With the rising emphasis on environmental, social, and governance (ESG) investments and initiatives, producers are more and more in search of progressive methods to boost power effectivity and sustainability of their operations, in keeping with the NVIDIA Technical Weblog.
Enhancing Operational Effectivity with OpenUSD and AI
Wistron, a number one international provider of data and communications merchandise, has made important strides on this space. The corporate has developed a digital twin platform and AI-enabled simulation instruments to optimize the design, efficiency, and power effectivity of their run-in take a look at rooms in new NVIDIA DGX and NVIDIA HGX factories. This development has the potential to cut back power consumption by as much as 10%.
To attain this, Wistron’s builders employed the NVIDIA Omniverse platform, leveraging Common Scene Description (OpenUSD) to construct their Digital Twin platform. This platform connects to their constructing administration system and IoT hub, offering real-time information from 1000’s of bodily sensors throughout their amenities. These sensors monitor essential parameters reminiscent of core temperature and air con return temperature.
The combination of OpenUSD has facilitated real-time collaboration for distant groups, streamlining the evaluate of facility layouts and accelerating decision-making processes in facility planning and operations.
Moreover, Wistron’s builders have built-in physics-informed AI fashions into their Digital Twin platform utilizing the open-source NVIDIA Modulus framework. These fashions assist speed up simulation work, enhance thermal dynamics, and scale back operational dangers, guaranteeing that cooling programs carry out optimally even beneath demanding circumstances.
Constructing Digital Twins of Check Rooms
Wistron has utilized OpenUSD to unify their information pipeline and streamline workflows, enabling the creation of 3D fashions by their consultants. This method standardizes asset creation and ensures compatibility throughout varied software program and simulation instruments. John Lu, plant supervisor at Wistron, highlighted that OpenUSD provides versatile information modeling, permitting the mixture of numerous information and outcomes from varied 3D modeling and simulation instruments.
For instance, OpenUSD has facilitated the connection between Wistron’s Digital Twin platform and the Autodesk FlexSim simulation software program, enhancing their groups’ potential to simulate, analyze, and experiment with essential manufacturing processes. Builders have constructed a customized extension to import information and parameters from FlexSim, utilizing an OpenUSD-native Omniverse Connector to combine this information into their Digital Twin platform.
Accelerating Simulation and Predicting Danger with Physics-Knowledgeable AI
Wistron’s simulation consultants depend on computational fluid dynamics (CFD) simulations to assist the design and administration of their run-in take a look at rooms. Nonetheless, the standard CFD approaches, working on general-purpose computing architectures, typically proved rigid and resource-intensive.
To deal with these challenges, Wistron built-in further capabilities into their Digital Twin platform, together with:
- Physics-informed neural networks (PINNs) utilizing NVIDIA Modulus, which considerably hurries up airflow simulations from 15 hours to only 3.6 seconds—a 15,000X enchancment.
- An AI-based extension for high-fidelity visualization and evaluation of CFD simulations, serving to to reduce cooling system masses and scale back working prices.
- A advice system for his or her automated storage and retrieval system (ASRS) to determine optimum testing areas and autonomously place new supercomputing baseboards in areas with the least thermal threat.
These enhancements allow Wistron to approximate the underlying physics of their thermal programs, offering quick and correct predictions of temperature distributions and thermal behaviors inside their run-in take a look at rooms. Groups can now determine facility hotspots and forecast core temperatures as much as half-hour upfront.
Learn the most recent NVIDIA announcement at COMPUTEX to learn the way Wistron is adopting NVIDIA know-how to construct and function digital twins.
Picture supply: Shutterstock
. . .
Tags
