Understand the Digital Twin and How to Use It

By John Pritchard, digital engineering business lead, and Leo Kilfoy, director of SaaS and digital design at Rockwell Automation

“Digital twin” wasn’t in anyone’s vocabulary when engineers in NASA’s Apollo program used the first “living model” to investigate the Apollo 13 oxygen tank explosion in the 1960s. However, fast forward to today, and digital twins are engrained in manufacturers’ conversations as they plan and implement digital transformation initiatives.

In short, a digital twin is a living digital replica of a physical object or system. And it’s redefining what’s possible in manufacturing. With a digital twin of a machine, for example, you can:

• Optimize the machine’s design and refine its performance before you order parts or cut steel.
• Test and debug the machine’s controls well before you connect the physical controller and machine at a customer’s site.
• Reimagine training, maintenance and other activities by bringing them into a virtual reality (VR) or augmented reality (AR) environment.

Despite the high level of interest in digital twins, a lot of confusion remains about what a digital twin is and how to use it. By understanding it and the technology driving it, you can make sure you’re investing in the right tools for your applications and using them efficiently to improve design processes, machines and manufacturing operations.

Digital Twin Explained

A digital twin isn’t a static digital model. It changes over time just like its physical counterpart, based on factors such as operating and environmental conditions.

A digital twin can also learn. Using historical operating data, it can predict the health and performance of its physical counterpart over time.

There are two types of digital twins: physics-based and data-driven.

A physics-based digital twin is an analytical model that’s built from first principles, or our understanding of how the physical world works. Sticking with the example of a digital twin of a machine, the twin is typically built during the design stage from a CAD (computer-aided design) model. And by incorporating control or automation code, it can help you understand how the machine will behave in certain situations before commissioning it.

With a physics-based digital twin, you can test, tune and prove a machine before you build it. This can reduce your need to build physical prototypes and shorten your development timeline. A physics-based digital twin can also help you deliver better machines by adjusting elements like speed, orientation and layout until you achieve optimal performance.

Ideally, the digital twin also incorporates other physics, such as product loads and environmental stressors. This can help you predict how the machine will perform in different production scenarios and environments.

A data-driven digital twin is built using historical time-series data from sensors. It’s ideal for maximizing the productivity of operational assets.

For instance, a data-driven digital twin can create predictive models to anticipate failures on a machine before they happen. Technicians can then schedule required maintenance during a planned shutdown. You can even “rewind and replay” production to review issues that occur and help improve asset performance.

Physics-based and data-driven twins aren’t mutually exclusive. You can combine them for a range of purposes.

For example, a data-driven model can help you calibrate a physics-based model to help make sure it’s producing real-life data. A data-driven model can also provide complex inputs to a physical model. This can help you rewind and replay production issues, or test and validate production changes before you make them. One company even runs its orders through these combined models to find the optimal scenarios for its operations.

Physics-based models can also augment data-based models in useful ways. They can train data-based models when insufficient data is available or to augment off-nominal conditions. And they can provide an output, such as a virtual sensor reading, to help inform system behavior.

The Right Tool for You

The growing demand for digital twins has led to an explosion in digital twin tools on the market. Which one is right for you? When evaluating tools, consider these factors.

Do they support how your teams work? The COVID-19 pandemic changed work as we know it. More people are working remotely, but they still expect to be able to collaborate with their teams and access tools they need no matter where they are.

With digital twin tools now available in the cloud, along with design and digital engineering capabilities, teams can access the tools they need no matter where they’re working. Cloud-based tools can also improve collaboration across designers and other employees to increase your speed of innovation.