Digital Twin of an Industrial Plant: The Missing Data

IN SHORT

A digital twin of an industrial plant is a virtual replica of the facility that brings together its geometry, its technical data and, when needed, real-time operating data, so you can simulate changes and make decisions without walking the floor. How far you can trust it depends less on the software than on the geometry it was built on.

If that geometry comes from as-designed drawings (the original design, not what was actually built) that nobody corrected after the last expansion, the twin inherits the error. That is why we start with 3D laser scanning: our terrestrial scanners work at an accuracy of ±2 to 10 mm (roughly ±0.08 to 0.4 in), depending on the equipment and site conditions. On the projects we deliver, starting from that data translates into up to 35% less rework and 25% time savings, according to the results we report.

8-minute read. By the end you will know what separates a useful digital twin from a good-looking dashboard, the three decisions where it outperforms traditional documentation, and the plants where the traditional method is still the right call. If you need the general definition and its applications first, start with what a digital twin is.

You have probably seen one of these. Maintenance flags abnormal vibration, and the plant model doesn’t show that stretch of pipe because it was rerouted two years ago and nobody updated the file. An engineer wants to run a new line next to an aisle; the drawing shows 1.2 meters of clearance, but on the floor there are 70 centimeters because an electrical panel sits there that appears on no drawing. A manager approves an automation project on an eight-year-old layout and finds out, halfway through installation, that the structure meant to carry it is gone. Three scenes, one root cause: the model was built on data that stopped being true the day something moved on the floor.

At Foundtech our job is capturing reality: we turn the plant into the geometric layer that a digital twin runs on. We capture the plant with 3D laser scanners, model it in BIM and deliver the base that plant data connects to. We have completed more than 200 projects across Europe and the Americas and modeled more than 10 million m², with ISO 19650 certification and in alliance with the Swiss firm BIM Facility AG. Before switching platforms or adding sensors, check what geometry you fed the model that supports your decisions today.

The real criterion isn’t the digital twin software: it’s the geometry underneath

Most of the digital twin conversation is about platforms, IoT sensors and dashboards. That is the visible layer, the one that sells in demos. But judged by the decisions it supports, a digital twin is only as good as the geometry under that layer. If the base is an as-designed drawing with eight years of undocumented changes on top, the twin inherits the error and presents it with an air of precision.

A survey with 3D laser scanning and Scan to BIM gives the digital twin its ground truth. The output is a point cloud, a three-dimensional record of the plant as it stands today, usually captured without stopping operations. With it, comparing as-designed against as-built (what was actually built and modified) stops being the memory of whoever made the last change and becomes a measurement with millimeter-level precision. The BIM model is then built on what exists, not on what was once designed.

A digital twin doesn’t fix bad data. It scales it in real time.

Digital twin vs. traditional documentation: three decisions where the gap shows

The useful comparison goes decision by decision: what you will decide with the model, and how well the real geometry supports it.

1. Reconfiguring a line without breaching a safety clearance

The traditional method checks the original drawing and a spreadsheet layout, then confirms distances with a tape measure on a one-off walk. What it lacks is the actual clearance between structure, panels, ducts or piping added after the drawing was last updated. 3D capture provides measured distances on the point cloud that anyone can check whenever a clearance is questioned, without walking the line again.

2. Telling a process problem from a space problem

The traditional method logs the nonconformance, opens a ticket and assumes the cause is the process, because the layout says there is enough room. What it lacks is whether the documented space matches the real one, or whether a geometric deviation (a moved guard, a relocated conveyor) is the root cause. Quality management tools organize those records, but none of them measures space. 3D capture adds that comparison before you redesign a process that may never have been wrong.

3. Siting a new line without shutting down to measure

The traditional method walks the area with a tape and a camera, eating up engineering days and partial stoppages to measure critical zones. It is also selective: people measure what they think matters, not everything that is there. 3D capture surveys the whole area and delivers it as a model you can query. That was the problem at the Werdhölzli wastewater treatment plant in Switzerland: when planning its modernization, the owner had no up-to-date drawings or engineering records, and a traditional survey of a facility that complex would have taken too long. We scanned the plant and modeled its piping, tanks and structures, and the resulting digital twin is available for maintenance and asset management.

