AI is transforming renewables


The CEO of startup Green Gravity prototypes his ambitious vision in a warehouse 60 miles south of Sydney, Australia, and simulates it in NVIDIA Omniverse, a platform for creating 3D workflows and applications.

The concept requires some heavy lifting. Solar and wind power will pull steel blocks weighing up to 30 cars each down shafts taller than a New York skyscraper, storing potential energy that can spin turbines at any time.

Distributed energy network

Swinnerton believes this is the optimal way to conserve renewable energy, as there are nearly a million abandoned mine shafts scattered around the world, many of which are already connected to the grid. And its mechanical system is cheaper and more environmentally friendly than alternatives such as massive lithium batteries, which are better suited to electric vehicles.

We have a huge opportunity to repurpose a million mines,” said Swinnerton, who switched gears after a 20-year career at BHP Group, one of the world's largest mining companies, committed to tackling climate change.

Digital design

A longtime acquaintance saw an opportunity to accelerate Swinnerton's efforts with a digital twin.

"I was intrigued by the idea of ​​Green Gravity and suggested taking a digital approach, using data as a differentiator," said Daniel Keys, IT expert and CEO of xAmplify, a provider of accelerated computing services.

AI-powered simulations could speed up the design and deployment of a new concept, said Keys, who met Swinnerton 25 years ago in one of his first jobs flipping hamburgers at a fast-food stand.

Today they have a digital prototype cooking on the Scaile xAmplify computer, based on NVIDIA DGX systems. It is already accelerating Green Gravity's proof of concept.

“With what we derived from the digital twin, we were able to save 40% of the cost of our physical prototype by going from three weights to two and moving them vertically 10 meters instead of 15 meters,” said Swinnerton.

Use cases permitted by Omniverse

It's the first of many use cases that Green Gravity is developing in the Omniverse.

Once the prototype is complete, the simulation will help scale the design to mines as deep as 7,000 feet, or about six Empire State Buildings stacked on top of each other. Ultimately, the team will build a control panel in the Omniverse to control and monitor the sensor-laden devices without the safety risk of sending a human down the mine.

Virtual visualization tools

Operators will monitor equipment remotely using visualization systems equipped with NVIDIA A40 GPUs and can stream their visuals to tablets thanks to the Omniverse Spatial Framework's TabletAR extension.

The xAmplify workflow uses a number of software components such as NVIDIA Modulus, a framework for physics-based machine learning models.

“We also use Omniverse as a core integration framework that allows us to connect half a dozen third-party tools that operators and developers need, such as Siemens PLM for sensor management and Autodesk for design,” Keys said.

Omniverse makes it easy to integrate third-party applications into a single 3D workflow because it is based on the OpenUSD standard.

In the process, AI sifts through reams of data on thousands of available mines to select optimal locations and predict their potential for energy storage. Machine learning will also help optimize designs for each site.

Together, this is a digital journey that Swinnerton believes will lead to commercial operations for Green Gravity over the next few years.

It is the latest customer for xAmplify's Canberra data center, which serves Australian government agencies, national defense contractors and an expanding group of enterprise users with a range of NVIDIA-accelerated software.

Learn more about how artificial intelligence is transforming renewables, including optimizing wind farms, solar power generation and fusion power.

Post a Comment