Digital Twin

Digital Twin: Bridging the physical-digital divide

Share this post:

It’s impressive enough just to consider a few of the technologies coming together in next-generation autonomous vehicles—visual radar and image recognition, vehicle-to-vehicle communication, IoT sensor capabilities, and the layers of software that will orchestrate the gradual process of taking our hands off the wheel. Industry estimates predict we’ll see more than 10 million such vehicles on the road by 2020. But creating a car that can safely drive itself is really just an early benchmark for 21st-century automakers to achieve. There’s another technology coming into play that looms as potentially the biggest key to the long-term success of autonomous transportation—as well as dozens of other industries in the years to come. It’s called the digital twin—a capability that has been incubating for more than a decade in academia, IT and R&D, but is only now starting to show its potential in the marketplace.

Digital twin – the virtual doppelganger

In brief, the digital twin is a virtual doppelganger of the real-world thing, or a complex ecosystem of connected things, such as an autonomous car in the middle of rush-hour traffic in Los Angeles. It’s not just a 3D model—it’s a living model in 3D that sees the car as part of a complex technology ecosystem of electronics, navigation, communication and entertainment, collision avoidance, climate control and so on. Engineers can analyze how a car performs not just in its physical environment, under every condition imaginable, but over its entire lifecycle, from an early-stage digital prototype on screen to its last day on the road.

Aviation engineers today can use the digital twin to pinpoint when and under what conditions, or after how many hours of flying a critical part or sensor will fail. Similarly, rocket engineers are putting twins to work to predict and verify performance of the lightest possible materials and payloads, long before they ever perform costly tests on the launch pad. Digital twins are being used not just to model how such physical assets perform, but—longer-term—how ever-more complex systems of assets (and people) will behave together as a whole.

Three emerging technologies that enable digital twin

First is simulation software and tools—the heart of the digital twin—which has come a long way since the primordial days of CAD and 3D design. Simulation tools today can codify, replicate and virtualize the performance of physical products and systems, all based on the hard-wired laws of physics. Digital twins are essentially complex simulations of any number of components in action—aircraft jet engines during takeoff, wind turbines in a shifting breeze—but based on true operational data generated, over long periods of time, by sensors from every critical part.

Second is the enabling force of all those sensors—the Internet of Things—and the rapidly growing volume and breadth of data that can now be captured from any set of physical assets through embedded, ultra-cheap connectivity.

Third is the emerging power of machine learning and predictive analytics in systems such as IBM Watson®, which when set to work on all those data streams, serve as a critical tool of engineers—providing the intelligence and predictive capabilities that were previously accomplished through costly trial and error. Those newfound abilities, which can now be applied to everything from power plants to autonomous vehicles over their entire lifecycles, hold the potential to unlock trillions of dollars in economic value in the coming decades.

The value digital twins can bring

How might that value emerge in a few years with an autonomous vehicle? Engineers will construct a digital twin before they design or build the real thing—enabling dramatic cost savings and accelerating time to market. Designers will collaborate from the outset with operational teams and data analysts to begin gathering different data types to start modeling (and verifying) how their future pride and joy will perform under every condition; how different types of drivers will interact with it; what its vulnerabilities are from a maintenance and breakdown standpoint.

Building the twin might start with the physical components. Engineers would pool data on type of motor, suspension, chassis and aerodynamic body they want to tap into, and the materials they’re built from. Then they would start adding new data layers—such as operational data logs of similar models, or traffic and road data from major cities (to model performance in different climates). Engineers will pile up all that data and start tapping into machine intelligence tools to design and model out their ideal product—long before anything hits the assembly line.

Then comes the promise of the digital twin over the lifecycle of the vehicle. With digital twins, engineering and operations teams can see not just what is happening at any given time, but why. They can speed up simulations in operation and productivity, pinpoint when, why and how breakdowns will occur, and reduce the costs and risks of unplanned downtime. Think about the millions of man-hours—and physical road miles—that will be saved by digital twins in autonomous vehicles, without the need for nonstop physical road testing.

With digital twins, engineers can accurately predict the future performance of a product as it operates within a larger system—a wing on a plane as it travels from San Francisco to London; a rocket engine as it undergoes the violent progression from launch to stage separation; an office building as it manages power, energy and HVAC systems through the course of a day; and extending much further, a driverless car, navigating city streets at rush hour with hundreds of other virtual cars, each one represented by complex sets of very real physical and operating data.

Where else will digital twins and simulation models plot out our futures? Consider the building you may be sitting in right now. Humans have long engineered buildings a little like our bodies—with plumbing that circulates through walls, wires connecting the rooms like nerves, concrete and I-beams providing the skeleton. But until recently, these indispensable bedrocks of the modern world have lacked the “brain”—the layer of machine intelligence that systems such as IBM Watson now provide. For years, it’s been left to humans to manage the lights, power and temperature, to service the elevators and other equipment, monitor security cameras and keep rooms stocked with supplies. Today, IBM Watson and the digital twin promise a dramatic upgrade.

IBM has even begun to build digital twins of physical structures—dynamic, simulated models of the real thing, powered by the massive amounts of data that a single structure generates around the clock, such as physical specs, energy consumption and cost data, equipment parameters and live occupancy data pulled from elevators. The promise of all this is to use digital twins for everything from predictive maintenance and optimized facility management to streamlining workspace design based on the data flows showing how real people truly use the physical space.

The digital twin concept itself may not be new—it was first introduced about 15 years ago. What is new is the explosion of connectivity in the physical world, which, thanks to the IoT, makes the digital twin finally ready for prime time. And not just for cars, but potentially everything in the manmade world.

Learn more about IBM and Digital Twin.

Add Comment
No Comments

Leave a Reply

Your email address will not be published.Required fields are marked *

More Digital Twin Stories
By Matthew Mikell on October 24, 2017

Digital Twin of the Month: Twinning the GP38-2 iron horse

Sept 18 was an epic day in 1830. This was the first head-to-head race of the iron horse and bio-horse; the steam locomotive was battling for acceptance.  While the locomotive of the day, called Tom Thumb, lost largely on mechanical issues, it eventually won the next race.  The locomotive had started on its path to dominate […]

Continue reading