Let’s face it: making decisions is hard. Just consider the following hurdles to overcome:
- You need to figure out how to actually evaluate decisions. What performance metrics truly matter, how would you rank or weight them?
- Once you know what constitutes the quality of decisions, you need to estimate how various decisions score on the defined metrics.
- The consequences of decisions typically reveal themselves only in the future. There is no instant feedback mechanism.
- Decisions are rarely made in isolation, but they interact with the environment and decisions made by others.
Naturally, with decisions come mistakes, and mistakes are both costly and painful. As Warren Buffett phrases it:
"It’s good to learn from your mistakes. It’s better to learn from other people’s mistakes."
Essentially, this quote captures the spirit of the Digital Twin. Failing in real life is expensive, but failing in the virtual world is cheap. A digital twin is a virtual representation of a real-world system, which can range from a physical system (e.g., a machine component or a wind turbine) to a full supply chain or even a complete city. In this virtual proxy, you can easily test many decisions and run many scenarios to observe how decisions pan out.
But wait, isn’t that just a simulation environment?
At this point, you might wonder whether a digital twin is not simply a simulation environment. Haven’t those been around for decades? Yes and no. Although you might argue it’s just a fancy new name for an old concept, the nuance is in the scale and richness of the model. In particular, the digital twin is rooted in real-time data, maintaining its similarity to the physical counterpart over time.
Let’s look at some definitions on digital twins.
IBM:
"A digital twin is a virtual representation of an object or system that spans its lifecycle, is updated from real-time data, and uses simulation, machine learning and reasoning to help decision-making"
Forbes:
A digital twin is a virtual model of a process, product or service. This pairing of the virtual and physical worlds allows analysis of data and monitoring of systems to head off problems before they even occur, prevent downtime, develop new opportunities and even plan for the future by using simulations.
Royal Haskoning:
"Driven by relevant data, a digital twin is a virtual mirror of a real-world asset, process or even an entire system, like a supply chain or city. This mirror provides insight into how that asset behaves under a variety of simulated conditions, helping you to improve decision-making and optimise business processes."
Twi-Global:
A digital twin is, in essence, a computer program that uses real world data to create simulations that can predict how a product or process will perform. These programs can integrate the internet of things (Industry 4.0), artificial intelligence and software analytics to enhance the output.
Wikipedia:
"A digital twin is a virtual representation that serves as the real-time digital counterpart of a physical object or process. […] The digital twin of a physical object is dependent on the digital thread – the lowest level design and specification for a digital twin – and the "twin" is dependent on the digital thread to maintain accuracy."
To paraphrase: despite some differences in interpretation, all stress the importance of data, the degree of realism and the application for decision-making. In particular given the rise of IoT techniques and the emergence of data platforms, we nowadays have sufficiently rich and granular data to progress from abstract simulation models towards representative digital twins.
The Digital Twin
So, when was this digital twin born? David Gelenter may be the spiritual father, outlining the concept in his 1991 book Mirror Worlds. The first manufacturing application followed in 2002. In 2010, NASA was the first to truly adopt the digital twin in its spacecraft program. It is only fairly recently that the concept really took off though, spurred by the surge of IoT technology, machine learning breakthroughs, and the explosion of data availability.
In 2017, Gartner identified digital twins among the Top 10 Strategic Technology Trends. Indeed, they have been deployed across a wide array of industries and problems since. Digital Twin Technology has the potential to improve existing products and processes, as well as identifying untapped sources of innovation. I will refrain from further lavish projections here, but digital twins surely will be booming business in the years to come.
For data scientists, the symbiosis between the physical and digital world is of key interest. A digital twin is not a parallel reality – it has a continuous feedback loop with the real world, with data driving decisions and decisions producing new data. With a data scientist’s job being to explore data and produce actionable insights, the closer semblance between both world will encourage new collaborations between domain experts and data scientists.
Applications
It would be easy to list a number of engaging examples here, but let’s stick to a single one. Recently, General Electrics opened up a "digital wind farm" that mimics the real world. For each physical windmill in the real world, a virtual counterpart runs in the cloud. With each passing seconds, new operational data is gathered, creating an ever-richer digital proxy as we speak. Before a new wind turbine is constructed, a virtual twin is being run first in the digital world, allowing to fine-tuning its configuration in advance.
