YarifyStart a conversation

Sensor-level data access, without redesigning the historian

Pressure, temperature, vibration and flow sensors already stream raw and pre-processed readings through OPC-UA or MQTT into a plant historian. The Data Act asks for a governed way to hand a scoped slice of that same stream to the plant operator and the maintenance vendor they choose — not a new sensor network.

Readings and status are in scope. Process recipes aren't.

Chapter II covers the sensor's own readings and the metadata needed to interpret them — not what the plant's control system does with that data once it arrives, and not the sensor OEM's cross-plant benchmarking analytics.

Field groupTypical sourceDelivery format
Raw & pre-processed readingsSensor via OPC-UA (IEC 62541) or MQTT brokerTime-series export or streaming API
Calibration metadataSensor configuration recordStructured export alongside each reading batch
Diagnostic/fault codesSensor self-diagnostic, often NAMUR NE107 status classesStructured export with code, timestamp and status class
Sensor health statusDevice heartbeat via gatewayStatus API for connectivity and last-calibration checks

Usually in scope

  • Raw and pre-processed sensor readings
  • Calibration metadata needed to interpret the reading
  • Diagnostic and fault codes generated by the sensor
  • Sensor connectivity and health status

Usually outside Chapter II

  • The plant's process-control setpoints and recipe logic
  • Drift-correction or compensation algorithms
  • Cross-plant benchmarking analytics the sensor OEM runs internally
  • Readings never captured or retained by the historian

The plant operator's right is already live

The Data Act has applied generally since 12 September 2025. Where the sensor or its gateway is itself a product with digital elements, the Cyber Resilience Act applies in parallel to how that new API gets secured.

Operator access-on-request

Provide an authenticated request path for readings, calibration metadata and diagnostics already retained.

Operator-directed third-party sharing

Let a plant operator route sensor data to a predictive-maintenance or analytics vendor of their choice.

Pre-contract data transparency

Disclose what a sensor records, at what resolution and retention, before purchase.

Accessible by-design architecture

New sensor platforms placed after 12 September 2026 should ship the access API built in.

Same access-layer pattern, sensor-level granularity

If you've read our industrial machinery access-layer page, this will look familiar — the difference is the catalog stops at the sensor, not the whole machine.

Today: a closed loop

Product

Industrial sensor

Backend

Plant historian / sensor OEM cloud

  • Raw & pre-processed readings
  • Calibration metadata
  • Diagnostic/fault codes
  • Sensor health status

Only exit

Manufacturer's own app

  • No user-facing export or API
  • No route for the user to authorize a third party
  • No permissions model, so every request is a one-off exception
  • No audit trail if a user or regulator asks what was shared

With a Data Act access layer

Product

Industrial sensor

Backend

Plant historian / sensor OEM cloud

  • Raw & pre-processed readings
  • Calibration metadata
  • Diagnostic/fault codes
  • Sensor health status

New layer

Data Act access layer

  • Field catalog
  • User & device entitlements
  • Consent & grant registry

Customer export & portal

CSV/JSON, self-service, same quality as the manufacturer sees

Third-party API (OAuth)

Purpose-bound grant to a repairer, analytics or energy provider

Audit & compliance log

Request, decision, recipient and delivery evidence for review

What we implement, concretely

Raw sensor-data access, historical export, API keys or OAuth for a third-party maintenance vendor, and an audit layer.

CapabilityWhat it does
Raw sensor-data APIAuthenticated access to current and historical readings per sensor or asset
Historical bulk exportScheduled or on-demand export sized for a predictive-maintenance vendor's ingestion pipeline
API keys & OAuth for third partiesPurpose-bound, revocable access for a maintenance or analytics provider the plant operator names
Diagnostic & health feedStructured fault-code and connectivity stream, mapped to NAMUR NE107 status classes where used
Entitlement & audit layerPlant/operator-aware access control with a reviewable decision log

Start with one sensor family and one maintenance vendor

Prove field-level classification and the trade-secret boundary on one plant before extending the catalog across the sensor estate.

  1. Trace the OPC-UA/MQTT pipeline

    Map every field from sensor through gateway, historian and analytics, and classify raw vs. derived.

  2. Separate readings from control logic

    Confirm which fields are sensor output and which belong to the plant's own process-control system.

  3. Design the operator-facing API

    Define catalog, request, third-party grant and delivery endpoints separate from internal engineering APIs.

  4. Build the decision engine

    Turn catalog, entitlement, trade-secret and privacy checks into explainable allow, partial and refuse outcomes.

  5. Ship export and streaming

    Deliver a historical export and, where feasible, a live feed matching the historian's own refresh cadence.

  6. Test the edge cases

    Run through integrator-operated historians, mixed-OEM sensor fleets and disputed calibration-algorithm fields.

Industrial sensor Data Act FAQ

How is this different from the industrial machinery page?

Scope, mostly. The industrial-machinery page covers whole-machine telemetry — production counts, cycle times, machine-level alarms. This page covers the sensor layer that feeds those systems and many others: individual pressure, temperature, vibration and flow readings that get aggregated into machine-level and plant-level views. A plant operator's request can land at either level, and your catalog should be able to answer both. See the machinery page for the whole-machine version of this same access-layer pattern.

Is calibration data a trade secret?

Rarely the calibration constants themselves — those are usually needed just to interpret the raw reading correctly, which makes them metadata the Act expects you to disclose alongside the reading. The proprietary part, if any, is more likely the compensation algorithm or drift-correction model that consumes calibration data internally, which is a much stronger candidate for the trade-secret process than the calibration record a user needs to make sense of their own readings.

Is a predictive-maintenance vendor an eligible third-party recipient?

Yes, provided the plant operator is the one directing the sharing and the vendor isn't a Digital Markets Act gatekeeper. Article 5 exists precisely for this case: a user pointing their own sensor data at a third-party analytics or maintenance provider they've chosen, for a purpose agreed between them, rather than staying locked into the sensor OEM's own analytics offering.

Do we have to deliver raw sensor streams in real time?

Only where continuous or real-time delivery is both relevant to the request and technically feasible. The Act requires delivery of the same quality available to the holder, but it doesn't create a new duty to build real-time infrastructure that doesn't already exist. If your historian only refreshes hourly, a compliant export can reflect that same cadence rather than inventing sub-second delivery no internal system actually has.

We're a systems integrator, not the sensor manufacturer. Are we the data holder?

Often yes, in practice, if you're the one who receives and can make the combined sensor data available — for example running the plant historian that ingests readings from several sensor OEMs. The Data Act looks at who actually holds and controls the data flow, not only who stamped their name on the sensor housing, so an integrator operating the aggregation layer should expect to carry data-holder obligations for that layer.

Does exposing a new sensor-data API create Cyber Resilience Act obligations?

Potentially, where the sensor or its gateway is itself a product with digital elements under the CRA. A new external API is a security-relevant change to that product's attack surface, and should go through the same vulnerability-handling and secure-development process as the rest of the connected product rather than being treated as a pure data-access feature outside CRA scope.

Send us your sensor fleet and historian setup.

Tell us your sensor types, current OPC-UA/MQTT pipeline and historian, and who's asking for access — a plant operator or a maintenance vendor. We'll scope a pilot that proves one field-level access decision end to end.