Back to Programs

National Energy Market — Grid Intelligence

Context

National-scale energy market operator governing electricity dispatch and gas markets across multiple market regions. Systems underpin real-time grid operations where availability is a public safety obligation and architectural decisions carry regulatory and operational consequence. Details abstracted in line with sensitivity requirements.

Role and mandate

Sole architect with full-stack ownership across cloud infrastructure, OT/SCADA security, on-premises systems, and cyber governance. Responsible for designing and delivering agent-based Grid Intelligence architecture as a principal programme of work.

Architecture problem

Introduce AI/ML-driven anomaly detection and agent-based intelligence into OT field networks without compromising determinism, control plane integrity, or real-time operational throughput. Integrate with Azure and Edge cloud infrastructure while maintaining OT/IT boundary discipline and compliance with critical infrastructure obligations.

Constraints

  • SOCI Act critical infrastructure obligations and ASD Essential 8 compliance requirements.
  • IRAP and NIST CSF alignment across all market systems and grid operations infrastructure.
  • Real-time availability constraints — degraded operation is not acceptable for national energy dispatch.
  • OT/IT boundary integrity — AI/ML integration must not introduce unsafe control paths or non-deterministic failure modes.
  • Sovereign data and operational continuity requirements for cloud-connected components.

Decisions and trade-offs

  • Agent-based architecture scoped to observation and anomaly detection — no control plane write access from intelligence layer.
  • Azure IoT Edge deployed at field network boundaries — local inference with selective cloud telemetry to preserve latency and sovereignty requirements.
  • IEC 62443 zone and conduit model applied to AI/ML integration points — treating intelligence pipelines as new attack surfaces requiring explicit boundary controls.
  • Purdue model assessed and adapted for cloud-connected OT — hybrid architecture acknowledging where the traditional model breaks down at scale.
  • Security zoning designed for assured degradation — loss of intelligence capability must not impair core grid operations.

Outcomes

  • Agent-based Grid Intelligence architecture in delivery across national energy market systems.
  • OT/IT boundary integrity maintained through explicit architectural controls on AI/ML integration paths.
  • Improved anomaly detection capability across field networks without operational throughput impact.
  • Compliance architecture aligned to SOCI Act, ASD Essential 8, IRAP, and NIST CSF across all markets.
  • Reusable architectural patterns for cloud-connected OT security applicable beyond the immediate programme.