Building digital twins in clean tech - when water and electricity meet in a world with climate change
Digital twins for water, electricity and smart cities have become more than just a buzzword in the past couple of years. As we’ve discussed before, a “digital twin” of any system is when data from the system, whether through sensors, satellite imagery, drones, robots or SCADA systems, are integrated to form a digital representation of what the system looks like over time. While there have always been researchers in academia and the scientific community working on aspects of these systems, we have seen an explosion of interest across all sectors in the past couple of years.
Some of that interest is due to Covid, of course. When the shutdowns happened across the world, a key question that was asked of all physical facilities, especially critical infrastructure like water treatment plants, wastewater plants and power plants, was to develop methods that would allow for remote monitoring of these systems. The answer to that question was a combination of sensor data, machine learning both for outlier detection and for predictive monitoring of the system and “human in the loop” where expert judgement was used to integrate and evaluate data produced by these systems.
These methods form the basis of digital twins. However, one of the limitations of building digital twins in clean tech sectors (agriculture, water, energy, urban sustainability for example) is the lack of data. While high-frequency data from sensors may be available for the last couple of years, often there is not enough historical data to be able to build algorithms that can detect outliers in system behavior accurately. This becomes a bigger issue when we try to build resilience to climate change in our existing infrastructure.
The signal from climate change has only become evident in the last 5 years or so - which is a relatively small section of the 100+ years of historical data that are often used to build predictive models for our infrastructure. Scientists have tried a few different approaches to heighten the signal from climate change - increasing the weight assigned to data from the last few years, building algorithms that use only the last 10 years worth of data instead of the last 100 and creating synthetic data using physics-based models (also known as process models) of earth systems under different climate change scenarios.
Of these approaches, using a combination of historical data from the past 10 years together with synthetic data from process models has been the most useful in building out digital twins of electrical and water infrastructure under climate change. As the climate changes, both electrical infrastructure and water infrastructure separately have vulnerabilities that need to be addressed. For example, some areas are more susceptible to fires sparked by power lines as climate change causes greater periods of dryness and drought. Similarly, some areas are more susceptible to flooding and impacts from sea-level rise - which in turn, impacts the water pipes and treatment plants located there.
One aspect that has not been studied until recently is the combined vulnerability of energy and water infrastructure. A study conducted by scientists and presented at a workshop held by the National Academies of Sciences and Engineering showed that in the regions where energy infrastructure is vulnerable, there are likely to be failures in the availability of power for pumping stations - which in turn is likely to increase the vulnerability of water infrastructure. Scientists developed digital twins for energy and water infrastructure as connected systems and evaluated the vulnerability when the failure in one system led to failures in the other system. For example, a failure in water infrastructure could result in greater flooding near a power plant - which in turn would result in the power plant needing to be shut down and hence a failure in the electrical infrastructure.
By building models that simulated the connections between these systems, scientists were able to identify the regions that were most vulnerable to multiple failures as shown in the figure below. This would enable cities and communities to build alternatives such as backup power generators or decentralized water and energy microgrids that can function as community systems and backup systems. This method also enables communities to create early-warning systems that are more accurate in assessing the threat from combined failures as opposed to single system failures.
Figure 1: Digital twin of electrical and water systems using synthetic data (source: National Academies workshop, 26th July 2023)