Bringing AI to the power grid

Artificial intelligence (AI) enabled servers are one requirement for the adapting needs of the utilities market as substation computing moves to the edge. To meet this need we designed the Crystal Group Energy Series (ES), with the flexibility to integrate the precision and power of NVIDIA RTX GPUs based on the NVIDIA Ampere architecture. ES3604L24 server

Combining our hyperconverged, virtualization-enabled substation servers with the NVIDIA RTX™ A5000 GPU introduces a new array of AI capabilities across a utility’s substation network.

  • Real-time image parallel processing of at least 10 simultaneous 4K streams
  • Enhanced security monitoring with inference capability
  • Predictive substation analytics
  • Substation health monitoring with hypervisor failover capabilities
  • Software-agnostic hardware designed for the demanding requirements of the power industry

“With more than 55,000 substations in the U.S. to upgrade and modernize, we are developing scalable, multi-faceted solutions to enable a more resilient, reliable and secure smart grid,” said Lindsay Palma, business development manager. “Adding the versatility and value of AI into their transformation efforts now equips them to meet today’s immediate needs with the flexibility to accommodate future demands.”

“As utilities add more distributed resources to the grid, and as complexity increases at the substation, engineers are going to need high-performance, edge compute solutions with GPU-level processing speed and accuracy to unlock real-time optimizations at the substation,” said Marc Spieler, head of Global Energy Business Development at NVIDIA. “Integrating NVIDIA GPUs with Crystal Group ES servers provides utilities with the optimal AI platform to protect, manage and secure remote substation installations.”

The GPU-accelerated substation servers are software-agnostic, hyperconverged systems that combine real-time automation, remote management, cybersecurity, auto failover, and zero-trust security features. The NVIDIA RTX A5000 GPUs combine the latest RT Cores, Tensor Cores, and CUDA core technology with 24 GB of graphics memory and a PCIe Gen4 system interface for rapid data transfer. These state-of-the-art visual computing capabilities accelerate deep learning, data science and compute-based workloads.

This comprehensive, ready-now solution supports and simplifies utilities’ current and evolving development needs, including virtual protection relay applications.

Our ES servers and embedded computers are IEC 61850-3-certified and feature industrial design standards for intelligent virtualization platforms. With rugged chassis built and tested to withstand the intense conditions inherent to remote locations—from extreme temperatures, shifting weather conditions, moisture and dust to electromagnetic interference and power transients—the ES line provides long, in-field operational life, seamless operation and reliable accuracy.