Digital twins: the key towards a highly automated and autonomous water sector
Unpredictability has been a trend in 2020, with COVID-19 being one of the greatest challenges in a generation. However, we must not forget the ongoing threat of climate change which is connected to a myriad of events such as the typhoon Haishen which has hit South Korea recently. These extreme events can impact lives especially by limiting access to water and sanitation. Digital twins give utilities actionable insights that help them predict performance and identify failures before they happen.
Process operation in the water sector is not yet completely automated considering that gigabytes of data are being collected over the years. However, only a small part of the collected data is actually used. Instead of creating data graveyards, we need to use the information sitting in these data for automation and decision support. Here is where digital twins can come to the rescue.
There are many advantages that digital tools can provide to the water sector. From my experience, digital twins can significantly improve the decision-making process in the operation of treatment trains and distribution systems. Just like a flight simulator, digital twins allow an operator to test several strategies and pick the “best” one under all circumstances. This not only helps to achieve operational excellence but is also useful for staff training. It could even be possible to include the operator’s knowledge in the digital twin thus avoiding losing knowledge when staff retires. This automatically leads to improved cost-efficiency, which allows keeping the customer bill minimal, but it also gives a better guarantee of the highest quality service.
The work at my research group BIOMATH (Ghent University) focuses on developing various process models of high predictive power to be used as digital twins and soft sensors. To achieve this, we develop so-called hybrid models that complement mechanistic models with data-driven models (AI) to boost predictive power. We use these models to develop advanced control strategies to allow operation at optimal conditions at all times.
Predictive power is a key prerequisite when using models such as digital twins, i.e. in an online fashion to support semi-autonomous decision making. Developments are both pursued for drinking/process water and wastewater treatment process trains as well as for the transitional paradigm shift to resource recovery.
Considering that on-line monitoring and visualization have now become mainstream, the future of digitalisation (and digital twins) is to use this data to optimise processes and systems at all times. But this is a much harder challenge: it means developing and embedding reliable (i.e. high predictive power) digital twins in the SCADA (supervisory control and data acquisition) itself and start making decisions based on their information.
The actual application and experiences in practice will point to weaknesses that we then need to address, but that will be partly automated through self-learning digital twins. Within 5-10 years, I see many systems in the water sector running in a semi-automated and semi-autonomous mode, just like robots partly took over manufacturing processes some decades ago. The role of staff will have shifted to understanding how these digital twins work, using them in decision making, and maintaining them.
Another trend I see, which will be emphasized by climate change effects becoming more and more pronounced, is that major industries will adopt similar digital technologies to increase water reuse in their production processes. Hence, the knowledge of the water sector will be integrated more into several industrial sectors. The versatility and automated controllability of these systems by means of digital twins will then be a huge benefit and even a requirement.
To learn more about Digital Twins and models for digital water, please watch The Use of Models in the Digital Era or read https://www.iwapublishing.com/books/bookarticle-author-editors/ingmar-nopens