Theory — the brief
The company
NorthWind Renewables (fictional) runs 600 turbines across 8 wind farms on three continents. Three SCADA vendors, two CMMS systems, one data lake nobody trusts.
The pain
Reliability engineer: "Vendor A calls the part gearbox. Vendor B calls it transmission. Vendor C uses a 7-digit code. My alarm rule has to enumerate all three."
Data lake owner: "Every new turbine model brings new tag names. My dashboards break every quarter."
Asset manager: "I want to ask 'how many gearboxes with vibration > X are installed in farms commissioned before 2020?'. Today that's a two-week consulting engagement."
What's different from MedaCore
Two big shifts shape every later decision:
- Telemetry volume. Tens of thousands of observations per minute. We cannot afford OWL 2 DL's exponential worst-case reasoning. OWL 2 EL — polynomial — is the right profile.
- Mature reusable standards exist. SOSA/SSN (W3C) for sensors and observations, QUDT for units. Minting our own
:Sensorclass would be unprofessional. The job is reuse-first.