Clash Detection Without BIM: Is It Possible?

Short answer: yes — and on most projects it’s the only clash detection that happens early enough to matter.

“Clash detection” has become shorthand for one specific workflow: load federated 3D models into Navisworks, Autodesk Construction Cloud, or Solibri, and let the software flag where the ductwork runs through a beam. That workflow is real and valuable. But it carries an assumption that quietly excludes most of the work actually being built: that you have coordinated 3D models, and that you have them early enough to act on what they reveal.

For a large share of projects — especially in multifamily, mid-market commercial, and renovation work — that assumption doesn’t hold. The conflicts are still there. They’re just sitting in 2D PDF drawing sets, waiting to surface as a change order. This post is about finding them there, before the field does.

Why “clash detection” came to mean BIM

The tools that popularized the term are model-based by design. Navisworks, Solibri, and ACC’s clash tooling all consume IFC or native model geometry and test it for spatial interference: does object A occupy the same coordinates as object B? That’s a genuinely hard problem that 3D geometry solves elegantly, and for projects with mature, well-coordinated BIM it’s the right tool.

The result, though, is that an entire category of thinking got welded to a single file format. Ask most teams “can we run clash detection?” and the honest answer becomes “only if we’ve got a federated model” — which reframes a project-quality question as a tooling question.

The reality: most projects don’t have a coordinated model early (or at all)

Three things are true on a lot of real jobs:

• The model arrives late. Design-side BIM, where it exists, often isn’t federated and coordination-ready until well into design development — after the budget and many sequencing decisions are locked.
• The trades model separately, if at all. MEP subs frequently coordinate in their own models on their own timeline. A truly federated, all-discipline model is a deliverable many projects never fully pay for.
• The documents are the source of truth anyway. The drawings and specs — not the model — are what gets permitted, bid, and built. Conflicts that live in the documents get built into the building.

So the practical window where you most want to catch conflicts — during design review, before bid, before mobilization — is frequently the window where you have the least model coverage. You have PDFs.

What “conflict detection” means on 2D documents

On a drawing set, the conflicts that cost money aren’t only the spatial clashes 3D is built to find. They’re a broader family, and most of them are document problems, not geometry problems:

• Spec-vs-drawing conflicts. The drawing calls one product or assembly; the specification section calls another. Neither sheet is “wrong” on its own — they just disagree.
• Cross-sheet conflicts. A dimension, elevation, or detail on one sheet contradicts the same element on another. Architectural says one floor height; structural assumes another.
• Cross-discipline conflicts. The classic one: structure, MEP, and architecture each solve their own scope, and the seams between them don’t line up. A shear wall lands where a duct main has to run. (See our MEP coordination process for how these seams form.)
• Missing-coordination gaps. Something required is simply absent on the sheets it should appear on — fire-alarm coverage not carried onto a new structure, an ADA clearance that nothing accounts for.

None of those require 3D geometry to find. They require reading every sheet against every other sheet and against the specs — comprehensively, without fatigue. That’s a document-review problem, and it’s exactly the problem AI is now good at.

2D AI review vs 3D model clash detection: what each actually does

These are complementary, not competing. They catch different things, at different times, at different cost.

	2D AI plan review	3D model clash detection
Input	The PDF drawing set + specs you already have	A federated, coordination-ready 3D model
Best at	Spec-vs-drawing, cross-sheet, missing-coordination, code/clearance gaps	Hard spatial interference between modeled objects
When it runs	At design review / before bid — as soon as drawings exist	After models are built and federated
Who can run it	Anyone with the PDFs (GC, developer, owner’s rep, sub)	Teams with BIM staff and modeled scope
Marginal cost	Low — no modeling effort required	High — requires the model to exist first

The honest framing: 3D clash detection finds the spatial clashes better, because that’s what geometry is for. But it can only look at what’s been modeled, and only after modeling happens. 2D AI review looks at the whole documented project — including the spec conflicts and coordination gaps a clash engine never sees — and it can do it the day the drawings are issued. For a deeper side-by-side, see 2D vs 3D clash detection.

When you still want BIM — and when 2D review is enough, or earlier

Use BIM clash detection when you have the models and the spatial coordination is the dominant risk: dense mechanical rooms, complex healthcare or lab MEP, tight structural-MEP interstitial space. If the model exists and is coordination-ready, run it.

Reach for 2D AI plan review when:

• You don’t have a federated model (yet, or at all).
• You need an answer at design review or before bid — not after coordination.
• The risk is as much in the documents (spec disagreements, cross-sheet mismatches, missing scope) as in the geometry.
• Stakeholders who need to see conflicts — the developer, the owner’s rep, the lender — don’t run modeling software.

In practice most projects want both at different moments: 2D AI review early and continuously as drawings evolve, BIM clash detection later where modeled coordination is critical. They’re not substitutes; they’re a sequence.

Flikt’s position: no BIM required

Flikt runs AI construction plan review directly on 2D PDF drawing sets — no model, no IFC export, no modeling staff. It reads the full set the way a senior reviewer would, sheet against sheet and drawing against spec, and surfaces the spec-vs-drawing, cross-sheet, cross-discipline, and missing-coordination conflicts described above.

This isn’t theoretical. On real projects the conflicts hiding in 2D sets have driven the expensive outcomes we document on the evidence page — including one project where more than 40 change orders traced back to conflicts between 2D plan sheets, and individual structural conflicts (a shear wall colliding with duct routing) carrying $28,000–$45,000 in rework plus two to three weeks of schedule. Design errors caught in review rather than in the field cost roughly an order of magnitude less to fix; the entire value of finding them in the documents is finding them before they’re built.

That’s the answer to “is clash detection possible without BIM?” Not only is it possible — it’s the version of conflict detection that runs early enough, and on the documents real projects actually have, to change the outcome.

---

Flikt reviews your 2D plan sets for cross-discipline conflicts before construction — no BIM required. See how AI plan review works, or if you coordinate trades, start with the MEP coordination process and the building plan review process. MEP subcontractors: see Flikt for MEP subs.