What is the difference between BIM and a digital twin?

A supplier shows a three-dimensional model of your building, its rooms shaded from blue to red by live temperature readings, and calls it a digital twin. It may be a useful dashboard, but it does not tell you whether you need BIM, a twin or neither. The short answer is that BIM is the managed body of information about a building, from design through to operation, while a digital twin connects part of that information to live data from the building to help someone make a specific operating decision. Settle which decision matters first, and the right purchase follows.

BIM is information, not a 3D model

The IMI Framework (formerly the UK BIM Framework) treats BIM under the ISO 19650 series as information management across design, construction, operation and maintenance. Geometry is only part of that information, and BIM does not stop at handover: the asset information the operator receives is part of it, and has to be kept current.

What makes that information useful in operation is fit, not detail. A technician who has to find a fire damper and its access hatch gains little from a precise model if the damper's tag, location and maintenance record are wrong. Asking for every property the software can hold has the opposite cost: more to produce, more to check, and little that anyone uses.

"Digital twin" needs defining on every project

The RIBA Smart Building Overlay to the Plan of Work points out that there is no official definition of a digital twin. For its own purposes it separates an asset twin, a digital representation of the building and its systems usually based on the asset information model, from a performance twin built on the data the building generates. That distinction is a good starting point for a brief.

Each project still has to say what it means by the term: which equipment and spaces are represented, which live data stays connected, and which decision the system supports. Without that, owner and supplier end up assessing different deliverables. Measured against such a brief, the coloured model in the opening paragraph is quickly placed. It displays readings, but it predicts nothing, raises no work order and does not on its own make a business case.

The two side by side

An operational twin of the same unit adds the live picture. It shows the supply-air temperature climbing from 14 °C to 17 °C after 2 pm while the cooling valve stays fully open, and it marks the gap between that and the behaviour expected for the day's weather. The engineer sees the pattern in context instead of trawling through trend logs. The twin still cannot say why. Someone has to go up to the roof and establish whether the coil is fouled, the valve actuator has failed or the chilled water is arriving too warm.

If the engineer cannot find the right unit and its manuals, better asset information comes first. A twin starts to earn its cost when the team has recurring problems to investigate and the time to act on what it shows.

From BIM information to a working twin

Keep the first step small: one operating question, the assets it involves and an honest check of the information that already exists. Equipment tags must match across the model, the controls and the maintenance system: a model whose objects cannot be matched to real equipment has little operational value, however detailed it is. Connect only the readings the first use case needs, test a few representative cases with the operations team, and name who maintains the mappings when equipment changes.

IFC, the open exchange schema published as ISO 16739-1, can carry model data between applications, so ask suppliers which model view definition their software supports. An IFC file is still a file. It is not a live connection, an analytics service or a guarantee that every property your use case needs will survive the exchange.

How often should a twin update?

As often as the decision requires, and no more. A monthly performance review and an operational alarm need very different data services. Ask suppliers to state the sampling interval, the transfer delay, how stale data is flagged and how long history is kept; "real time" is not an acceptance criterion without a stated tolerance.

Letting the twin control equipment is a separate decision. For a first investigation tool, recommendations reviewed by an engineer may well be enough. If the system is to write commands, its authority, limits, approval steps, rollback and responsibilities have to be defined explicitly. Calling something a twin grants it no permission to control anything.

Can an existing building have a twin without BIM?

For many use cases, yes. The test is whether verified equipment registers, surveys, system relationships and operational data can support the first use case. A detailed whole-building model may be justified where spatial information is essential to the decision, but establish that before rebuilding the model becomes the programme's first major cost.

Six questions for any proposal