From passive to active twin: you don’t need sensors to start

A digital twin doesn’t have to be born connected to everything. Our Digital Twin service works in steps. First a passive twin: the plant’s BIM model enriched with technical data (manuals, warranties, maintenance dates), useful for inventory and documentation. Then, as the plant’s digital maturity allows, an active twin that brings in IoT sensors to show temperature, vibration or flow on the 3D model, plus connectors to systems such as SAP, Maximo or other CMMS platforms so work orders appear directly on the model. Everything runs in a web platform, with no software to install.

The active twin is what enables predictive maintenance. On the service page we state that this approach can cut downtime by up to 30% and that return on investment is typically seen in under 12 months; both figures depend on the industry and on each plant’s starting point. If you are building that business case, our breakdown of what predictive maintenance costs in an industrial plant covers what goes into the investment.

The real cost of deciding on data nobody measured

The cost of a model built on outdated data doesn’t show up on the project invoice. It shows up later, in every decision made with it. The NIST GCR 04-867 study put the cost of inadequate interoperability in the U.S. capital facilities industry at $15.8 billion a year (2002 data). Owners and operators bear two-thirds of it, mostly during operations and maintenance, not during construction.

One caveat: that study measures data interoperability in general, not manufacturing digital twins. We cite it because the mechanism matches what we see in the field: bad data flows downstream, and the plant operator almost always pays for it. On large footprints the effect grows. At Migros Millas, together with BIM Facility AG, we digitized roughly 11,000 m² (about 118,000 sq ft) with interior scans and drone captures, down to every piping and electrical route; the resulting digital twin supports maintenance and expansion decisions with current information.

When traditional documentation is still the right call

Not every plant needs a laser survey before building its digital twin. If the plant is small, rarely modified, and the decisions the model supports are low-risk (minor adjustments, not layout changes or safety calls), a well-kept drawing with disciplined manual updates can be enough. The same goes for an area that was recently renovated and documented with the same rigor.

The question that matters is how often something moves in your plant without anyone documenting it, and how expensive the decisions you make about that space are. Where not knowing the real geometry is costly, the survey pays for itself; where it isn’t, it’s an unnecessary expense.

What a digital twin won’t solve

A digital twin on a validated base doesn’t replace part metrology, meaning the precision measurement of an individual component with shop equipment; that is still a job for your quality team or a specialized provider. It doesn’t generate temperature or vibration data on its own either: that comes from sensors, which are integrated into the model when the scope calls for it. And it doesn’t replace your eQMS (electronic quality management system); it gives spatial context to the records you already keep.

What the first project looks like in your plant

The survey follows four stages: on-site capture and registration with laser scanners (and drones when the scope calls for it); BIM modeling from the point cloud; validation, overlaying the model on the captured cloud; and delivery in native, editable files such as RVT (Revit) or PLA (Archicad), with drawings generated from the model and, if needed, free viewers so anyone on the team can consult it.

On timing, a small project can be delivered in one week and a complex industrial facility can take 3 to 5 weeks; the exact schedule comes with the quote. Because the files are delivered editable, your team or any other BIM provider can keep working on them.

Don’t start by scanning the entire plant at maximum detail. It is the expensive, slow route, and it usually ends in a model so heavy that nobody opens it again. Start with the area where not knowing the real geometry costs you the most, at the level of detail (from LOD 200 to LOD 500) that decision actually needs.

What to do on Monday

• List the three most expensive operating decisions coming up next year (a new line, a reconfiguration, an automation project) and note which drawing each one relies on and what year it is from.
• Ask maintenance how many changes were made since that drawing was last updated. If nobody knows for sure, that is your answer.
• Pick one area to start: the line that changes most, the zone with the most nonconformances, or the space you plan to use for the next project.
• Decide whether you need a passive twin (documentation and inventory) or whether you already have sensors and a CMMS worth connecting.

If someone in your last layout review said “we should go confirm that on the floor” and that task is still open, your digital twin is running on data nobody has measured. For background on where this technology comes from, read our history of digital twins in industry and our piece on the impact of Industry 4.0 on productivity.

Frequently asked questions

Do I need IoT sensors to start a digital twin?

No. You can start with a passive twin for documentation and inventory and scale to an active twin with sensors when the plant is ready.

Can it connect to my ERP or CMMS?

Yes. We build connectors to systems such as SAP, Maximo or other CMMS platforms so you can see work orders on the 3D model.

Do I have to shut down the plant for the scan?

Usually not. Capture is non-intrusive and modeling runs in parallel, without affecting daily production.

Tell us which area of your plant carries your most expensive decisions, and we’ll assess with you whether it is ready for a digital twin, starting with the survey, so your next decision is made with certainty. Book an assessment.