Still some more examples? Consider urban planning (Sim City, but next level), healthcare (constructing a ‘virtual patient’ based on detailed medical data), manufacturing (Product Lifecycle Management) and the automotive industry (designing car features based on observed driving styles). At this point, the only limitation seems to be our own imagination.
Design characteristics
Digital twins have a number of interesting and rather unique attributes. The following are key design characteristics that define a true digital twin.
Connecting
The key distinction between a simulation environment and a digital twin is data. The IoT, with all its sensors and app data, allows to track physical processes on an unprecedented scale. As such, it pulls the physical and the digital world closer together, with the digital proxy increasingly becoming a – well, twin – of reality. Consequently, barriers between organizations, products and customers can gradually be decreased.
Data decoupling
In a notion highly similar to that of data platforms, a digital twin by nature decouple data from its physical entity. Data nowadays can be collected, transmitted, processed and stored at very low costs. By segregating data as a separate business layer, the door is opened to apply a wide array of analytical tools and learning algorithms. In short: information becomes freely accessible to all stakeholders, without the need to physically observe the real system.
Diagnosing
With digital twins being such an accurate representation of reality, the virtual proxy may be used to diagnose and resolve problems. For instance, if a car crash occurs, the collected data of the car may be used to trace back the root of the problem. As such, safety features may be finetuned in future iterations of car components. Similarly, digital twins may be used to, e.g., diagnose bottlenecks in manufacturing processes.
Reprogramming
Data acquired from reality is not only intended to enrich the virtual world, nor are digital twins restricted to passively provide decision support to those who look for it. Based on the information obtained from the real world, the digital twin can actively identify and test improvements, directly feeding back the results to the physical system. For instance, software can be updated on the fly, based on received user data. Machine Learning algorithms are particularly suited to the reprogramming task, continuously tweaking and improving processes based on the ever-continuing stream of data.
Scaling and shaping
As you may have surmised at this point, data twins can play a variety of roles. Testing a product prototype before it physically exists demands a different environment than monitoring a factory process, for instance. A twin may be as small as an individual component or as large as a macro-process. Digital twins can be molded in many shapes according to their application in decision-making.
Closing words
The potential of digital twin technology is no less than fascinating. The richer and more accurate virtual worlds become, the more potential they harness to drive real-life decision-making. The – somewhat polygamic – marriage between IoT, Simulation, machine learning, data platforms and human experts proves to be particularly fruitful. Modeling such a detailed environment takes time, but from there onwards real-time data ingestion and machine learning can be deployed to automatically build ever more accurate representations of reality. At the moment, the opportunities to enhance decision-making seem truly boundless.
So next time you don’t know what to decide: just ask your digital twin.
Takeaways
- A digital twin is a virtual computer model that mimics a real-world environment, representing anything from a product prototype to a factory.
- Digital twin technology may be seen as a simulation model, infused with real-time data and logic. Machine learning algorithms and IoT techniques are used to build rich and intelligent representations of reality.
- Digital twins can be used to test a wide array of actions under many scenarios, enhancing real-world decisions by first observing them in the virtual world.
- The technology is one of the contemporary top trends in business, projected to greatly enhance decision-making in the years to come.
Further reading
Forbes (2017). What Is Digital Twin Technology – And Why Is It So Important? https://www.forbes.com/sites/bernardmarr/2017/03/06/what-is-digital-twin-technology-and-why-is-it-so-important/?sh=249c9d3a2e2a
Gartner (2017). Gartner’s Top 10 Technology Trends 2017. https://www.gartner.com/smarterwithgartner/gartners-top-10-technology-trends-2017
General Electric (n.d.). Digital Solutions For Wind Farms. https://www.ge.com/renewableenergy/wind-energy/onshore-wind/digital-wind-farm
IBM (n.d.). What Is A Digital Twin? https://www.ibm.com/topics/what-is-a-digital-twin
Royal Haskoning(n.d.) Digital Twins – Where Physical Meets Digital. https://global.royalhaskoningdhv.com/digital/trends/digital-twin
Twi-Global (n.d). What Is Digital Twin Technology And How Does It Work? https://www.twi-global.com/technical-knowledge/faqs/what-is-digital-twin
Wikipedia (2021). Digital Twin. https://en.wikipedia.org/wiki/Digital_twin