Digital twins, explained: a living model of a real place

A digital twin is a virtual copy of a real place or asset that stays connected to the original. Sensors, surveys and records feed the copy, so it shows how things are now, not just how they looked on the day it was built. A building manager can see which rooms are empty. A city planner can see where water would flow in a storm. A museum can see the condition of a fragile wall over time.

The term is used loosely, and often applied to things that are really just 3D models. This article separates the two, explains what goes into a twin, and helps you decide whether one is worth building.

Twin or just a 3D model?

• A 3D model captures shape and appearance. It is built once, and it is excellent for presentations, planning discussions and virtual tours.
• A digital twin adds data that keeps changing, such as temperature, occupancy, water level, equipment condition or maintenance history, and it uses that data to answer questions.

The test is simple. Ask whether the virtual copy would still be correct a month from now without anyone touching it. If yes, it is a model. If it updates itself from the real world, it is a twin.

What goes into a twin

1. 01A 3D base. It is built from laser scans, drone or ground photographs, or design files. The level of detail should match the decisions it supports.
2. 02Data feeds. Sensors, GIS layers, asset registers, maintenance logs or manual surveys connect the model to what is happening.
3. 03A way to see and use it. A web viewer, a dashboard, a map or a VR headset, depending on who needs it.
4. 04Rules and analysis. Alerts, simulations and forecasts that turn the data into advice.
5. 05Care and ownership. Someone responsible for keeping the model and its data accurate.

Stages of maturity

A twin does not need to be sophisticated on day one. Many successful projects grow through four stages.

• Descriptive: an accurate 3D copy that people can explore and measure.
• Monitoring: live or regularly updated data is shown on the model, for example room temperatures or sensor alerts.
• Predictive: the twin forecasts what will happen, such as when equipment might fail or how a flood could spread.
• Prescriptive: it recommends or triggers actions, for instance rescheduling maintenance or adjusting ventilation.

Climbing one stage at a time keeps cost under control and lets you check that each stage is actually used before building the next.

Where twins pay off

• Campuses and buildings: plan maintenance, manage space and find faults faster.
• Heritage sites: record condition precisely, track change over years and give researchers remote access.
• Cities and utilities: test flood, growth or traffic scenarios before spending on construction.
• Factories and plants: train staff and trial layout changes without stopping production.
• Agriculture and forestry: combine terrain, crop and weather data to guide field decisions.

What it takes to build one

Capturing the place is only the first step. Many projects begin with drone photography or laser scanning, which produce dense 3D data. That data is then cleaned, simplified and turned into a model that loads quickly. The model is then linked to data sources and to a viewer that fits the people who will use it. A model that takes two minutes to open will not be opened.

The bigger cost over time is keeping it current. Buildings change, sensors fail and records fall behind. Plan who updates what, how often, and what happens when data is missing.

Risks to plan for

• Stale data. A twin nobody updates quickly becomes an expensive 3D model.
• Security. Connecting sensors and systems to a model creates new routes for attack, so access control and monitoring matter.
• Lock-in. Proprietary formats can trap your data. Prefer open formats and exportable data.
• Over-building. Detail beyond what decisions need adds cost without adding value.

Questions to answer before you build

1. 01Which decision will this improve, and how will we know it did?
2. 02Who will use it every day, and what do they need to see?
3. 03Where will the data come from, and who is responsible for it?
4. 04How often must it update to stay useful?
5. 05What will it cost to run each year, not only to build?

A small example: a campus water tank

Take a university campus that struggles with water shortages. Its first digital twin does not model the whole campus. It covers a single rooftop tank and the pipes that feed three buildings.

• The 3D base: a simple model made from drone photographs and the plumbing drawings.
• The data: a level sensor in the tank and flow meters on the pipes, reporting every few minutes.
• The view: a web page that shows the tank level on the 3D model and colours each pipe by flow.
• The rules: an alert when the tank falls below a threshold, and a weekly report on unusual night-time flow, which often means a leak.

Within a few months the team finds a leak that had gone unnoticed, and learns when each building really uses water. That evidence, not the 3D model, is what justifies extending the twin to the next system.

Build, buy or partner?

• Buy a platform when your needs are common, for example facility management, and a product already fits them.
• Build when your data or decisions are unusual, or when you need full control. Expect to maintain it for years.
• Partner when you need specialist skills, such as scanning and modelling, but want to own the data and the decisions.

Whichever route you take, insist on open data formats and the right to export your model and data, so that you are never locked in.

Frequently asked questions

How accurate does the 3D model need to be?

Only as accurate as the decisions require. Maintenance planning may need centimetre-level detail. A public tour of a heritage site may not. Ask what the least detail is that still supports the decision.

Does a digital twin need sensors?

Not always. A twin can be updated from surveys, records or periodic scans. Sensors help when conditions change quickly and you need live information.

How much does one cost?

Costs range from a few thousand dollars for a scanned model with a viewer, to large sums for a connected city-scale system. The annual cost of keeping data current is often larger than the cost of the first build.

Is it the same as BIM?

No. Building information modelling describes a building’s design and construction. A digital twin can use BIM as its starting model and then adds live data and analysis during operation.

Digital twins are not a technology to adopt for its own sake. They are a way of giving people a clear, current picture of a real place, so that they can make better decisions. If that picture is worth the effort of keeping it current, a twin is worth building.