A digital twin is a digital representation of a real asset, process or system that is connected to it: data from the physical asset (sensors, maintenance records, operations systems) flows into the digital version, and the digital version is used to understand, predict and improve how the asset performs. What separates a twin from a model is this connection and the purpose it serves.
At a glance
| Built from | Asset information (the AIM), geometry, and live or regular data feeds |
| Data sources | Building management systems, IoT sensors, CAFM, meters, inspections |
| Used for | Monitoring, maintenance planning, energy optimisation, space use, scenario testing |
| Scale | One piece of equipment, a building, a campus, a network or a city |
| Related ideas | Asset information model, smart buildings, national digital twin programmes |
Model, shadow, twin
| Term | Connection to the real asset |
|---|---|
| Model | None: a description made at a point in time, such as a design model |
| Digital shadow | One way: data from the asset updates the digital version automatically |
| Digital twin | Two way, or at least used to act on the asset: decisions taken from the digital version change how the asset is run |
The terms are used loosely in industry; it helps to ask what data flows, how often, and what decisions it supports.
What a twin needs
- A purpose: a twin built "because we can" rarely survives. Good purposes are specific: reduce energy use, predict pump failures, plan cleaning by occupancy.
- Reliable asset information: identities, locations and relationships of the assets, which is what a good AIM provides.
- Data connections that are maintained when systems change.
- Ownership: someone responsible for keeping it current.
Where BIM fits
BIM delivery creates much of what a twin needs: a structured inventory of assets, their locations and systems, and consistent identifiers. A record model or AIM that follows the asset information requirements is a strong starting point. A design model with no asset data, or with identifiers that do not match the operator's systems, is not.
Common misunderstandings
- "Our as-built model is a digital twin." Without connection to operational data, it is a model.
- "A twin needs full 3D geometry." Many useful twins are mostly data with simple geometry.
- "It's a technology purchase." Most of the work is in data quality, integration and ownership.
Where to read more
ISO/IEC publications on digital twins and the Gemini Principles (Centre for Digital Built Britain) describe principles for twins of the built environment.