For us, at City Science, the Digital Twin for Infrastructure has always been the inevitable evolution of the Internet of Things (IoT). Providing a highly detailed model, driven by sensor data, digital twins have the potential to deliver profound effects on the way that systems are understood and managed over the next 5-10 years. This ranges from a ‘simple’ system such as a single wind turbine, to complex systems like road networks or cities. While ‘digital twins’ may seem like just another buzz word (to some extent it is a new name for detailed modelling), for the first time, these models enable us to apply scientific techniques to dynamic systems, previously beyond our reach. This short blog aims to explain what digital twins are, how they can be used to maximise lifetime asset values and what technologies are needed to deliver a digital twin for Infrastructure.
What is a Digital Twin?
A digital twin refers to the digital representation of a real-world entity or system. Digital twins are computer-based models, linked to their real-world counterparts through real-time sensor data. They can be used to understand the state of an asset or system and collect and store significant amounts of data about the system’s evolution through time. As a result, they can be used to develop new understanding, digitally run scenarios, evaluate appropriate responses to changes and ultimately add value through improved operation.
The concept of digital twins has emerged from Industry, based on the principles of the so-called ‘Industry 4.0’, IoT. Digital models now exist at a range of levels from the single-asset through to infrastructure-scale. Importantly, at the industrial level, the digital twin is based on massive, cumulative, real-time, real-world data measurements across an array of dimensions. This allows for advanced analytics and monitoring systems to be used to identify operational issues, extend asset life and provide enhanced learning from which to establish more accurate longer-term models.
“Over time, digital representations of virtually every aspect of our world will be connected dynamically with their real-world counterpart and with one another and infused with AI-based capabilities to enable advanced simulation, operation and analysis. City planners, digital marketers, healthcare professionals and industrial planners will all benefit from this long-term shift to the integrated digital twin world.” David Cearley, Gartner
Why is a digital twin useful?
Ultimately digital twins offer a compelling opportunity to reduce errors, reduce risk and save money. Digital twins can be used in the strategic phase – understanding what changes to make to a system to make it more efficient. Digital twins can be used in the design-phase – reducing up-front costs and understanding the impact of a change before making it. And finally, digital twin data can be used in the operational phase – enhancing in-use performance through predictive analytics and machine learning, minimising down-time and maximising asset utilisation. In short, digital twins can deliver considerable efficiency to any type of asset or system to which they are applied.
What does a Digital Twin for Infrastructure look like?
To date, digital twins have been developed at a range of scales. The most advanced examples relate to a single asset systems (such as a jet engine), or well-understood processes (such as within a manufacturing environment). Models have been developed that extend to more wide-spread dynamic systems but are critically differentiated from single-asset digital twins due to a paucity of time-series data. As the scale of digital twins increases both in terms of geospatial scope (kilometers vs millimetres) or the modelling time frame (e.g. 10 years vs. 10 minutes) relevant real-world data becomes scarcer. Current macro- or meso-scale models, at the infrastructure scale, use a range of in-filling techniques to back-populate absent data. As a result, true Digital Twinning at an infrastructure scale has not yet been achieved.
Digital Twins at a range of scales
Digital twins are increasingly being used in high-performance engineering and are likely to become more wide-spread in use-cases at greater scales. The following section, provides an overview of the types of modelling and benefits achieved where digital twins are currently being used.
Detailed Asset Performance e.g. Aircraft Engines
The use of digital twins is allowing aircraft engine manufacturers to move towards digital design-and-test processes. The technology is enabling them to significantly reduce engine development time and costs by reducing the number of demonstration physical engines required and enabling rapid virtual testing. For monitoring and maintenance, specialist data aggregation and alert systems, using onboard sensors and live satellite feeds are used to track the health of engines as they operate worldwide. Pro-active techniques predict when something might go wrong and take action to avert threat before it has a chance to develop into a real problem. The technology also enables new predictive analytics models to be developed through the use of machine learning to enhance asset life.
Multi-Asset Systems e.g. Formula 1
An individual Formula 1 car might have over 300 sensors, scientifically located to relay performance-critical information about the individual assets – the tyre pressure, the down force, every aspect of engine performance and many more. Information from these assets is aggregated to create a detailed engineering model of the entire system. These Digital Twin models can be used in design – for example understanding the impact of engine changes, or to understand the trade-off between comfort and handling. Digital twin simulations can also be used during races, to identify issues and understand how to maximise performance given changing conditions.
Larger-scale systems e.g. BIM
Moving up to the scale of the built environment, Building Information Modelling (BIM) creates digital representations of physical and functional characteristics of buildings. BIM models have been trialled and used across a range of construction phases including design, construction and commissioning. In the design phase BIM enables the understanding and visualisation of the asset before it is built; BIM can help understand the full construction before build and can reduce the cost of hand-over upon commissioning. Operational, in-use, or ‘Level 3’ BIM models are far more rare, but increasingly multiple IoT datasets are available within buildings.
City-scale systems
Beyond the building level, accessing real-time data of sufficient granularity and coverage becomes more challenging. However, a number of approaches can be used to assimilate and in-fill data and then model complex interactions and evolutions over time. An example simulation at an urban scale is UrbanSim – an open source land use model designed by the University of California. It was developed to create and assess what-if scenarios to enable long-term planning for infrastructure. It is one of a number of simulation models that aim to forecast changes at the city scale. Although relatively successful, limitations of UrbanSim arise from both data availability issues and model incompleteness. The infilling techniques alluded to above lead to uncertainty in the initial system state and subsequently in any fitted parameters. Error propagation through connected models then follows. Model incompleteness refers to both our limited understanding of the dominant factors influencing the decisions of households and firms and of the fact the cities are highly open systems. The former may arise from a lack of data to support a hypothesis or of our incomplete understanding of city systems (for example how operational aspects influence long term decision making).