Treadstone Associates
Case File · Field Operations & Documentation

Site team catches a superseded drawing revision

Anonymised, illustrative composite. A structural revision changed a beam-pocket dimension on one floor of a BC mid-rise. Three days later, two copies of the old drawing were still circulating in the field — until a status flag the drawing office had built itself caught one of them before hangers were cut to the wrong measurement.

Treadstone Associates · Updated 2026

At a glance

  • • BC mechanical/sheet-metal sub, structural steel Rev D superseded Rev C mid-project.
  • • Rev D changed a beam-pocket dimension affecting duct hanger locations on one floor.
  • • Two sets of Rev C were still circulating in the field trailer and on one foreman’s tablet three days after Rev D issued.
  • • Drawing office had built a custom “Superseded” status into the markup list, beyond Bluebeam Revu’s two fixed defaults.
  • • Avoided cost: an estimated $13,800 in refabrication for 46 hanger assemblies, plus roughly 4 working days of lost float.

The situation

Structural revisions happen mid-project often enough that most crews stop thinking of them as unusual — a beam pocket moves, a dimension changes, a new sheet goes out. What makes a revision dangerous is not the change itself but the interval between when it issues and when every printed or cached copy of the old sheet has actually been replaced. On this job, that interval ran at least three days, and during it, two copies of the superseded Rev C were still in active use: one pinned in the field trailer, one cached on a foreman’s tablet.

Both stale copies looked entirely legitimate. The trailer print had the same title block, the same sheet number, the same general layout as the current version — nothing about it visually announced that it had been replaced three days earlier. A crew working from either copy would have had no reason to suspect it was wrong, which is exactly what makes an unflagged superseded sheet more dangerous than an obviously outdated one.

The problem

Bluebeam Revu ships with exactly two unchangeable default status models for markups — “Review and Migration” — and lets a firm build its own custom statuses on top, with their own colors and labels. Nothing about “current versus superseded” is a system default anywhere in the tool; a firm has to define that distinction itself or nothing distinguishes an old sheet from a current one on a screen. The same is true of the Stamp tool, which ships with illustrative examples like “DRAFT” and “FOR REVIEW” rather than a fixed library of construction-specific designations. This drawing office had already built a custom red “Superseded” status into its markup list for exactly this reason, and paired it with a visible stamp on every sheet a revision replaced.

The numbers

When Rev D issued, the drawing office flagged Rev C “Superseded” and stamped it the same day. The foreman’s tablet synced that status change the following morning — before his crew began cutting hanger rod to the old beam-pocket dimension. Without that flag, an estimated 46 hanger assemblies would have been fabricated to a dimension the structural drawing no longer supported: roughly $13,800 in material and labour to refabricate, plus about 4 working days of lost float on that floor’s mechanical rough-in while the correct hangers were remade.

The trailer print of Rev C was pulled and physically marked the same day the tablet synced, closing off the second stale copy before it could be used. Neither copy had been touched by the crew in the intervening three days — the catch came from the status check happening before cutting started, not from anyone noticing an error partway through fabrication.

The rule that decided it

The tools involved — markup statuses, stamps — do exactly what they are built to do regardless of who uses them. What separates a firm that catches a superseded sheet from one that doesn’t is not the software; it is whether a specific, deliberately-built “current versus superseded” distinction exists at all, and whether it gets applied the same day a revision issues rather than whenever someone gets around to it. Revu will not flag an old sheet on its own. Nothing will, unless a firm has already decided that distinction matters enough to build and enforce.

Nothing about this near-miss would have surfaced quickly if the hangers had actually been cut wrong: BC’s Limitation Act gives a claimant up to 2 years from discovering a problem to bring it to court (s. 6(1)) — a superseded-dimension defect built into concealed ductwork is exactly the kind of thing that might not surface until well inside that window.

The outcome

No refabrication was needed. The document-control procedure now requires every superseded sheet to be status-flagged and stamped the same day a revision issues, before redistribution to the field. See how as-builts differ from record drawings, and for a related documentation discipline on the same kind of job, see how one GC cut its RFI turnaround time.

For how firms are using AI tools to catch superseded revisions automatically, see AI-assisted drawing revision checking.

Takeaways

  • • No markup or drawing tool ships with a built-in “current versus superseded” distinction — a firm has to define and enforce that status itself.
  • • The gap between when a revision issues and when every field copy is actually replaced is where rework risk lives; close it the same day, not eventually.
  • • A cached copy on a tablet is just as dangerous as a printed sheet in a trailer — syncing status changes fast matters as much as flagging them at all.
  • • Pair a status flag with a visible stamp: a status a system can filter on, and a mark a person glances at and understands immediately, catch different kinds of mistakes.
  • • A superseded sheet with an unchanged title block and sheet number looks identical to the current one — nothing about its appearance warns a crew it has been replaced.
  • • Print copies pinned in a field trailer are as much a document-control liability as digital ones; a status-flag policy has to cover both or it only solves half the problem.

See where AI pays off first in your business.

A 30-minute call is enough to tell you whether AI pays for itself here.