BIM coordination: catching clashes before they reach the site

A duct that clashes with a structural beam costs almost nothing to move in a model, a few clicks and a short conversation with the engineer, and days to move once it has actually been fabricated, delivered and partly installed on site. The gap between those two costs, multiplied across the hundreds of potential clashes on a typical building, is why coordination meetings exist, and why the discipline of running them properly is worth far more than the time they take.

Federate the models, not the drawings

Coordination works on a federated model, a combined view built by overlaying every discipline’s own separate model file, structure, architecture, mechanical, electrical and so on, and checking the combination automatically for physical overlaps between elements that were never meant to occupy the same space. A clash report generated from this federated model is only useful if every trade’s individual model is genuinely current when the check runs; a report built partly from last month’s mechanical model and this week’s structural model will flag clashes that were already resolved and miss new ones that have since been introduced, undermining confidence in the whole process.

Not every clash matters the same

Automated clash detection software typically flags many thousands of individual touches on a project of any real size, and the great majority of them are not real problems: a pipe passing through a zone reserved for future maintenance access, two elements that touch by a fraction of a millimetre due to modelling tolerance rather than a genuine physical conflict, or a duct crossing a wall at a point where the wall is known to be omitted in the final design. Someone experienced enough to tell real clashes from this background noise has to review the raw report before it goes anywhere, or the sheer volume of flagged items trains the team to stop trusting, and eventually stop reading, the report altogether.

Decide who fixes what

A genuine clash between two services needs an assigned owner, not just a flag sitting in a report that nobody has claimed responsibility for. Agree in the coordination meeting itself, with both relevant trades present, who adjusts their model to resolve each clash, and confirm the fix has actually been made before the next coordination run, rather than assuming it happened because someone nodded in the meeting. A clash log that tracks status, open, assigned, resolved, verified, keeps this from becoming a list that quietly grows every week without anyone noticing it was never actually shrinking.

Running an effective coordination meeting

The meetings that work best tend to be short, frequent and focused on a specific zone of the building rather than an attempt to review the entire model every time. Walking through one floor or one zone in detail, with the relevant trades in the room and the federated model live on screen, resolves far more genuine clashes per hour than a broader meeting that tries to cover the whole building and inevitably skims over the detail where the real conflicts hide.

Bring the model to site

A federated model is most useful, in the end, when the crew actually installing the work can see it too, on a tablet at the point of work, not only reviewed by designers in a meeting room days or weeks earlier. Site teams who can check the coordinated model against what they are about to install catch the clashes that slipped through coordination, and just as importantly catch cases where the model was right but the installation was about to deviate from it, before either becomes a rework item.

This article is general information and not engineering advice.