Digital twins and the circular building.

Most of the material value in the built environment is decided long before a wrecking ball or a saw arrives. It is decided in the drawings, in the joints, in the specifications that say whether a column is glued or bolted, whether a façade is a system or a composite, whether a floor is designed to last one lease or three. By the time a building reaches end of life, the possible endings have already been narrowed to what its design allows.

The problem is that almost none of that decision context survives the building. Drawings drift out of date, specifications sit on a drive nobody can find, the people who chose the fixings retire, and the demolition team turns up with a hammer and a landfill contract because that is the cheapest way to deal with something nobody understands anymore.

A digital twin, done properly, is the fix for that. Not the consultancy-brochure version of a twin — a spinning 3D model on a screen — but a live, structured record of what a building is, what it is made of, and how it can come apart again. It is the memory layer that lets the building be assembled once and then, decades later, disassembled without guesswork.

What a building's twin actually is

A digital twin of a building is three things stitched together. The first is a geometric and semantic model, usually rooted in BIM — the geometry of the building, the classification of its parts, and the relationships between them. The second is a data layer holding the properties of those parts: what they are made of, where the material came from, how they are connected, what maintenance they have had, what tests they have passed. The third is a link to the real world — sensor data, inspection reports, changes made during use — so the model stays in sync with the actual building rather than drifting into a pretty fiction.

TNO's Predictive Twin Lab for the built environment describes twins along exactly this spectrum: descriptive twins that mirror what a structure is, and predictive twins that forecast how it will behave, be used, or perform under stress. Their broader Digital Built Environment programme frames the twin as a shared substrate for builders, designers, managers and policymakers, not a private tool for one contractor.

The circular question is a specific one within that: can the twin answer, at any point in the building's life, "what is in this, where did it come from, and how do I get it back out?"

Assembly: capturing the memory while it is still cheap

The moment to build a circular twin is at assembly, because that is the only moment when the information is free. The bricklayer knows which pallet they are laying. The steel arrives with mill certificates. The timber has a chain-of-custody document. The installer knows the torque they used on the connection. Capture that information into the twin as the building goes up, and the building leaves site with a full material inventory attached to it, joint by joint.

Fail to capture it, and every subsequent attempt to recover it is more expensive than the material is worth. Later surveys can guess at what is behind the plaster. They cannot re-establish which of three identical-looking bolts was torqued and which was not.

This is the practical work behind TNO's Asset Lifecycle Information Management (ALIM) research: open, linked-data standards that make lifecycle information about a building portable across systems and durable across ownership changes. Without that standardisation, the "twin" ends up trapped inside a specific contractor's software, a proprietary format, or an export that quietly loses its semantics. The twin's real value is that it can still be read in forty years, by a company that does not exist yet, running software nobody has written yet.

The middle life: keeping the twin honest

A twin that is only accurate on handover day is not a twin. It is a snapshot. The building lives for decades, and every retrofit, every tenant fit-out, every roof repair either updates the twin or invalidates it. This is where most twins fail in practice — not for lack of ambition at delivery, but for lack of an owner during the long, unglamorous middle.

The way through is boring and structural: the twin has to be treated as part of the asset record, not as an IT deliverable. Any change to the building has to close a loop back to the twin, the way any change to a regulated document closes back to version control. Predictive twins earn their keep here — running scenarios on energy performance, structural stress, and material condition — but only if the descriptive layer underneath them stays true.

Disassembly: the twin as the demolition plan

At end of life, everything the twin has quietly stored becomes the demolition and recovery plan. A well-maintained twin can be read as a bill of materials for what is about to be released: how much timber, of what species and grade; how much steel, of what section, from which mill; how many façade panels, how they are fixed, whether the fixing can be reversed. That inventory is what a reverse-logistics operator, a component broker, or an urban mining platform needs before they can price a recovery.

Recent research is starting to make this operational. A 2025 Scientific Reports study by Kaewunruen and colleagues shows how a digital twin, coupled with BIM and lifecycle assessment, can drive a strategic demolition plan for bridge infrastructure — sequencing the takedown to maximise recovered value and minimise embodied carbon released. The Circular Twin Framework work out of TU Wien argues something similar for buildings: the twin is the ecosystem layer that lets circular building strategies actually run at scale, rather than remaining a per-project heroics exercise.

On the ground, projects like DISCOVER at EnergyVille/VITO are pairing digital twins with robotics on active demolition sites, scanning what is really there, updating the twin against reality, and using that live model to guide selective deconstruction instead of blunt-force demolition. The pitch is precise: turn the demolition site into a digital twin so workers are safer and materials flow back into use.

Where the twin meets the Digital Product Passport

The Digital Product Passport, arriving under the EU Ecodesign for Sustainable Products Regulation, will require structured, product-level data on materials, origin, recycled content and repairability. For construction products, that data has to come from somewhere, and the twin is the natural home for it. Each component sits in the twin with its passport attached — not as a separate PDF, but as a linked, queryable record that travels with the geometry.

Read together, the twin and the passport turn a building from a static object into a stock account. The building has an opening balance of materials at handover, a running ledger of changes during use, and a closing balance ready for recovery at end of life. That is what circular real estate finance is quietly starting to price — not the label on the building, but whether the numbers underneath it can be trusted.

The readiness question

A twin, like every other circular tool, only works inside an organisation that is ready to use it. In our Circular Readiness framework, keeping a twin honest across decades is CRL4 work: governance, data ownership, integration into how the asset is actually managed. Buying the software is easy. Building the discipline that keeps the twin true is not.

The organisations that will benefit most are the ones treating the twin as a strategic asset rather than a compliance artefact — the ones asking, before they commission a building, who will own its twin in 2065 and how it will still be readable then. Everyone else is building a very expensive drawing.

Where this leaves the sector

The technology is not the bottleneck. BIM is mature, linked-data standards are being hardened by TNO and its peers, sensors are cheap, and the regulatory pull from ESPR and the Digital Product Passport is real. The bottleneck is coordination: agreeing what goes in the twin, who owns it, how it survives changes of ownership, and how it hands off to demolition and recovery decades later.

That is a CRL5 problem, not a software problem. Standards bodies, developers, owners, insurers, demolition contractors and material platforms all have to point at the same data model. The countries and cities that get that alignment right will treat their built environment as a material reserve. The ones that do not will keep demolishing value they can no longer see.