What Is Clash Detection in Construction?

A hard clash is unambiguous: a duct passes through a beam, a pipe runs through a column. Software finds these reliably because the test is pure geometry, and they are the clashes everyone means when they say clash detection.

Soft clashes are more interesting and more often missed. Nothing physically intersects, but a valve sits where no one can reach it, a coil has no pull space, a panel has less clearance in front of it than code requires. These depend on clearance zones being modelled, and if nobody modelled the access space, the software has nothing to test. Workflow clashes are not geometric at all — two trades needing the same area in the same week — and they surface in sequencing review rather than in the model.

The mechanics are cyclical. Trades issue models on an agreed date, a coordinator federates them and runs the clash tests, results are filtered and grouped, and the coordination meeting works through them zone by zone deciding who moves. Trades update their models, and the cycle repeats — typically weekly or fortnightly during coordination.

The hard part is not the software. It is the filtering and the negotiation. A raw clash run on a large project returns tens of thousands of results, most of them duplicates of the same conflict seen from different angles, tolerance noise, or intersections between elements that are not real objects. Turning that into a few hundred decisions a room full of people can make is the coordinator’s actual job.

It only knows what is modelled. Elements omitted from a model cannot clash, and the classic omissions — hangers and supports, insulation thickness, access and pull space, temporary works, in-wall blocking — are exactly the things that cause field conflicts. A model with clean clash results and no insulation modelled will still produce field problems.

It also says nothing about whether the resolution is sensible. Software reports that a duct and a beam intersect; it does not know that lowering the duct puts it below the ceiling height the architect committed to. And it does not check the 2D documents, so a conflict that exists only in the drawings — which is where the contract lives — is invisible to it.

The return on coordination is measured in things that do not happen: the RFI not raised, the rework not performed, the ceiling not reopened, the delay not incurred. That makes it consistently undervalued, because avoided costs never appear in a budget line while the coordination effort does.

The projects that get the most from it are the ones that set the rules early — model element authorship, level of development by phase, clearance zones as modelled objects, a fixed issue calendar — and treat unresolved clashes as an item with an owner and a date rather than a number in a report.