted in time, according to their work schedule. Some people have to commute to work for an hour, others work next door. So their might be shifts in their wake-up times, however, both go out for work. Hence I chose a clustering algorithm with dynamic time warping, that can take those shifts into account.
      3. You can get the SIMDEUM model. It is written in Matlab by Mirjam Blokker (KWR), who developed the model during her PhD. Just contact her and she will be happy to share the code with you. At the moment, I am also working with Mirjam and other researchers from KWR to provide a Python version of SIMDEUM, which will be fully open-source. A prototype is already up and running, but a final release might still take some time. We planned to publish the source latest at the end of this year.

      Best wishes,


      • CC2: Reply to AC1, Jiada Li, 05 May 2020

        Hi David: Great! Thank you for your patient reply. I will check the paper and the SIMDUIM model out.