NVIDIA says its next-generation digital twin technology can speed up product design by up to 1,200 times, cutting timelines from weeks or months down to seconds.
The GPU maker just introduced what it calls “real-time physics digital twins” geared toward industries like automotive, aerospace and defense, and manufacturing, where simulations are the backbone of product development. The highly regulated industries rely on digital twins to model and test products before they’re ever built. A digital twin is a virtual replica of an asset that exists or is intended to exist in physical reality.
NVIDIA says real-time digital twins, enabled through its Omniverse Blueprint, transforms how engineers work. With the technology, they can explore countless design iterations, test ideas in real time, and refine their designs at a previously impossible pace.
Omniverse Blueprints combine the key NVIDIA tools:
- CUDA-X libraries for accelerating simulation solvers
- NVIDIA Modulus, a physics-AI framework for training digital models
- RTX rendering for real-time visualization and 3D data interoperability
The blueprint is compatible with software from developers like Altair, Ansys, Cadence, and Siemens. The platforms can use the Omniverse Blueprint for computer-aided engineering (CAE) of digital twins. NVIDIA introduced the blueprint at SC24 in Atlanta, an international conference for high-performance computing, networking, storage, and analysis.
“We built Omniverse so that everything can have a digital twin,” Jensen Huang, NVIDIA founder and CEO, said in a press release. “Omniverse Blueprints are reference pipelines that connect NVIDIA Omniverse with AI technologies, enabling CASE software developers to build groundbreaking digital twin workflows that will transform industrial digitalization, from design and manufacturing to operations, for the world’s largest industries.”
To showcase Omniverse Blueprints, NVIDIA shared renderings from Luminary Cloud, a simulation-as-a-service provider. As one of the first users, Luminary leveraged the platform to create a real-time virtual wind tunnel. With the detailed digital model, engineers can simulate windflow around vehicles and other objects almost instantly.
Luminary’s implementation uses data from traditional computational fluid dynamics (CFD) simulations to train an AI model. According to NVIDIA, the AI can predict airflow behavior with remarkable speed and accuracy, eliminating the need for step-by-step calculations in every simulation. Once trained, the AI handles variations in geometry, like changes to a car’s shape or a seam design, providing instant feedback on how those changes impact performance.
NVIDIA’s RTX-powered renderings bring the data to life. Engineers can visualize the results of airflow simulations in real time using hyper detailed, photorealistic holograms so they can quickly identify issues and optimize designs on the fly.
The Silicon Valley-based NVIDIA has been evangelizing digital twin technology for years. The Omniverse platform, a collaborative environment for 3D design and simulation, is popular among engineers, designers, and manufacturers who use it to test, iterate, and optimize their creations.