Energy & Utilities

Synchrophasor streams, smart meters, wind farms, batteries, operator actions, and incident memory. ZeptoDB gives grid agents one timeline for live evidence and durable context.

---

Why operators choose ZeptoDB

Sub-cycle latency

272μs queries over millions of rows. Detect grid disturbances within a single AC cycle — not on the next dashboard refresh.

PMU-scale ingestion

Thousands of phasor measurement units at 30–240 samples/sec each, with nanosecond timestamps. Lock-free pipeline, no dropped frames.

Millions of endpoints

Symbol-partitioned column store scales to millions of meters, inverters, and sensors without cardinality penalties.

Parquet historian

Years of telemetry on S3 / GCS / NFS. Query hot and cold transparently for forensic analysis and regulatory reporting.

Incident memory

Store operator notes, prior disturbances, mitigation steps, model calls, and retrieved context beside PMU and asset telemetry.

---

Use cases

Grid stability & wide-area monitoring

• Ingest synchrophasor (PMU / C37.118) streams with nanosecond timestamps
• Real-time ROCOF, phase-angle difference, and oscillation detection with window functions
• ASOF JOIN to align phasors across substations for wide-area analytics

Smart metering & demand response

• Minute-level and sub-second metering for millions of endpoints
• Rolling aggregates per circuit, transformer, or feeder
• Demand-response signal correlation with load response latency

Renewable asset telemetry

• Inverter, turbine, and BMS telemetry at 1–10KHz
• Performance ratio, capacity factor, and curtailment analytics in SQL
• Anomaly detection on vibration and temperature for preventive maintenance

Battery & storage operations

• Cell-level voltage, current, and temperature streams
• State-of-charge and state-of-health rolling estimation
• Thermal runaway early warning with multi-sensor ASOF JOIN

Grid agent workflows

• Retrieve similar disturbances before recommending an operator action
• Pair live PMU changes with prior incidents, mitigation steps, and asset notes
• Cache repeated explanations for common alarms and demand-response patterns
• Replay agent recommendations against the exact telemetry and operator timeline

---

Example: PMU oscillation detection

-- Detect sub-synchronous oscillations by comparing phase angles across substations
SELECT
    p1.ts,
    p1.substation AS sub_a,
    p2.substation AS sub_b,
    p1.phase_angle - p2.phase_angle AS angle_diff,
    ema(abs(p1.phase_angle - p2.phase_angle), 60) AS smoothed_diff
FROM phasors p1
ASOF JOIN phasors p2
    ON p1.ts = p2.ts
   AND p2.substation = 'SUB-B'
WHERE p1.substation = 'SUB-A'
  AND p1.ts > now() - interval '5 minutes'
  AND abs(p1.phase_angle - p2.phase_angle) > 10  -- degrees

---

Deployment options

Environment	Configuration
Substation edge	Single node, ARM Graviton, WAL + local Parquet
Control center	Multi-node cluster, 10/25 GbE, Parquet HDB on NAS
Cloud historian	EKS cluster, Parquet HDB on S3, cross-region replication
Hybrid	Edge ingests + aggregates; cloud keeps long-tail history

See Production Deployment and EKS Cluster Requirements for reference architectures.

---

Compliance & security

• Structured JSON audit logging suitable for NERC CIP / IEC 62443 evidence
• TLS 1.3 and mTLS on all endpoints, including cluster RPC
• RBAC with namespace isolation for operator / analyst / auditor separation
• Signed, offline-validated Enterprise license — no phone-home

See Security for the full posture.