Ansys Accelerates the New Era of Digital Engineering Using NVIDIA Technology

AI-driven and accelerated simulation will bring physics to the next era of industrial digitalization.

Now, Ansys has announced that we are expanding our long-standing collaboration with NVIDIA to include exciting new opportunities in artificial intelligence (AI). Ansys will leverage the full stack of NVIDIA’s accelerated computing technologies to help engineers scale up their use of AI.

We’ll use NVIDIA Modulus to further develop Ansys AI+ modules, which are powered by AI and provide some of the most advanced Ansys solver capabilities. Starting with Ansys semiconductor software products for the electronic design automation (EDA) market, NVIDIA Modulus will empower simulation users with advanced, physics-based machine-learning methods for computational fluid dynamics (CFD), thermal, electromagnetic coupling, and other multiphysics challenges.

In the data center, NVIDIA GPUs such as the NVIDIA Hopper architecture-based H100 Tensor Core GPU dramatically accelerate Ansys solvers, a dozen of which are already supported. Ansys will also utilize NVIDIA GH200 Grace Hopper Superchips that are designed for giant-scale AI and high-performance computing (HPC) applications. This support is a research and development priority across Ansys. For example, Ansys is porting our semiconductor design tools, Ansys Fluent computational fluid dynamics (CFD) software, and Ansys LS-DYNA explicit simulation software to NVIDIA processors, including NVIDIA GH200 Grace Hopper Superchips, which deliver massive performance boosts for applications running terabytes of data. NVIDIA’s latest superchips empower scientists, researchers, and engineers to discover solutions to extremely complex problems faster than ever before.

Ansys SimAI, the artificial intelligence (AI)-based software-as-a-service (SaaS) application that was recently launched, combines the predictive accuracy of Ansys simulation with the speed of generative AI and GPU computing. SimAI enables you to rapidly test design alternatives, around 10x to 100x more depending on the size and complexity, without the constraints of traditional solvers across all design phases. For example, a fairly large simulation like the aerodynamics of the vehicle in the video below could be tested 20x faster with SimAI, enabling you to speed development or try 20 additional design iterations in the same amount of time.

In the application shown in the video, high-fidelity Fluent simulations were accelerated on NVIDIA H100 Tensor Core GPUs. Approximately 50 accurate CFD results, incorporating exterior shape variations and topological changes like rear mirrors, ski racks, and spoilers, were fed into SimAI, and NVIDIA GPUs were used to train the model. The trained model was then deployed via NVIDIA GPUs, illustrating that SimAI achieved less than 0.5% drag error compared to traditional CFD.

Don’t Just Visualize Simulations, Experience Them

Ansys is also in on the next technological era: the 3D internet, which includes technologies like spatial computing, digital twins, and extended reality. You can think of the industrial metaverse as a digital ecosystem — a real-time, 3D, virtual world. The possibilities for industrial sectors, entertainment, finance, and just about everything else are mind-boggling.

The time- and cost-saving benefits of using simulation to predict what will happen before physical assets are built are well-proven. The same is true of simulating systems and processes. But imagine being able to digitally create and walk through a factory before you even break ground. You could simulate workflows and the interaction of each machine via digital twins, or even ensure you will meet sustainability targets.

Now imagine that the factory is a virtual human being and that the “machine” is a 3D digital heart, enabling a surgeon to experience how an upcoming operation will go, predict future health risks, and treat them before symptoms even arise. Or maybe the virtual ecosystem is a city where you can see 6G communication challenges via antenna simulations that take place in real time as cars autonomously navigate around smart city streets. The possibilities are endless. We stand at the precipice of a new era for simulation to solve problems that were previously impossible.

We may be getting a bit ahead of ourselves, but the building blocks of the next era of digitalization, such as accelerated computing, artificial intelligence, and multiphysics simulation, are advancing at incredible speeds. Like the 2D internet, which uses hypertext markup language (HTML) as a standard, an open language and platform for the 3D internet is critical to helping the indistrial metaverse flourish.

Speaking the Language of Collaboration

Ansys has joined the Alliance for OpenUSD (AOUSD) — comprising members such as NVIDIA, Pixar, Apple, Autodesk, Adobe, and others — to help build an open, extensible ecosystem for simulating in 3D. Universal Scene Description (OpenUSD) describes geometric, material, physical, and behavioral representations of 3D worlds. AOUSD is an open, non-profit organization dedicated to promoting the interoperability of 3D content through OpenUSD.

Ansys is bringing its AI-enabled simulation expertise to the Alliance. Developers using the NVIDIA Omniverse platform can benefit from unique Ansys solver capabilities that will run directly within the platform to significantly increase the pace of innovations — from safer autonomous vehicles to improved understanding of the human heart to 6G communications and more — that will benefit everyone.

Work has already begun. Ansys’ AVxcelerate Sensors simulation software has been integrated with NVIDIA DRIVE Sim. The integration will provide high-fidelity sensor simulation outputs generated with AVxcelerate Sensors for development of advanced driver assistance systems and autonomous vehicles.

“Perception is crucial for AV systems, and it requires validation through real-world data for the AI to make smart, safe decisions,” said Walt Hearn, senior vice president of worldwide sales and customer excellence at Ansys in a recent press release. “Combining Ansys AVxcelerate Sensors with NVIDIA DRIVE Sim provides a rich playground for developers to test and validate critical environmental interactions,  to accelerate AV development.” 

Multiple teams at Ansys are working with Omniverse — from LS-DYNA to Ansys STK for digital mission engineering, Ansys RF Channel Modeler for radio frequency system design, and the new Ansys Perceive EM, which was made for synthetic data generation to drive AI/ML applications. Perceive EM introduces wireless channel modeling that captures the behavior of antennas and the signals traveling through a virtual twin of the environment. It provides 4G, 5G, and 6G network architects with the means to simulate — in real time — radio network systems in bespoke urban and rural locations, accounting for materials, as well as motion within the scene.

Ansys customers will benefit from our integration with an expanding set of NVIDIA Omniverse features and technologies such as photorealistic visualization, augmented and virtual reality, generative AI, and massive data manipulation. Ansys will enrich the Omniverse developer ecosystem with unique physics-based solvers that will help unlock cross-industry simulation and digital twin use cases within the platform, including for automotive, aerospace and defense, healthcare, and communications companies.