With the growing emphasis on environmental, social, and governance (ESG) investments and initiatives, manufacturers are increasingly seeking innovative ways to improve energy efficiency and sustainability in their operations, according to the NVIDIA Technology Blog.
Improve operational efficiency with OpenUSD and AI
Wistron, a leading global supplier of information and communications products, has made significant progress in this area. The company developed a digital twin platform and AI-enabled simulation tools to optimize the design, performance, and energy efficiency of test rooms in its new NVIDIA DGX and NVIDIA HGX factories. These developments have the potential to reduce energy consumption by up to 10%.
To achieve this, Wistron’s developers adopted the NVIDIA Omniverse platform and leveraged OpenUSD (Universal Scene Description) to build a digital twin platform. The platform connects to building management systems and IoT hubs to provide real-time data from thousands of physical sensors throughout the facility. These sensors monitor important parameters such as core temperature and air conditioner return temperature.
OpenUSD’s integration facilitates real-time collaboration among remote teams, streamlining facility layout reviews and accelerating the decision-making process in facility planning and operations.
Wistron’s developers also used the open source NVIDIA Modulus framework to integrate physics-based AI models into the digital twin platform. These models help accelerate simulation tasks, improve thermal dynamics, and reduce operational risks, ensuring cooling systems operate optimally even under challenging conditions.
Building a digital twin of your test room
Wistron leveraged OpenUSD to unify its data pipeline and streamline workflows to enable experts to create 3D models. This approach standardizes asset creation and ensures compatibility between different software and simulation tools. John Lu, Wistron’s plant manager, emphasized that OpenUSD provides flexible data modeling that can combine a variety of data and results from various 3D modeling and simulation tools.
For example, OpenUSD facilitated the connection between Wistron’s Digital Twin platform and Autodesk FlexSim simulation software, increasing the team’s ability to simulate, analyze, and experiment with critical manufacturing processes. Developers built a custom extension to pull data and parameters from FlexSim, using the OpenUSD-based Omniverse connector to integrate this data into the digital twin platform.
Accelerate simulations and predict risk with physics-based AI
Wistron’s simulation experts use computational fluid dynamics (CFD) simulation to support the design and management of commissioning test rooms. However, traditional CFD approaches running on general-purpose computing architectures are often shown to be inflexible and resource-intensive.
To address these challenges, Wistron has integrated additional capabilities into its digital twin platform, including:
- Physical Information Neural Networks (PINNs) using NVIDIA Modulus dramatically improve airflow simulation speeds from 15 hours to just 3.6 seconds, a 15,000x improvement.
- AI-powered extensions for high-fidelity visualization and analysis of CFD simulations, helping to minimize cooling system load and reduce operating costs.
- A recommended system for an Automated Storage and Retrieval System (ASRS) to identify optimal test locations and automatically place new supercomputing baseboards in areas with the lowest risk of opening.
These enhancements allow Wistron to approximate the fundamental physics of thermal systems and quickly and accurately predict temperature distribution and thermal behavior within a test room. Now, teams can identify hotspots in a facility and predict core temperatures up to 30 minutes in advance.
Read the latest NVIDIA announcements at COMPUTEX and learn how Wistron adopts NVIDIA technology to build and operate its digital twin.
Image source: Shutterstock
. . .
tag