Build a Digital Twin for Interactive Fluid Simulation

Experience a Real-time Wind Tunnel

This experience shows an interactive virtual wind tunnel. Simulating airflow in a virtual wind tunnel with computational fluid dynamics (CFD) requires millions of complex calculations. Without AI, it can take minutes to see the result of a single design change. Developers building the next generation of AI-powered CAE tools are combining simulation AI with immersive virtual environments to enable real-time digital twins where design changes instantly update in the simulation, as you see in this demonstration. NVIDIA is making it easier to create interactive design tools by introducing Omniverse Blueprint for interactive aerodynamics.

To enable real-time performance in a virtual wind tunnel, simulation AI models are first trained offline on representative datasets. For this demo, the training dataset was created using Luminary Cloud’s GPU-accelerated CFD solver, which models complex airflow over diverse geometries. The simulation AI learns the complex relationships between car geometries (STLs) and airflow. This blueprint is compatible with industry-standard CFD solvers and can connect to third-party tools for meshing and geometry morphing, creating watertight meshes for simulation-ready geometries.

The blueprint also integrates NVIDIA PhysicsNeMo (our framework for simulation AI) with CFD solver data. This enables developers to train surrogate models from scratch or fine-tune NIM™ foundation models, reducing AI training time and cost. Once trained, the AI runs simulations orders of magnitude faster than traditional CFD, enabling real-time aerodynamic flow simulation. This speed provides designers with creative freedom, allowing designers to innovate and explore changes interactively.

Architecture Diagram

What’s Included in the Blueprint

NVIDIA Blueprints are comprehensive reference workflows designed to streamline AI application development across industries and accelerate deployment to production. Built with NVIDIA AI and Omniverse libraries, SDKs, and microservices, they provide a foundation for custom AI solutions. Each blueprint includes reference code for constructing workflows, tools, and documentation for deployment and customization, and a reference architecture outlining API definitions and microservice interoperability. By enabling rapid prototyping and speeding time to deployment, these blueprints empower enterprises to operationalize AI-driven solutions like AI agents, digital twins, and synthetic data generation, and more.

Included NIM Microservices

• DoMINO-Automotive-Aero 1.0

World State Controller

This reference solution implements an Omniverse Kit application controller that maintains the application world state and connects the 3D world stage with simulation results produced by the surrogate model.

In summary, this NVIDIA Omniverse Blueprint offers a starting point for building real-time digital twins for computer-aided engineering (CAE) workflows combining CUDA-X™ accelerated solvers, NVIDIA PhysicsNeMo for simulation AI, and Omniverse for high-quality rendering.

Minimum System Requirements

Hardware Requirements

The real-time wind tunnel blueprint supports the following hardware:

• At least 2x RTX™ GPUs with at least 40GB of memory each, e.g., 2xL40S or 2xA6000
• 128GB RAM
• 32 CPU Cores
• 100 GB Storage

OS Requirements

• Linux - Ubuntu 22.04 or 24.04

Software Requirements

• Git: For version control and repository management.
• Git Large File System (LFS): For large files that are too large to efficiently store in a Git repository.
• Python 3: For scripting and automation.
• Docker: For containerized development and deployment. Ensure non-root users have Docker permissions.
• NVIDIA Container Toolkit: For GPU-accelerated containerized development and deployment. Installation and configuring docker steps are required.

Ethical Considerations

NVIDIA believes Trustworthy AI is a shared responsibility, and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their supporting model team to ensure the models meet requirements for the relevant industry and use case and address unforeseen product misuse. For more detailed information on ethical considerations for the models, please see the Model Card++ Explainability, Bias, Safety & Security, and Privacy Subcards. Please report security vulnerabilities or NVIDIA AI concerns here.

Terms of Use

GOVERNING TERMS: The software and materials are governed by the NVIDIA Software License Agreement (found at https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-software-license-agreement/) and the Product-Specific Terms for NVIDIA Omniverse (found at NVIDIA Agreements | Enterprise Software | Product Specific Terms for Omniverse).