Digital model, digital shadow, and digital twin
When papers sort plant digital copies, they look at automated data flow. A model has none, a shadow feeds one way into the copy, and a twin also sends control back to the hall.
What the three levels mean
Manufacturing literature that follows the Kritzinger integration ladder uses three labels. A digital model is a virtual copy with no automated link from the physical hall. A digital shadow receives automatic updates from sensors or systems into that copy, and does not write control back. A digital twin, in that stricter sense, exchanges data both ways so the virtual side can also change the physical side.
Public definitions that stress continuous sync with the physical counterpart sit on the same ambition. Reviews of manufacturing papers keep finding more work at model and shadow than at true bidirectional twins. Plant purchasing is often one level lower still: a finished 3D file sold as a twin before any automated feed exists. How one purchased slice gets declared finished sits under the Wikipedia digital twin vs what plants can actually buy.
Where plants actually stop: the model
Digital models are the default object called a twin on site. CAD, BIM, meshed scans, offline what-if scenes, and photoreal walkthroughs stay models until something on the floor updates them without a person re-exporting. They earn their keep before you buy the equipment. They go quiet once the layout is committed and the shift needs live state. Leadership still hears digital twin because the 3D model of a factory looks finished.
Plants stop there for a staffing and ownership reason, not because the silent file unlocks the most twin value. A capable engineer, a workstation, and a software license can ship a convincing hall file. Live shadow work needs sensors or location, integrations into plant systems, and agreement across logistics, production, IT, and facilities on naming, ownership, and who acts when the picture moves. That work is interdisciplinary on purpose. If the plant wants the wider value academy papers attach to digital twins, the spend and the staffing have to match that breadth, not only the modeling desk.
What a live shadow must carry
Academy twin value is not logistics alone. A live shadow into a shared 3D hall answers the search and fleet questions this shift. Under the same view it should store a spatially oriented dataset, the place-and-time record continuous improvement and later models need. That record is what correlation and causality analytics use for hall-wide pattern and root-cause work. Stopping at a silent model skips the search job, the stored replica, and the pattern work at once.
Plants may already own live tables: MES completes, WMS moves, ERP stock, SCADA alarms, even a forklift telematics list or an IoT dashboard. Those streams rarely land as continuous updates on the same hall model teams open in a design review. Dashboards and the 3D file stay separate. Until location, plant signals, or both refresh the model automatically, and until that picture is kept as history, the plant has not reached shadow, no matter what the slide says. That is also why many smart manufacturing and smart factory packages still feel unfinished: the label promised a twin while the hall still runs on disconnected screens.
Bidirectional twins that push setpoints or dispatch without a human are rarer than live shadows. They need a control owner and a safety case. Most halls that reach a useful live picture still stay on the shadow level, with a lead closing the loop by radio or by an existing PLC or WMS rule outside the twin. That is progress from a silent model. It is not the academy's highest level. The control trap is covered under bidirectional twin control on the slide, radio on the floor.
Why the levels matter for the next budget
If sponsorship thinks every 3D file is a twin, the next euro (dollar) buys prettier geometry instead of a data link into that geometry. Naming the level as model makes the missing step visible: sensors, integrations, and a habit of opening one live hall. Naming a rare live map as a full bidirectional twin oversells the other direction. Textbook label confusion across departments is under textbook labels rarely match the live floor.
Provider demos rarely open with model, shadow, twin. They open with the product name. Ask whether the 3D updates by itself from the floor, how often, and who still owns the next action. That filter sits beside different providers sell different twins.
BIM-first programs especially freeze on the model level while operations still run without a live hall picture. That case is under a BIM twin is not an operational twin of the shift. Simulation files without floor talk-back stay models too, covered under simulation is not a twin until the floor talks back.
How teams decide which level they are funding
1. Ask whether the 3D changes without a person updating the file - If not, you are on a model, even if live systems exist elsewhere.
2. Match the level to this quarter's decision - Fit before purchase needs a model. Seeing the wait, storing the shift, and asking which pairings keep preceding stops need a shadow. Automatic write-back needs a true twin with ownership, closer to a closed-loop digital twin with a named control path.
3. Say the level out loud in the go-live note - "We have a digital model of Hall B" or "we have a digital shadow of Hall B" beats "we have a digital twin" when the board still expects control loops. Say whether the shadow also keeps the spatially oriented record, or only shows a live screen.
4. Price the next step from where you actually are - Most plants fund model to shadow first: bring live data onto the hall and keep it. Budget people and integrations across teams, not only another license. Shadow to bidirectional twin is a later authority problem, not the first gap.
twinzo's usual job is that first real step: turning a silent hall model into a live operational digital twin that behaves as a shadow, with integrations into systems plants already run, and a clear scope that humans still assign the next move. That view supports internal logistics optimization and production monitoring on equal footing with the spatially oriented dataset underneath and the correlation and causality work that dataset enables. Capability depth is on the features overview. How the operational type carries all three jobs is under operational digital twins show the shift. Twin types sorted by plant job are under not every digital twin is the same. How a rollout is priced is on pricing.
Get in touch if you want to walk through whether your current "twin" is still a model, and what it takes to put live floor data onto that hall for the first time.