Forty-two hard clashes, all in one plant room, all found before a single steel order went out. This is the walkthrough of how that number came about — the model structure, the clash matrix, and why the first coordination cycle found so much at once.
The brief on this contract was ordinary enough on paper: federate the architectural, structural and MEP models for a shopping mall in Yangon, run coordination through to construction issue, and keep the contractor's programme intact. The plant room is where that brief got tested. It's a small footprint carrying a disproportionate share of the building's risk — chillers, pumps, electrical distribution, fire services and structural transfer beams all competing for the same volume above a suspended ceiling that wasn't going to move.
Setting the clash matrix.
Before any detection run, we agreed the clash matrix with the design team and the contractor: which discipline pairs mattered, what tolerance counted as a hard clash versus a soft one worth flagging but not blocking, and who owned resolution for each pair. MEP-to-structure and MEP-to-MEP were both set to zero tolerance in the plant room specifically — everywhere else on the project, a small soft-clash allowance was acceptable.
That distinction mattered. A generic, one-size-fits-all clash tolerance across the whole model would have buried the plant room's real problems under hundreds of insignificant results elsewhere in the building.
Why the first cycle found so much.
The plant room's forty-two clashes weren't spread evenly across the project — they were concentrated in one federated run, in one room, because that room had never actually been coordinated in 3D before. The structural transfer beam had been sized and drawn independently of the mechanical riser layout; the fire services main ran a route that assumed a ceiling void the electrical containment had already claimed. None of this was anyone's mistake in isolation — it's what happens by default when four disciplines design in parallel without a shared, up-to-date volume to check against.
Running the clash detection early, at LOD 350, on a model that was federated weekly rather than at milestone gates, meant the plant room's problems surfaced while the fix was still a model change — a beam reroute, a riser relocation — instead of a site instruction once ductwork was already fabricated.
BIM ACE found forty-two hard clashes in the mall's plant room before steel was ordered. That is the contract paying for itself in week three.
Project Director — Main contractor, Shopping Mall, YangonWhat changed going forward.
Two things carried over into the rest of the project after this cycle. First, the plant room's zero-tolerance clash rule was extended to every other MEP-dense zone on the building — risers, ceiling voids above corridors, and the loading dock. Second, the coordination cadence itself moved from milestone-gated to weekly, so a discipline change never sat unchecked against the federated model for more than a few days.
- Clash tolerances set per zone, not uniformly across the whole model.
- Weekly federation instead of milestone-only checks, so issues surface while they're still cheap to fix.
- A named owner for every open clash, reviewed in the same weekly cycle — not a report that gets filed and forgotten.
None of this is unusual practice for a properly run BIM contract. What made the difference here was doing it in the plant room specifically, early, before the sequence that makes a clash expensive — procurement and fabrication — had already started.