We may be waiting for the consumer metaverse to fully materialize, but in some pockets of industry, digital replicas have been a thing for decades, enabling businesses to see virtual models of products, tools, and even cities.
Take the aerospace industry, for example. In the business of designing planes, each component and part must be tested and certified for safety before the plane is ready for service, which can be very expensive. As a result, some in the industry have turned to digital twin technology over the last two decades as a way to potentially help lower costs, speed up testing cycles, and model aircraft before building and testing physical prototypes.
To do this sort of in-depth virtual testing, many aerospace companies look to external organizations like the National Institute for Aviation Research (NIAR), the aerospace R&D arm of Wichita State University.
NIAR works with government agencies, eVTOL manufacturers, and commercial aircraft OEMs like Boeing to test parts for compliance with FAA regulations, and with the FAA itself on certification by analysis methodologies for airframe crashworthiness and ditching, according to Gerardo Olivares, senior research scientist and director at NIAR. The industry has outsourced parts of these processes to organizations like NIAR in an effort to lower costs.
Olivares told Emerging Tech Brew that NIAR uses digital twins for flight testing, design, and test safety in devices like pilot seats, and to assist in FAA certification. He said its digital twin tech is developed with the help of Altair, a tech company that specializes in simulation software, among other things.
According to a McKinsey estimate, digital twins “can accelerate time to market by up to 50%,” as companies can spend less time building and testing physical models.
“[Digital twins are] a game changer for the aerospace industry, because now you don’t have to spend many millions of dollars building physical prototypes in order to be able to develop, and then test new aircraft concepts,” Olivares said.
The flight path
The institute, founded in 1985, has worked with some version of digital-twin tech for decades.
In the earlier days of the tech, NIAR used what Olivares called “data-driven” digital twins, which he said “are very good for processes where you have repetitive tasks,” like a plane flying one specific route, and used data from thousands of sensors in aircraft.
Keep up with the innovative tech transforming business
Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.
As data was gathered through precise repetition, NIAR was able to extrapolate and use it through AI and machine learning tools to try and model future outcomes, schedule maintenance programs, and gain a better understanding of when to replace certain parts.
But since 2007, as computing, sensor, and connectivity tech has improved, NIAR has been shifting its operations to what Olivares called “physics-based” digital twins, which are full, high-fidelity models of every part and piece of an aircraft, and which can simulate specific conditions like weather or repeated use or inform potential maintenance issues like when to replace certain parts virtually.
“We can do all the development virtually: Test, fly structures, aerodynamics, and everything so by the time that we start building components, we already have a very high degree of confidence that the system is going to work as intended,” he said. “And the only purpose of the full-scale test is to validate that, in fact, all the simulations that we ran were accurate.”
NIAR can run thousands of virtual missions in a short period of time, amassing 10 to 20 years of real-world testing data within a few weeks, Olivares said.
Olivares said that developing a pilot seat typically takes 24–32 months “from conceptual design to certification,” and can cost up to $3.5 million “between physical prototypes and engineering,” but with digital-twin models, he said that testing and design were done virtually, cost less than $1 million, and took only six months to complete.
Olivares said that NIAR has been building data-driven models since 1985, but emphasized that without modern high-performance computing, digital twins as we know them would simply not be possible.
“You can have good physics-based models that predict things. But if you’re not able to run them quickly and accurately, then it slows down things. So that’s another advantage of the digital world: If you have enough funds to have a big cluster, you can scale things greatly. You can look at thousands of cases [where] the physical testing [would] be impossible from a planning point of view,” he said.
Update: This piece has been updated 12/09/22 to provide additional context for Altair’s technology use by NIAR.