Industries / Industrial equipment and services / IoT and industrial automation / Digital twins
Connecting data
Connecting data means getting readings out of machines and building systems, naming each one so that software knows what it measures, and carrying it to the model.
6
Milestones achieved, from 1996 to 2018.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1996
One standard driver
In the 1990s users of plant software were "trying to break free from the vendor lock-in", in the OPC Foundation's words. In 1995 a task force from Fisher-Rosemount, Rockwell Software, Opto 22, Intellution and Intuitive Technology set out to write a shared specification, named OLE for Process Control because it was built on Microsoft's COM and DCOM software. It "acted like a device driver to enable PLC controllers to deliver live data, alarms and historical data" to any program. The OPC Foundation was incorporated on 22 April 1996, and version 1.0 of OPC Data Access followed in August 1996.
Credit
Al Chisholm (Intellution), David Rehbein (Fisher-Rosemount), Thomas Burke (Rockwell) and the rest of the task force
1999
Short messages over thin links
Arlen Nipper, then of Arcom Control Systems, and Andy Stanford-Clark of IBM "drew up the first version of MQTT at the start of 1999", by the history written by IBM's Ian Craggs. In MQTT, devices publish short messages to a broker, a server that passes each one to every program that has asked for that topic. It was meant to replace the "hundreds of proprietary poll/response protocols" of industrial devices. The standards body OASIS approved MQTT 3.1.1 as an OASIS Standard on 29 October 2014.
Credit
Arlen Nipper; Andy Stanford-Clark; IBM; OASIS
2006
Plant data with its meaning
OPC UA (Unified Architecture) "integrates all the functionality from the existing OPC Classic specifications" into one specification, which includes an information model describing what each value is. The OPC Foundation's history says version 1.0 "becomes available on July 28th, 2006"; its own FAQ says the specification was "first released in 2008". It became the international standard IEC 62541 in October 2011.
Credit
OPC Foundation
2016
Names that software can read
Building control systems label each point (a sensor reading or setting) in their own way, and the Brick paper says building applications "rarely experience widespread adoption due to the lack of a common descriptive schema". In November 2016 Bharathan Balaji and 14 co-authors presented Brick, a schema that describes sensors, equipment and the relationships between them. They tested it on "six diverse buildings across different campuses, comprising 17,700 data points", running "eight complex unmodified applications". Brick is maintained as "an open-source effort".
Credit
Bharathan Balaji, Arka Bhattacharya, Gabriel Fierro and co-authors from universities and companies
2018
A platform its owner skipped
In 2015 GE described engineering models of each customer machine, used "to continuously tune and increase insights", with its Predix platform as the part that "organizes and oversees all the other sections" (Control Global, October 2015). CIO Dive reported in August 2017 that Predix had "technical issues" and that GE's own cloud storage for it "fell flat". In December 2018 GE announced that it would spin off GE Digital. The analyst Josh Greenbaum told TechTarget that GE "struggled and failed to make it a standard -- even inside of GE", and Forrester's Paul Miller called building and hosting the platform in GE's own data centres "one of the biggest mistakes".
LessonA twin platform did not spread to customers while its maker's own divisions were not using it.
Credit
GE; GE Digital
2018
Principles for connected twins
The Gemini Principles, published in 2018 by the Centre for Digital Built Britain at the University of Cambridge, set nine principles for twins of the UK's built environment under three headings: purpose, trust and function. The paper found that "few digital twins at present are connected or share data across organisations, sectors or geographies. Lack of interoperability is a key constraint."
Credit
A. Bolton, M. Enzer, J. Schooling and co-authors; Centre for Digital Built Britain and Digital Framework Task Group
Watch for
A standard building data model
ASHRAE Standard 223P is a proposed semantic model, built on the web standards RDF, SPARQL and SHACL, that would let building software discover a building's systems and set itself up. NIST says 223P "has completed an initial public review and will be published in FY26", the US government year that ended on 30 September 2026. Its page, updated in November 2026, says the team is "waiting for the committee to vote for the second public review".
Achieved whenASHRAE publishes 223P and control systems from at least two makers export models that an outside party checks against it.
Claimed by
NIST and the US Department of Energy, November 2026
A twin builder in Fabric
Microsoft's digital twin builder is "an item in the Real-Time Intelligence workload in Microsoft Fabric" that "creates digital representations of real-world environments to optimize physical operations using data". Its documentation, dated May 2026, says "This feature is in preview", and notes that it is a different product from Microsoft's older Azure Digital Twins service.
Achieved whenit is generally available and a named user publishes measured results with its method.
Claimed by
Microsoft, May 2026
Open
01
Brick models from point lists
A tool that takes a building automation system's export (point names, units, descriptions and a month of trend data) and writes a Brick model, which an engineer then accepts or corrects point by point.
The briefCredit · unclaimed02
One twin on three platforms
A small test twin, such as one air-handling unit with about 20 points, written once in open formats and loaded into several twin platforms, with converters between them and a fixed set of questions that should get the same answers everywhere.
The briefCredit · unclaimed
Open 01 · the brief
Brick models from point lists
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A tool that takes a building automation system's export (point names, units, descriptions and a month of trend data) and writes a Brick model, which an engineer then accepts or corrects point by point.
Why it is still open
Pacific Northwest National Laboratory (PNNL) tested automatic mapping on eight of its own buildings with more than 7,429 point names, checked against expert-made answers. Classifiers scored 28.3%, the Llama 3 model 26.6% and retrieval-augmented prompting 35.6%. The report says "manual tagging remains time-consuming and error prone".
Built with what exists
the Brick schema and the brickschema Python library, with SHACL checks; the Mortar collection of Brick-labelled buildings, which PNNL used for training; features from the trend data (units, range, how a value moves) computed with pandas; a strong language model such as Claude Fable 5.2 to read names and descriptions; Gemini Decision 1, which gives probabilities for fixed-answer questions, to flag the points a person should check (see the Note "A thousand questions, one pass").
First step
Ask a university facilities office for one building's point export and a month of trend logs, and have an engineer label 500 points.
Done looks like
On points from at least ten buildings not used in development, each labelled separately by two engineers, at least 90% of points get the right Brick class. The engineers' review time per 1,000 points and every error are published.
Open 02 · the brief
One twin on three platforms
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A small test twin, such as one air-handling unit with about 20 points, written once in open formats and loaded into several twin platforms, with converters between them and a fixed set of questions that should get the same answers everywhere.
Why it is still open
The Gemini Principles found that "few digital twins at present are connected or share data across organisations" and called the lack of interoperability "a key constraint". Microsoft alone now documents two separate twin products and notes that one "is different from" the other.
Built with what exists
Brick and brickschema; DTDL, the modelling language of Azure Digital Twins; Eclipse BaSyx, an open-source server for Asset Administration Shells; Eclipse Ditto, an open-source twin service; OpenUSD for the 3D view; a coding agent to write the converters.
First step
Write the 20-point air-handling unit in Brick by hand, with its 25 questions and their correct answers.
Done looks like
The same 25 questions, such as "which valve serves this coil" or "what was the supply-air temperature over the last hour", get identical answers on every platform after a round trip, every field lost in conversion is listed, and the converters are published under an open licence.