Digital twin as experience accelerator

In which situations can a digital twin of the distribution process play a valuable role in the training of new staff? This is the central question in the project entitled: ‘On the use of a digital twin as an experience accelerator for the operation of distribution.’ The boundary conditions for the needed models and data are to be elaborated for the most promising situations.  As a result, this project will offer drinking water utilities support in the concrete configuration of their digital twins. The outcomes will also help academia by providing its members with a benchmark problem, allowing them to test and focus their research with reference to the practice of Dutch drinking water utilities.

Better prepared for incidents

Although digital twins are already in use in many sectors, this is not yet the case for the drinking water distribution process. Digital twins could for instance be very useful in practicing response measures to distribution incidents. This would deepen the insight of modellers and give them confidence as to how to respond when something actually happens. Exercises with digital twins can provide the basis for the development of a script with the measures that need to be taken and optimised. In the event of a real incident, the response can then be managed with greater speed and confidence.

Knowledge transfer

Digital twins also contribute to an accelerated development and retention of knowledge. This is necessary in view of the staff outflow due to retirement, for example, or because new colleagues are nowadays no longer expected to spend their entire careers working at the same water utility. The modelling of operational situations is not only of interest for knowledge transfer to new colleagues, but it can also help improve these processes. For academia the models are potentially useful as benchmarks in research on the optimisation and automation of operational processes. By supplying a benchmark model that perfectly matches the (Dutch) drinking water practice, we will help the science by providing developments that are of maximum possible relevance, as well as connections with questions from water practice.

Elaboration of operational questions

On the basis of interviews with drinking water experts and of literature research, we will compile a list of operational distribution questions that qualify for a model-based approach with a digital twin.

A single operational question will be selected and elaborated as a concrete model-based example. To make it possible to effectively represent this question, we will:

  • Select or set up one (anonymised) hydraulic model and determine its necessary granularity and simplifications.
  • Collect the necessary input data and their connection to the model, with a determination of the required resolution, precision and variability.
  • Elaborate the learning objectives of the operational question, both in terms of the model behaviour and of the interaction of the staff member with the model (‘knobs that need to be turned’).
  • To create a suitable model environment, pay special attention to a clear description, the availability of sufficient (meta) data with minimal data cleaning (ready-to-use data), having a realistic problem, and a model of sufficient complexity, without being too computationally intensive.

The operational question is to be tested with colleagues from drinking water utilities who work daily on water network management. They can provide practical and valuable input to the formulation of solutions for the scenario. Subsequently, the operational question will be developed into an operational benchmark problem. This means that the model, data and learning objectives will be bundled and described quantitatively. Example solutions to this operational question will also be formulated.

What will it yield?

This project aims to produce the following outcomes:

  • A detailed overview of use cases. For each case, a description is given of which kind of drinking water utility staff member can acquire which valuable experience with operational questions, when use is made of a digital twin of the drinking water distribution system.
  • A quantitative elaboration of one of the most promising use cases, with a particular focus on:
    • A systematic (possibly quantitative) elaboration of the question/learning objectives/scenarios behind the use case.
    • The necessary model components with related specifications (for example, regarding resolution, variation, representativeness, and the modelling of specific components like pumps or basements).
    • The necessary input data with related specifications (for example, regarding uncertainty and completeness), and how these can be best collected.

The elaborated use case will be bundled into a so-called benchmark problem which will be shared not only with the Joint Research Programme (BTO) partners, but also with academia. The approach, outcomes and effort that were required to formulate the training problem, will be reported and shared with the international research community through a contribution to a scientific conference. The outcomes and possible follow-up steps aimed at actually organising staff training in this manner will be studied with the BTO Distribution theme group.