Please enable JavaScript.
Coggle requires JavaScript to display documents.
AVEVA PI ↔ Microsoft Fabric (Power Industry) - Coggle Diagram
AVEVA PI ↔ Microsoft Fabric (Power Industry)
Power Use Cases
Fleet Performance & Availability
Asset Health (turbines, boilers, transformers)
Heat Rate / Efficiency KPIs
Emissions & Compliance (NOx, SO2, CO2)
Condition-Based & Predictive Maintenance
Outage Planning & Reliability (EFORd, MTBF/MTTR)
Renewable Forecasting & Curtailment (Wind/Solar)
Market Bidding & Dispatch Optimization
Integration Paths
AVEVA Connect → OneLake
Cloud-native ingestion from PI/Edge
Contextualize to assets; publish to Fabric Lakehouse
Low-code pipelines; API-based
PI Integrator for Business Analytics → ADLS Gen2/Fabric
Publish time-series views (wide/long tables)
Batch or incremental schedule
Write to Parquet/CSV -> Lakehouse bronze
PI Web API / PI SQL Client via Fabric Data Factory/Notebooks
REST/ODBC queries for tags, AF attributes, Event Frames
Load into OneLake (bronze) using pipelines
Delta/Change Data Capture with timestamp filters
Streaming (Kafka/Event Hubs) → Fabric Eventhouse/KQL DB
OMF/Edge to Kafka or directly to Event Hubs
Real-time tables & KQL queries
Near–real-time dashboards & anomaly detection
File Export (CSV/Parquet) → Fabric Ingest
Drop to Blob/ADLS landing zone
Dataflow Gen2 or Notebook ingestion
Good for restricted networks
Data Modeling & Transformations
Tag → Fact table (long-form), quality flags
AF Hierarchy → Dimensions (Plant/Unit/Asset)
Event Frames → Fact Events (trips, starts)
Resampling & Gap-Fill (1m/5m/15m)
Time Zones (UTC vs local), daylight handling
Engineering units & validation rules
Security & Governance
Entra ID (AAD) SSO/OIDC for PI & Fabric
RBAC & Least Privilege across pipelines
Purview lineage & data catalog
PI Trust removal, certificate-based auth
PI-to-cloud network via Self-hosted IR
Phased Roadmap
Phase 0: Discovery – scope assets, KPIs, SLAs
Phase 1: Land – bronze ingestion (Integrator/Connect/DF)
Phase 2: Model – silver/gold, semantic model
Phase 3: Real-time – Eventhouse & alerting
Phase 4: ML/AI – anomaly detection, forecasting
PI Data Sources
PI Data Archive (tags, compression)
PI Asset Framework (AF hierarchy, attributes)
Event Frames (start/stop, trips, alarms)
Interfaces/Connectors: OPC DA/UA, Modbus, DNP3, IEC 61850
SCADA/EMS/DCS Historians
Edge/OMF Publishers
Fabric Targets
OneLake / Lakehouse (Bronze/Silver/Gold)
Warehouse (SQL) for BI/Reporting
Eventhouse / KQL Database (real-time)
Notebooks (PySpark) for engineering & ML
Data Factory Pipelines & Dataflows Gen2
Power BI semantic models & Direct Lake
Power-Specific Calculations
Heat Rate, HHV/LHV adjustments
Net Generation, Auxiliary Load, Station Service
Ramp Rates & Startup Curves (Cold/Warm/Hot)
Transformer Health Indices (DGA, temp)
Reliability KPIs (EFORd, SAIDI/SAIFI where relevant)
Operations & FinOps
Data Quality (stale tags, flatlines, outliers)
Monitoring & Alerting (pipeline failures)
Cost controls: file sizes, partitioning, compaction
Retention & Tiering (hot/warm/cold)
Disaster Recovery & HA patterns