When we talk about a digital twin, we are referring to the digital representation of a unique thing that exists in reality. Digital twins leverage the ‘twin’ concept. While they are not exactly physically identical, they represent characteristics and appearance.
The concept of using a representation as a means of research and simulation is not new. Physical models, prototypes, and even digital simulations have been used for a long time in product and process.
However, today’s technological evolutions go way beyond using representations for product engineering use. The ways that we spoke about this in the past are no longer sufficient. Models can include many more parameters than they formerly could and can be applied to mirror and represent unique individual physical objects. This evolution has allowed the emergence of the unique digital twin.
Image courtesy: Pixabay
A digital representation: past, present & predictive
Let’s take the example of a digital twin of an individual production machine like a press.
For this press, the digital model contains the history of its past: the machine’s historical data and genealogy, where it was made, by which manufacturer, delivery date, etc. It then tracks the lifecycle history of the press: historical performance, what repairs were executed, and possible relocations and modifications.
Through the Internet of Things (IoT) and live connectivity with the physical object, the digital twin mirrors the current machine state. It represents current machine settings like position of levers and dials. And it will – on an ongoing basis – capture and represent machine readings like temperature, pressure, vibration, stamp count, and more.
Including the machine characteristics and modeling the machine behavior lets the digital twin predict, with some accuracy, the machine’s future. A model, potentially augmented by artificial intelligence (AI) and machine learning (ML), can use temperature, pressure and vibration readings to predict quality of the product produced by the machine or even predict machine failure. These insights can be used for various simulations – to improve output, product quality, and predictive maintenance.
The challenge with huge amounts of data is how to make them easily accessible. Modern visualization techniques allow representing the machine in 3D from different angles. The information can be overlaid on the machine representation to show context, making it more intuitively available. Advanced representations like virtual reality or augmented reality further contribute to this intuitive access.
Technology drives the emergence of digital twins under the Industry 4.0 umbrella
Digital twins are a culmination of the use and the combination of multiple technologies including:
- Sensors, whether built in by default in new equipment or as an add-on to older equipment become ubiquitous.
- IoT, and its rapid adoption, allows sharing and centralizing the gathering of plenty of data points.
- Data lakes, which allow for the capturing the huge volume and variety of data in an unstructured way that can still be consumed by visualization and machine learning tools.
- Machine learning, which leverages the data for advanced simulations, for predictive alerts and automated recommendations.
- Advanced presentation and visualization technologies, which are crucial to present the digital twin, its history, its current state and recommendations in an intuitive and accessible manner.
- Augmented reality, which allows overlaying the digital twin on the image of its real life twin – directly highlighting attention points or relevant data on the image of the physical object.
Different purposes of a digital twin
A digital twin can be leveraged for different purposes, and possibly a combination of purposes, many which are critical to the future of electronics manufacturing and the supply chain.
In its most traditional form it is used as a registration / observation model. In this case digital twins are shadowing their real-world counterparts, i.e. monitoring temperature, pressure and vibration of a machine. When these values go beyond an acceptable range it triggers a maintenance intervention. While troubleshooting, the maintenance technician has full access to the stream of historical data, which helps him to assess the situation.
Combined with the observation model, a digital twin can also serve as a user interface for driving / operating its real-world counterpart. Virtual switches on the digital twin can be flipped and they will steer the behavior of the actual equipment. A smart thermostat can be controlled remotely to set the temperature. Lights can be switched on and off by interacting with a floor plan layout of the building.
Finally. the digital twin can be used for simulations and optimizations. The simulation model will use the key parameters of the object. With the use of Machine Learning these models become dynamic and self-learning. The simulations can also be used for training – where the real-world object is either too costly or too dangerous to be used in real training.
A digital twin is a computerized simulation of a physical asset. It represents historical, current and predictive views of its real-world counterpart. It is highly dependent on a stream of data, often supported by sensors and IoT. Further, it leverages advanced visualization.
Digital twins are used for monitoring and registration, as user interface and for simulations. They increase efficiency in the interaction with a physical asset, improve operations and provide support for business-improving simulations.
The concept of digital twins is identified as an important transformer under the Industry 4.0 umbrella. Which is all about how people, interconnected sensors, machines and artificial intelligence can work together more effectively.