Nvidia, Utilidata Partner on Software-Defined Smart Grid Chip Development
Nvidia and energy meter software vendor Utilidata are collaborating to make regional electrical grids smarter and more energy-efficient by developing special software-defined smart grid chips that aim to dramatically upgrade the nation’s power-distribution system.
The chips, which will powered by Nvidia’s AI platform, will be embedded in smart meters to enhance grid resiliency and integrate distributed energy resources (DERs) including solar, storage and electric vehicles (EVs).
The smart grid chips are being developed to give future electrical meters the computing capabilities they will need to work on upgraded and modernized electrical grids, according to the partners.
Once completed, the nascent smart grid chips will be tested by the U.S. Department of Energy’s (DOE’s) National Renewable Energy Laboratory (NREL), which will evaluate the software-defined chips to determine if they can scale and commercialize the lab’s Real-Time Optimal Power Flow (RT-OPF) technology. RT-OPF enables highly localized power load control to integrate additional power consumption while ensuring stable and efficient grid operations.
The use of the coming smart grid chips is also expected to help accelerate the transition to a decarbonized grid where pollution from burning non-renewable fuels is reduced to almost nothing.
“A software-defined device has functionality and capabilities that are not static nor limited by the original design, but can adapt to new workloads and demands by updating the software or deploying new applications,” Marc Spieler, head of global energy business development at Nvidia, told EnterpriseAI. “Similarly, a software-defined smart grid chip has the functionality to solve problems in the grid today and the compute for workflows at the grid-edge needed to solve problems that will arise tomorrow.”
To make this happen, the chip will attach an Nvidia Jetson Xavier NX module, made up of an Arm CPU and an Nvidia GPU, to a utility-grade device that samples voltage signals streaming from the grid at a very high resolution, said Spieler.
“By combining a Jetson module with a metrology device, we are able to process the voltage signal and run it through many machine learning, deep learning and physics-based applications in real-time,” he said. “These applications will increase grid resiliency, distributed energy resources (DER) integration and drive value for residential and business consumers of energy.”
In today’s systems, most of the data streaming through the grid-edge is unusable because current smart meters lack the processing power for real-time analysis, said Spieler, which is why a new approach is needed from utilities. “The software-defined smart grid chip provides the compute required to process high resolution voltage signals, detect dangerous anomalies, optimize power flow, allow transactive energy, surgically shed load, help consumers lower their energy bill and provide much needed visibility and flexibility at the grid-edge,” he said.
For software-defined smart grid chips to be successful, they require more than just compute, according to Spieler. “It requires an open, robust software platform for third-party independent software vendors (ISVs) and utilities to develop and deploy new applications securely and at scale,” he said. “As a leader in building open source ecosystems for accelerated computing, Nvidia is uniquely positioned to deliver the hardware and software platform needed.”
The first chips created and made through the partnership will be targeted for residential smart meters and will then be extended to other grid assets, such as transformers, substations and transmission grid lines.
Utilidata is finalizing the design of the chips now and expects to deliver a prototype during Q1 of 2022, with pilot deployments planned for late 2022. The chips have been under development for almost one year.
Utilidata chief technology officer Marissa Hummon told EnterpriseAI that the coming chips will be revolutionary for the industry. “The smart grid chip will offer unparalleled computing power with embedded software that will enable grid operators to manage rapid decarbonization, electrification and more extreme weather,” said Hummon. “Additionally, it will be an easy to deploy [chip technologies] that will reduce redundancies in existing solutions and systems.”
Using a special adapter, the smart grid chips will be able to work with new smart meters that are being rolled out, or with existing smart meters that are already deployed, said Hummon. “Eventually, we hope to expand beyond smart meters and offer this solution to companies that manufacture other devices that operate at the edge of the grid, including smart inverters, EVs and charging stations.”
The new chips will be designed to work specifically at the edge of the grid, which is “far more variable than the rest of the system, making it far more complex from a computation standpoint,” she said. “There are approximately 32,000 measurements in one second of data at the grid edge and that amount of data cannot be transmitted over the communications systems. The GPU allows us to process those measurements locally and in parallel.”
This capability is especially important for enhancing system resiliency, added Hummon. “We will be able to process high resolution measurements that show anomalous power flow in real-time and identify events that could result in outages before power goes out.”
Karl Freund, the founder and principal analyst at Cambrian AI Research, called the work being done by Nvidia and Utilidata on smart grid chip development valuable for utilities and their customers.
“Clearly, adopting Nvidia Jetson and enterprise AI lets Utilidata concentrate on their value-add and leverages the hardware and software Nvidia has built over the years,” said Freund. “Edge AI is about to get real.”
Other energy industry uses for AI have also been getting attention in 2021.
In February, Shell, C3 AI, Microsoft and Baker Hughes announced a partnership to create an Open AI Energy Initiative (OAI) that aims to grow AI use across the energy and process manufacturing industries. The OAI is envisioned by the partners as an open ecosystem of AI technologies that will provide a framework for energy operators, service providers, equipment providers and the software vendors who serve them to create new AI and physics-based models, monitoring, diagnostics and more to help solve critical industry needs, according to the group.
In November, IBM and Amazon Web Services announced a partnership to bring together IBM’s Open Data for Industries software platform and the AWS cloud to help energy companies find ways to solve vexing issues including varying compliance and data residency regulations around the world.