What Is a Digital Twin?

Digital twins have become indispensable for increasing production efficiency in the manufacturing industry. One example of a digital twin in manufacturing is creating a virtual replica of a production line in an automotive factory.

Each component and machine in the production line will be fitted with sensors to collect data on performance metrics, such as temperature and operational speed. This data is then fed into a digital twin, to build a digital representation of the whole production line.

Engineers can use the digital twin to monitor the real production line’s performance, simulate potential scenarios, and predict mechanical or maintenance issues. For example, you can use the digital twin to simulate changing the order of the production line, to see how it would affect operational efficiency and staff downtime, before implementing the changes in the real world.

Digital twins were first discussed in David Gelernter’s 1991 book, Mirror Worlds. But it was in 2002 that Dr. Michael Grieves, who was on faculty at the University of Michigan at the time, formally introduced and announced the digital twin software concept at a Society of Manufacturing Engineers conference in Troy, Michigan.

Several years later, in 2010, NASA’s John Vickers gave the concept its name, "digital twin," in a 2010 Roadmap Report. Beyond coining its name, NASA is widely recognized for pioneering use of the technology during its space missions in the 1960s. Digital twins were used to create, test, and study virtual simulations of spacecraft.

Since then, the utility of digital twins has continued to grow its profile. In 2018, Gartner spotlighted digital twins as one of the top 10 strategic technology trends for 2019. Over time, the technology has proven to be one of the most dynamic process mining tools (process mining is a data-driven technique used to understand, track, and improve processes by analyzing data from information systems to see how a process works).

Although the terms “digital twin” and “virtual twin” are often used interchangeably, there are notable differences between the two.

A digital twin is a real-time, digital replica of a physical object or system. Sensors are placed on the physical objects, and by gathering their performance data, you can monitor and simulate their behavior. On the other hand, a virtual twin is a static model of a real-world product, asset, or process. The model is often based on theoretical scenarios, used from design through to testing.

There are two key differences between the technologies:

1. Application: Digital twins provide real-time data updates to enable continuous monitoring and real-world analysis. In contrast, virtual twins work best for hypothetical scenarios, to simulate potential outcomes.
2. Interaction: Digital twins perform well with minimal manual interference, enabling you to passively receive and monitor its data. Whereas, virtual twins require more active involvement to adjust parameters or even directly engage with the virtual environment.