Every large, multi-discipline project has at least one sheet where a mechanical routing decision and a structural member are drawn as if the other doesn't exist. Most of the time, someone catches it before it matters. Sometimes nobody does — and a single missed MEP/structural drawing coordination conflict, invisible on paper, turns into one of the most expensive line items on the job.
The conflict itself is rarely dramatic. A duct run drawn at one elevation on the mechanical sheet. A wide-flange girder drawn at a conflicting elevation on the structural sheet, twenty feet away in the same set. Neither drawing is wrong on its own terms — the mechanical engineer routed to serve the space, the structural engineer sized and placed the member to carry the load. The conflict only exists in the overlap between the two sheets, and if no one is reading both sheets together, looking specifically for that overlap, it survives all the way to the field.
What actually happens when the duct and the beam disagree
Picture the sequence on a real job. Structural steel goes up first, on schedule, inspected and approved. Weeks later, the mechanical sub shows up to hang ductwork and finds the girder sitting exactly where the duct was supposed to run. There's no more redlining a sheet — the steel is already bolted, welded, and signed off. What happens next is the same sequence on almost every project this happens on:
- The mechanical sub stops work on that run and files an RFI, because there's no way to install as drawn.
- The GC's field team and the design team scramble a field fix — reroute the duct, notch the flange, add a transition fitting, or in the worst case, request a structural modification.
- The structural engineer has to sign off on anything that touches the member, which means an engineering review cycle before anyone can cut into approved steel.
- The fix gets priced as a change order, because it's additional scope nobody bid on and nobody budgeted for.
- Every day that duct run sits unresolved is a day of schedule float on a critical path activity, which the GC's general conditions cost tracks whether or not anyone's actively working.
None of those five things happen when the same conflict is caught on the sheet before steel goes up. It's a redline. It costs the time it takes to move a line on a drawing. Caught in the field, it's a demo-and-rebuild sequence with an engineer of record's stamp required before anyone touches it again.
Add those up and a single clash — one conflict, on two sheets, that a document-level review would have caught in an afternoon — lands in the $300,000–$400,000+ range once schedule impact is priced in. That's not because the physical fix is expensive. It's because everything downstream of "steel's already up" gets priced as a claim instead of a correction.
Why this conflict wasn't caught by anyone already looking at the set
This is the part owners find hardest to believe: a set this size already had multiple qualified people reviewing it before it ever reached the field, and the conflict still made it through. As we've written about separately, the architect's QC checks the design against its own intent — did the mechanical sheet match what the mechanical engineer meant to draw, did the structural sheet match what the structural engineer meant to draw. That review, done correctly, still doesn't catch a conflict between the two disciplines' sheets, because it was never scoped to compare them against each other.
The GC's own coordination review reads the set differently, and for a different purpose: to price it accurately. An ambiguity or overlap found during that pass is easy to price around rather than flag — raising it during bid risks a competitor undercutting on a cleaner number. That's not a GC acting in bad faith; it's the same structural incentive we've described in why RFI count is the wrong metric: the people closest to the conflict aren't necessarily the people scoped to surface it before it's locked in.
FMI and PlanGrid's 2018 industry study, based on responses from nearly 600 construction professionals, found that time spent on rework, hunting for project information, and resolving conflicts — what the report calls "non-optimal" activity — costs the U.S. construction industry an estimated $177.5 billion a year. Poor communication between project participants alone accounted for 26% of that rework, a roughly $17 billion annual figure. A duct that conflicts with a beam because two disciplines' sheets were never actually checked against each other is exactly the kind of gap that study is describing.
The math that makes "just one clash" a real number
It's tempting to treat a single missed clash as an outlier — bad luck on one sheet out of hundreds. The industry data doesn't support treating it that way. Separate research on construction rework consistently puts total rework cost in the range of 5–9% of project value, and design-related change orders specifically — the category a missed MEP/structural conflict falls into — run an estimated 3–5% of total construction budget on their own, with total change order costs on major projects frequently reaching 10–15% of contract value. On a project this size, the low end of just the design-error share of that range can still mean roughly $1M in exposure, spread across however many conflicts like this one are sitting in the set, uncaught, right now.
That's the frame that matters for an owner deciding whether a review is worth the cost: it's not "will this specific clash happen." On a set this size, some version of it almost always does. The question is whether it's found while it's still a redline, or found by a mechanical sub standing under a beam that was drawn to occupy the same six inches as the ductwork.
What proactive coordination is actually worth, in a documented case
The upside isn't hypothetical either. A published ROI case study from Haskell, a national design-build firm, tracked a $230 million design-build food and beverage project where the VDC team ran proactive BIM coordination throughout design and documented every conflict caught before construction using an issue-management platform. The result: a roughly $200,000 investment in coordination returned more than $2.5 million in avoided cost and schedule impact on that project alone — north of a 10x return.
