Digital twin – a good idea but a long way off

“Digital twin” sounds so appealing.  But it’s phenomenally complicated and we’re at least a decade away from having any good ones in the water industry.

The value of digital twins is clear.  Although they’re a recent invention, less than 20 years old, they have helped to improve the designs of power generation turbines, jet engines and locomotives.

But don’t be fooled by pre-existing solutions with a “digital twin” label stuck on them.  There’s a lot more work to be done before this is merited.

Developing a success framework for digital twins” sounded like a promising title but the content didn’t live up to it.  Whilst digital twins don’t have to be perfect to be valuable, let’s see try to create a first draft of the necessary components.

  1. As many components of the system need to be included in the model as possible.  Every pipe length, material and diameter.  Every pump and valve.  Every connection and inter-connection.  Every end customer.
  2. As many actions that change the system need to be recorded as possible.  Every repair, replacement, and addition.  Every outlet pressure adjustment.
  3. As many parts of the system need to report their status as possible.  Every pressure, flow, acoustic, temperature and quality reading; every burst, leak and customer complaint.
  4. Digital twins must be able to analyse ‘what if’ scenarios and issue predictions about the future, in order to guide decision making.  That means the digital twin cannot be built on data alone, it needs to include both data and predictive models.
  5. Models need to run at speed.  “Current digital twins are largely the result of bespoke technical solutions that are difficult to scale,” say healthcare and aerospace experts at King’s College London, The Alan Turing Institute, the University of Cambridge, and the Oden Institute for Computational Engineering and Sciences at UT Austin in Texas in Nature Computational Science.
  6. Models need to account of uncertainty in data, and uncertainty in the model in predictions.

It’s a steep hill to climb.  Whilst the solutions are being developed, water companies would be best to ensure that their GIS records are complete and accurate; instrument and monitor the network; and start to build up registers of network changes.