Initiative

GridCRED

Trusted data. Repeatable analysis. Better grid decisions.

By uniting data, models, and computational tools in a secure and consistent framework, GridCRED empowers public agencies, utilities, and grid operators to analyze performance, plan infrastructure investments, and coordinate long-term strategies.

GridCRED is a power system modeling and analysis platform under development at Pacific Northwest National Laboratory. It brings data, models, analytical tools, and computing resources together in a governed environment so teams can conduct consistent, scenario-based analysis and collaborate without rebuilding workflows for each project.

A shared foundation for grid analysis

Grid planners and decision-makers often work across disconnected datasets, tools, and workflows. GridCRED is designed to reduce that fragmentation by providing persistent access to data, integrated modeling tools, and scalable computing resources. 

Save. Seek. Retrieve. Share.

Save data in a secure, governed environment. Seek trusted datasets through a centralized catalog. Retrieve tools and computing resources for operational, planning, and future-state analyses. Share data, workflows, assumptions, and results with approved collaborators.

GridCRED can support work involving resource adequacy, load growth, transmission planning, market uncertainty, emergency preparedness, and related grid research. Current access is focused on PNNL teams and selected collaborators.

Platform Requirements

  1. Data catalog and management system for multiple petabytes of data
  2. Strong support for proprietary and/or business sensitive data
  3. Some form of external user access
  4. Some form of cloud presence
  5. Ability to run industry standard and research simulation tools
  6. Governance board to establish management policies around data use

Platform Architecture

GridCRED Scalable Infrastructure Diagram

Draft Computation Specification

icons for CPU, GPU and storage servers with requirements; see text below image for details

CPU Server x25

  • Dual 3.65 GHz, 48 cores
  • 2.3 TB of RAM
  • 15 TB of NVMe storage
  • 100 Gb of Ethernet

GPU Server x4

  • Dual 3.65 GHz, 52 cores
  • 3 TB of RAM
  • 20 TB of NVMe storage
  • 8 NVIDIA H200 GPUs
  • 1.1 TB of HBM3e RAM (each)
  • 100 Gb of Ethernet

Storage Server x3

  • Main, hot spare, backup
  • ~4 PB of storage each
  • 100 Gb of Ethernet

Put GridCRED to work

Bring the GridCRED team a use case, dataset, or analytical challenge. Together, we can explore whether the platform is a good fit and identify potential next steps.