A BIM model can “look right” while its underlying data fails the requirements — empty properties, inconsistent naming, and so on. Finding those by eye alone doesn’t scale. That’s what model checking (validation) is for.
- Model checking = mechanically verifying that a model meets its requirements
- Rules catch "missing required property," "value out of range," "naming violation," and more
- Not the same as clash detection — that's physical collisions; this is data correctness
- Run it early and often to cut downstream rework
Why eyeballing isn’t enough
On a model with thousands or tens of thousands of elements, visually confirming “every fire door has a fire-rating property” is simply not realistic. Manual checks suffer from oversights, person-dependence and poor repeatability.
Model checking defines those confirmations as rules and verifies them mechanically, all at once. The value is repeatability: the same rules, any number of times, the same result regardless of who runs them.
What kinds of rules?
The common rule shapes:
✅ Required property
Detects missing properties that should be there — e.g. fire rating on fire doors.
📐 Value range
Flags values outside a sensible range — e.g. wall thickness 50–1000 mm.
🔤 Value equals
Confirms values match an allowed vocabulary — e.g. occupancy from a fixed list.
🏷️ Naming rules
Element names / classification codes follow the convention, so schedules and links hold.
Many tools combine built-in integrity checks (elements with no storey, openings hosted on nothing) with project-specific rules you define. Bimly, for example, evaluates both built-in checks and user-defined rules (required, range, equals).
How it differs from clash detection
Often confused, but the goals differ:
Data correctness
- Verifies properties, naming, value validity
- Asks "does the information meet the requirements?"
Physical collisions
- Finds element-to-element collisions (MEP × structure)
- Asks "is the geometry clashing?"
Both protect quality, but one looks at data and the other at geometry.
How to run it in practice
Turn requirements into rules
Translate client requirements and internal standards into checkable rules (required, range, equals).
Run early and often
Check periodically from mid-modeling, not just before submission.
Share and fix findings
Share problems as issues (BCF) and track them to resolution.
Final check before delivery
Confirm the model passes the submission rule set and keep the result on record.
Summary
- Model checking (validation) mechanically verifies a model against its requirements.
- Rules cover required properties, value ranges, equals checks and naming.
- It looks at data correctness, distinct from clash detection’s physical collisions.
- Run early and often to reduce downstream rework.
“Looks right” and “data is right” are not the same. Model checking is the quiet step that keeps BIM as usable data. IDS (Information Delivery Specification), which writes the requirements themselves machine-readably for automatic checking, is the emerging standard that takes this idea further.