Peeling Back the Digital Twin Onion

Unreal Engine provides the visualization layer for data-driven digital twins.

Epic Games has sponsored this post.

Building a digital twin in Unreal Engine. (Image courtesy of Buildmedia.)

Building a digital twin in Unreal Engine. (Image courtesy of Buildmedia.)

“It’s probably the most contentious topic I’ve ever approached in my life,” began David Weir-McCall, a business development manager at Epic Games.

Was he talking about religion? Politics? How to pronounce gif?

Nope. Weir-McCall was talking about digital twins.

Digital Twins: What Are They, Anyways?

Weir-McCall is not the only one to have recognized the conceptual fuzziness of digital twins. Answering the question “what is a digital twin” is easy. Getting everyone else to give the same answer is not. In 2018, several prominent software, hardware and engineering companies banded together to form the Digital Twin Consortium in order to bring consensus to the concept. Slowly, it seems to be working.

“Anyone can get started in the digital twin space if you strip it back to the core elements,” Weir-McCall continued. “If you look at any of the definitions by the Digital Twin Consortium, they talk about a 3D representation of a physical asset.”

Epic Games is no stranger to 3D representation. The company’s core software platform, Unreal Engine, began as a tool for 3D video game development. But it’s not all fun and Epic Games. Like other game engines, Unreal Engine has found massive opportunities in industries beyond gaming. Industries that design, simulate, collaborate and build in 3D. Industries that are scrambling to reap the potential benefits of digital twins.

Architecture, engineering and construction is perhaps the most ready example.

“There’s a sudden explosion of interest in AEC around digital twins,” Weir-McCall said. “I think there’s a larger opportunity within AEC because it affects so much of the built world and how we live our lives. Every potential building in the world could have a digital twin.”

Visualizing building data. (Image courtesy of Lunas Inc.)

Visualizing building data. (Image courtesy of Lunas Inc.)

The Digital Twin Onion

Let’s strip digital twins back to the core elements, as Weir-McCall suggested. As with ogres, digital twins are like onions. They’re layered, and the very core of the digital twin onion is data.

Weir-McCall recalls once hearing digital twins described as “glorified Excel spreadsheets.” It’s a funny comparison, because when you think about it, it’s not that far off. The whole point of the digital twin is to capture all the data pertaining to the physical twin, and leverage it to maintain and optimize that real-world asset. If an engineer was so inclined, she could go visit the asset a dozen times a day, take a few hundred measurements, and keep track of them all in a spreadsheet. A few handy pivot tables and a what-if analysis may even provide some actionable insight.

Of course, no one would be crazy enough to call this a digital twin. The core of the digital twin onion—the data—is not enough on its own. It needs another layer on top, and that’s where Unreal Engine comes into play.

“Our approach is to be the visualization layer,” Weir-McCall explained. “The act of being able to visually represent data and visually tell a story lowers the cognitive load of understanding data.”

The Visualization Layer

Let’s look at an example. The first thing we’ll need is some data, but don’t worry if you don’t have any—as Weir-McCall points out, there are plenty of open data sets just waiting to be played with.

“One of our customers took open weather data and applied wind values to our wind physics in Unreal,” Weir-McCall described. “It was all about analyzing power lines and how they interact with their surrounding environment.”

The company used this digital twin to ascertain key characteristics of their power lines. They could accurately simulate that if the wind speed rose to a certain level, then this or that power line would swing far enough to make contact with a nearby tree. By plugging in the real-world weather data, the company could see problematic power lines in real-time.

“The potentials are endless. And that’s just one data set,” Weir-McCall said.

A digital twin of power lines. (Source: Julian Crespi from Vuela.tv.)

Just one data set—wind information—is enough to create a valuable digital twin. But this is where the digital twin onion starts to grow from a tiny pearl into a full-size Vidalia. There is no rule stating that you are limited to one data set. You could keep wrapping data layers all over your digital twin. For example, the power line digital twin could also include thermal, electrical or mechanical data of the line.

The really interesting part is that each new data layer can have its own visualization layer, and different users of the digital twin could take advantage of each. It would be as easy as switching between a third-person and first-person view in a video game, or a wireframe and shaded view of a CAD model. The underlying data remains the same, but the way that data is visualized creates a whole family of digital twins around it.

“That is the last layer of digital twins which is so exciting, that custom made user story,” extolled Weir-McCall.

And there’s another bonus: while human vision is limited to the reality of our optical system, Unreal Engine is not. Digital twins provide a way for us to visualize the invisible.  

Visualizing the Invisible

Unless you have a case of chromesthesia, you probably don’t know what sound looks like. But it could be very helpful if you did. Imagine you are an urban planner trying to strike a balance between resident comfort and traffic throughput—the noise of a highway, and how it interacts with the surrounding environment, is an important variable.

The right type of digital twin can make this variable visible. A research group from Chalmers University of Technology in Sweden demonstrated just that with a proof-of-concept tool they call DemoVirPEN. Built in Unreal Engine, DemoVirPEN modelled a simple urban environment with cars, pedestrian, and buildings. By simulating the traffic noise, which changes based on speed limits and other variables, DemoVirPEN can add a visualization layer to the data by presenting the sound level as a color-coded spectrum. Such a tool could one day be used to facilitate planning policy.

While the Chalmers team focused on noise in their demonstration, there are plenty more invisible phenomena to be exposed. The same idea could be applied to visualize air quality, electromagnetic radiation, or even—to pick an example completely at random—hotspots of infection by a novel coronavirus.

How to Build Digital Twins in Unreal Engine

This is all well and good, you may be thinking. I’m ready to sauté this digital twin onion. But how?

Weir-McCall has plenty of advice, but he is quick to reiterate that Unreal Engine is made for a very specific layer of the digital twin—visualization. This, he says, is what allows us to contextualize the underlying data, bringing it to life in a way that all users—not just data scientists—can readily understand.

“Getting started relies on you having access to a 3D model, having access to data, and then bringing those two together to build a user experience,” Weir-McCall said. “That’s what Unreal Engine provides, and from there you can do a number of really fascinating things.”

(Image courtesy of Epic Games.)

(Image courtesy of Epic Games.)

Both 3D models and open data sets are freely available online, but enterprise users will likely have some of their own. If not, the data connection is usually the trickiest part. Weir-McCall recommends that anyone interested in digital twins start thinking about what data they want to collect for their own application, and working with the right partners to do so. “We are there to complement a very complex, very structured backend IoT by our partners and by our customers,” he said.

It’s a point Weir-McCall refuses to gloss over.

“I need to drive home that if you’re interested about getting started with digital twins, the first thing that you can do is start owning and collecting your data. That makes the later stages of visualization and interaction so much easier than working with nothing,” he emphasized.

Just like peeling an onion, it’s best to take digital twins one layer at a time. But if you’ve got the data layer down, visualization is where you get to have fun with it. If you’re at that stage, Weir-McCall has written a guide on getting started with digital twins in Unreal Engine.

To learn more about Unreal Engine, visit unrealengine.com.

Written by

Michael Alba

Michael is a senior editor at engineering.com. He covers computer hardware, design software, electronics, and more. Michael holds a degree in Engineering Physics from the University of Alberta.