Digital twins are virtual representations of objects, processes, or environments. They are created by scanning data related to important functional areas of the examined physical entity, analysing it with algorithms, and then transferring it to the digital copy.
The concept may have been first applied to manufacturing in 2002 by Michael Grieves, then a lecturer at the University of Michigan. However, the term “digital twin” was only introduced around eight years later by John Vickers of NASA.
4 out of 10 industrial companies already rely on digital twins
Currently, 4 out of 10 industrial companies already rely on digital twins. The industry has high expectations for this technology. According to a survey by the digital association Bitkom, 63 percent of industrial companies believe that digital twins are indispensable for surviving in international competition. Meanwhile, 44 percent are already using digital twins, 8 percent are planning to do so, and a further 14 percent can imagine it. Only 10 percent cannot envision its use in the future, and one in five industrial companies (20 percent) has not yet explored the technology.
Moreover, 59 percent of German industrial companies assume that digital twins contribute to sustainable production. Almost half (49 percent) believe that digital twins enable completely new business models. Only 17 percent of industrial companies in Germany consider digital twins a passing hype.
Much more than a simple simulation
Digital twins have several key properties that define them. Timo Kistner, NVIDIA’s EMEA Industry Lead for AI for Manufacturing and Industry, counts them as “Representative Ground Truth – a single source of truth for virtual datasets; Physical Accuracy – adheres to the laws of physics; Perfectly Synchronised – precisely timed and perfectly synchronised with the real world; AI-Enabled – optimised by AI and enables training of AI.”
These properties make digital twins far superior to simple simulations, as they can analyze numerous processes simultaneously, allowing for more precise future scenarios.
“As teams simulate possible future scenarios, they need to be confident in the accuracy of their simulations. To achieve this, their simulations should be as close to reality as possible. Therefore, the digital twins they use for their simulations should represent the ground truth of their operations and correspond to the laws of physics,” explains Kistner.
Costly errors can be avoided
The technology brings many advantages, allowing every company in every industry to benefit. For example, digital twins can be used by companies to transform their business, such as reducing costs and waste, increasing quality and performance, or accelerating market launches.
“Digital twins also increase productivity as they enable teams to collaborate in real time to accelerate and improve decision-making,” says the expert. Companies can design, test, and operate new products entirely as digital twins in real time.
“Planning, design, construction, and facility teams use digital twins, for example, to design new products and systems and to monitor, simulate, and optimise production environments. They use the technology to analyze important events or problems that have arisen and better understand, monitor existing processes in real time, and simulate future scenarios,” says Kistner.
The use of AI and learning models makes it possible to simulate complex situations in a highly realistic manner and significantly reduce risks for end-users. Errors and weak points can be identified early.
“Traditionally, product development and testing were a mix of digital and physical steps, often carried out sequentially and comprising a large part of the work between each participant. Similar to early industrial production cycles, a mistake at any point in the process meant that production lines had to be stopped, and perhaps a step backward had to be taken to correct the mistake or even start over. Testing, especially of safety-related products and scenarios, could have a huge impact on the launch of a product or, worse, lead to a recall if problems are discovered post-launch,” explains Kistner.
Digital twins will be indispensable
The applications and possible uses of digital twins are very broad. For example, Digitale Schiene Deutschland (DSD), a part of Deutsche Bahn, is building the first digital twin to simulate the surroundings of railway lines and stations, enabling the full simulation of automatic train operation across the entire network. The digital twin is created based on a high-resolution digital map containing precise measurements of real routes.
“This involves creating a photorealistic and physically accurate replica of the entire rail network, including tracks through cities and landscapes, along with details from sources like platform dimensions and vehicle sensors,” says Kistner.
Using the AI-enabled digital twin created with NVIDIA Omniverse, DSD can develop highly capable perception, prevention, and management systems to detect and respond to irregular situations in daily rail operations optimally.
On the other hand, car manufacturer BMW has been testing production with virtual models. In 2021, the company created the cockpit for the BMW iX using a digital twin. Utilizing artificial intelligence, the production of the support structure for the BMW iX cockpit was simulated as if a real component were being manufactured using an injection molding process. This allowed the entire manufacturing process, including all physical properties, to be digitally simulated. The technology is now firmly established in BMW vehicle factories.
Due to the wide range of possible uses, digital twins will become an indispensable part of business and industry.
“The results are truly transformative, from reducing costs and waste to increasing productivity, improving quality and inventory, and accelerating production and lead times,” concludes Kistner.