That case ran BIM-based clash detection through design and preconstruction, which is the right tool for exactly what it's good at: catching two modeled elements that physically occupy the same space. But clash detection has a specific blind spot, and it's the same blind spot that let the duct-and-beam conflict above survive two separate professional reviews.
Why BIM clash detection alone still misses these
Clash detection software compares geometry. If the duct and the beam are both modeled, and both modeled correctly, most coordination workflows will catch the physical overlap before it's ever built — that's precisely the mechanism behind the Haskell numbers above. The gap is what happens when the conflict isn't a geometry problem. A general note on the structural sheet that prohibits penetrations through a member type used elsewhere for exactly that purpose. A spec section that specifies a clearance the drawn detail doesn't allow for. A connection that was never modeled at all, because no one on either team realized it needed coordinating in the first place. None of those trip a clash report, because none of them are two solid objects sharing space in a model — they're contradictions between what's written and what's drawn, sitting in the issued document set, waiting for someone to read both sides of the sheet at once.
That's the gap an independent, document-level review is built to close — not instead of BIM coordination, but for the category of conflict BIM coordination structurally can't see. As we've noted in comparing pre-bid and IFC-stage timing, the earlier that read happens, the cheaper every conflict it finds stays. A redline on an unissued sheet costs nothing. The same conflict, caught after steel is up, costs whatever demo, re-fabrication, a re-stamp, and a few weeks of float add up to — which, as the buildup above shows, clears $300,000 without anyone doing anything unusual at all.
Key takeaways
- A single MEP/structural conflict caught in the field — not on the sheet — routinely compounds into $300,000–$400,000+ once demo, re-fabrication, engineering re-stamps, and schedule float are priced in.
- The architect's QC and the GC's coordination review are both real, and neither is scoped to catch a conflict between two disciplines' sheets specifically.
- FMI and PlanGrid's industry study puts total U.S. construction rework cost at $177.5 billion a year, with poor communication between parties responsible for 26% of it.
- Design-related change orders run an estimated 3–5% of total project budget industry-wide — a real number on a large project, not a rounding error.
- BIM clash detection reliably catches modeled geometry conflicts (a documented $230M project case study returned 10x on coordination spend) but structurally misses contradictions between written notes, specs, and drawn details — the gap a document-level review is scoped to close.
None of this requires assuming anyone on the project did their job badly. The architect's QC did what it was scoped to do. The GC's coordination team did what it was scoped to do. The BIM model got clash-checked the way BIM models get clash-checked. The conflict still made it to the field, because nobody's job was specifically to read the issued set for the contradiction between two disciplines' sheets before the steel went up. That's a gap in scope, not a gap in effort — and it's the specific gap an independent, owner-side review before bid is built to close.
Frequently Asked Questions
How does a duct-versus-beam clash even survive the architect's QC and the GC's coordination review?
Each of those reviews is scoped to a different question. The architect's QC checks that the drawings match design intent within each discipline. The GC's coordination review reads the set to price the work accurately. Neither is specifically scoped to compare the mechanical sheet against the structural sheet looking for a contradiction between them, which is exactly the kind of conflict that makes it to the field undetected.
Does BIM clash detection catch this kind of conflict before it happens?
It catches it when both elements are correctly modeled — a duct and a beam occupying the same modeled space will typically surface in a clash report, which is why disciplined BIM coordination has a strong, well-documented track record. It doesn't catch a written note that contradicts a drawn detail, a spec clearance the drawing doesn't allow for, or a connection nobody modeled in the first place, because none of those are two solid objects overlapping in a model.
Why does the same conflict cost so much more once it's found in the field instead of on the drawing?
Because everything downstream of "steel's already up" turns into a claim instead of a correction. A field-caught clash typically means demoing and rebuilding installed work, a structural engineer's review and re-stamp before anyone can modify approved steel, expedited re-fabrication at a rush premium, and schedule float that the GC's general conditions cost accrues regardless of whether active work is happening. Caught on the sheet, the same conflict is a redline that costs the time to move a line.
Is a $400,000 change order from one clash realistic, or is that an exaggerated number?
It's a realistic upper-middle range once the full cost stack is priced in — demo and rework, an engineering re-stamp, expedited fabrication, and several weeks of schedule float on a project at that scale routinely add up to $300,000–$400,000+ for a single significant clash. It isn't a guaranteed outcome for every conflict, but it isn't an outlier either; it's closer to what industry rework and change-order cost data would predict for a conflict caught this late.
What actually closes this gap, if the existing reviews already miss it?
An independent, document-level review scoped specifically to find cross-discipline conflicts — reading the mechanical, structural, and spec sections against each other, not just each one against itself — before the set goes out to bid or construction. It doesn't replace BIM coordination; it covers the category of conflict, written contradictions and omissions rather than modeled geometry, that clash detection isn't built to see.