Engineering & Industry 4 min readORG-F21

OPC UA Information Models: Make Multivendor PLC Data Understandable

Share

Connecting multiple PLCs through OPC UA does not by itself create a common meaning. An information model turns vendor-specific addresses into governed objects, types, relationships, units, states and semantic identifiers so applications and people can interpret the same machine or process concept consistently.

The short answer

Begin with use cases and stable domain concepts, not a flat tag export. Reuse the OPC UA base model and applicable Companion Specifications before creating custom types. Define namespace ownership and versioning, ObjectTypes, VariableTypes, DataTypes, references, engineering units, ranges, states, quality and optionality. Map each vendor source explicitly to the semantic node and retain source address, transformation, timestamp and quality rules. Publish machine-readable model artifacts where appropriate, test browse paths and type conformance with representative clients, and control compatible versus breaking changes across server, gateway and consuming applications.

  • Model domain meaning independently from one controller address or vendor naming convention.
  • Reuse standard types and Companion Specifications where they fit the actual use case.
  • Give every namespace and type a version owner and change policy.
  • Test semantics, units, state behavior and quality—not only successful reads.
A common connection becomes a common meaning only through an explicit model and mapping.
Text-free OPC UA semantic stack from multivendor PLC addresses through governed mappings, types, instances and consuming applications

A common connection becomes a common meaning only through an explicit model and mapping.

Decision guide

LayerPurposeEvidence
Source mappingConnect PLC address and raw representation to the governed conceptMapping table, conversion, time and quality rules
Type modelDefine reusable assets, variables, methods, events and relationshipsNodeSet/type definition and modelling decisions
Instance modelRepresent this line, machine and equipment hierarchyBrowse paths, identifiers and instance conformance test
LifecycleKeep producers and consumers compatible through changeNamespace version, change log, compatibility and migration test

A practical workflow

  1. 01

    Define consumers, questions, commands prohibited or allowed, latency and lifecycle requirements.

  2. 02

    Inventory source data and identify stable domain entities, properties, states and relationships.

  3. 03

    Select base and Companion models, then design only the necessary custom namespace and types.

  4. 04

    Build governed mappings with units, conversions, timestamps, quality and missing-data behavior.

  5. 05

    Validate with representative servers and clients; publish version, owner, changes and migration evidence.

Model governance continues after the first successful client connection.
Text-free information-model lifecycle covering namespace ownership, reuse, mapping, conformance, versioning and consumer migration

Model governance continues after the first successful client connection.

Evidence and acceptance

Keep use cases, system and responsibility boundary, namespace URI/version/publication date, reused specifications, modelling rules, type and instance diagrams, NodeSet or equivalent artifact, source-to-node mapping, units and conversions, identifier policy, timestamp/quality behavior, access constraints, client/server versions, conformance cases, exceptions, change log, approval, deployment baseline and consumer migration record.

Common mistakes

  • Mirroring PLC tags into folders and calling the result an information model.
  • Creating custom types where a suitable standard or Companion model already exists.
  • Using display names as stable identifiers or embedding one vendor’s addresses in consuming applications.
  • Changing namespace semantics without versioning and consumer impact analysis.
  • Testing that values are readable but not their units, states, quality and relationships.

Responsibility boundary

Primary sources

Technical claims and product boundaries were checked against these primary sources. Fact-checked

  1. OPC Unified Architecture Part 3: Address Space ModelOPC Foundation
  2. OPC Unified Architecture Part 5: Information ModelOPC Foundation

Frequently asked questions

Is an OPC UA tag tree an information model?

Not necessarily. A useful model defines types, relationships, semantics, units, identity and lifecycle beyond a browsable list.

Should every project create a custom namespace?

Only when the required concepts are not adequately covered by the base model or an applicable Companion Specification.

Can the same model hide every vendor difference?

It can normalize agreed meaning, but source capability, timing, quality and behavior still need explicit mappings and exceptions.

Does AutoNexit host the OPC UA model?

No. It can preserve the controlled engineering record, while OPC UA products serve and consume the model.

Related articles

ORG-F15

OPC UA vs MQTT: Which Belongs Where in an Industrial Data Architecture?

OPC UA and MQTT solve different parts of an industrial information problem. OPC UA defines interoperable services, information models, security, and client-server or PubSub behavior; MQTT is a lightweight publish-subscribe transport through a broker. Many architectures use both, but only when semantics, topic governance, security, quality, and ownership are explicit.

ORG-F20

Industrial Edge Gateway Selection: Interfaces, Workloads, Security, and Handover

Selecting an industrial edge gateway begins with data sources, destinations, workload, latency, availability, offline behavior, environmental limits, ownership and support—not CPU and protocol logos alone. The accepted system needs a documented security boundary, buffered-data policy, maintainable software baseline and tested replacement procedure.