The Crew That Built Off a Superseded Sheet

Why this matters

A crew that builds from an old sheet does good work in the wrong place, and the rework is almost always the shop's, because the revision was issued and the shop received it. This case is reconstructed from the record afterward: the email trail, the truck set, and the inspector's note. The interesting part is not that the sheet was old, but how many separate moments the new one was in somebody's hands without reaching the crew, and what the shop's records could and could not prove about it.

The correction notice

A light-commercial HVAC shop had the mechanical scope on a single-story fitness studio fit-out: steel joists, a metal deck, two new rooftop units. In job week 8 the building inspector did the above-ceiling rough-in inspection of the west half and wrote a correction: the supply main from rooftop unit RTU-2 was "not installed per approved revised drawings, M-201 Rev 2." Specifically, the main hung 6 in higher than the revised sheet showed along its whole run, and it ended at a drop location 12 ft west of where RTU-2's curb now sat.

The crew had hung 46 ft of rectangular supply main, west to east, exactly as their sheet drew it. Their sheet was M-201 Revision 1, the set issued for permit and construction before work began. Revision 2 had been issued three weeks earlier.

Reading the record backwards

The shop had no sheet log. Everything below came from email, the general contractor's (GC) document portal history, the truck, and the people involved. Each step answers one question: where was Revision 2 on that day?

TIMELINE   job weeks; source of each line in []
wk 1  Rev 1 set printed for truck, 2 copies of
      M-201: one in set, one taped up for layout
      [print shop receipt]
wk 5  Bulletin 2 issued by architect via GC
      portal: RTU-2 moved 12 ft east, new beam
      at new opening, M-201 Rev 2 reroutes main
      6 in lower, drop moved 12 ft east
      [portal history]
wk 5  Portal notice to shop PM: not opened,
      notifications filtered to a folder [PM]
wk 5  GC super emails PDF to PM, 2 days later
      [email]
wk 5  PM forwards to lead's phone: "FYI
      bulletin 2, mostly structural" [email]
wk 6  Steel erector sets new beam at the new
      opening; roofer cuts the new opening
wk 7  Crew hangs 46 ft of main from Rev 1
      [daily log, labor hours]
wk 7  Roofer sets RTU-2 curb at new opening
wk 8  Rough-in inspection: correction issued
      [inspection report]

Week 5, the portal. The GC's portal posted Bulletin 2 and sent the shop's PM a notification. The PM's mail rules filed portal notices in a folder they checked weekly, and the bulletin sat there. This is the first gap: the shop's official receipt of the revision, which the GC could later prove to the day, was invisible to the shop.

Week 5, the email. Two days later the GC's superintendent emailed the bulletin PDF directly. The PM read the cover page. Bulletin 2's narrative listed three sheets, S-101, S-501 and M-201, and described the M-201 change as "revised RTU-2 duct connection". The PM read that as a fitting change at the unit, a detail the crew would sort out at the curb, and forwarded the PDF to the lead with the line "FYI bulletin 2, mostly structural." That phrase is the hinge of the whole case. Two of the three sheets were structural, which made the sentence true by count and wrong about the one sheet that mattered to the shop.

Week 5, the phone. The lead opened the PDF on a phone. M-201 is a 24 by 36 inch sheet. On a 6 inch screen at full-page view, a revision cloud, the scalloped line drawn around a change, is a faint wavy edge a few pixels wide, and the revision triangle beside it is smaller still. The lead remembers scrolling the structural sheets, seeing a new beam, and closing it. The forward's subject line had told the lead what to expect, and the lead saw what it said.

Weeks 1 to 7, the truck. The truck set held M-201 Rev 1, plus a second copy taped to a sheet of plywood as the layout copy. The revision block, the table in the title block listing each revision's number, date and description, read "1 PERMIT AND CONSTRUCTION" on both. Rev 2's block read "2 BULLETIN 2". Nobody compared them, because nothing in the shop's routine asked anyone to read a revision block. See related: The Revision Block and Why It Decides Everything, which covers how to read one.

Week 6, the field. The new beam and the new roof opening were in place before the crew started. Why didn't the crew notice? The main ran west to east, and the crew stopped at the old drop location, 12 ft short of the new opening. The opening was in an area they had not reached, and on Rev 1 an exhaust fan opening sat in roughly that part of the roof, so a hole there did not look wrong. The main never touched the beam. The field gave the crew no physical clash to stop them.

Ruling out the other explanations

Before the shop accepted the record above, it checked three other readings, because each would have put the cost somewhere else.

The GC failed to distribute the revision. The portal history showed delivery to the shop's PM in week 5, and the superintendent's email was a second delivery. The revision reached the shop twice. This one was dead the moment the portal record was pulled.

The revision was not clouded, so a careful reader could have missed it. Rev 2, printed full size, had a cloud around the entire main route and a triangle 2 tag at each changed dimension. The change was marked properly.

The inspector was holding a set the shop never got. The inspector's set was the GC's permit revision submittal, which included Bulletin 2's sheets exactly as posted to the portal. Same sheets.

What remained was the shop's own handling: received twice, summarized wrongly once, viewed on a screen too small to show the mark, and never printed.

What the rework cost, in hours

The original 46 ft run took 2 technicians 13 hours each, 26 labor hours. The rework took 2 technicians 15 hours each, 30 labor hours. They dropped the main, re-cut every hanger rod 6 in longer, rehung it at the Rev 2 elevation, scrapped the drop fitting at the old location, and extended 12 ft to the new drop with new duct and a new fitting. The rework hours were 30 / 26 = 1.15 times the hours of the original install, so the run consumed 26 + 30 = 56 labor hours, 2.15 times what it should have. Add one reinspection trip and 3 working days of ceiling schedule that the drywall contractor lost, which the GC noted in its schedule update and the shop had no grounds to dispute.

What it did not cost, and why. The main had not yet been insulated or connected to the branch ducts. Had the inspection come a week later, after branches, insulation and the ceiling grid on the west half, the same correction would have meant taking down branch connections and reworking insulation as well, which the PM estimated at roughly twice the hours actually spent. Early inspection limited the loss. It does not count as a control.

What the record could not prove

Reconstructing the case exposed how thin the shop's own record was, and the gaps were findings in themselves.

Nobody could say when the lead first saw Rev 2. The forward proved the PDF reached the lead's phone in week 5. It did not prove it was opened, which sheets were viewed, or for how long. The lead's memory was the only evidence, and it was honest, but memory is what a dispute attacks first.

Nobody could prove which revision was on the truck on the day the main was hung. The shop could show it only because both Rev 1 copies were still on the truck in week 8. Had someone tidied the truck after the correction, there would have been no physical proof the crew built from an issued sheet rather than a guess, and the GC's first question was exactly that.

The labor hours were provable because the daily log was good. The crew's daily log recorded hours by area and task, so the 26 original hours and the 30 rework hours could be separated cleanly. A log that recorded only total hours per day would have left the cost of the error as an estimate.

The SOP below closes the first two gaps on purpose: the truck sheet log records the revision in the crew's hands by date, and superseded sheets are stamped and archived rather than thrown out.

Why being careful would not have been enough

Every person in the chain was competent and none was careless in any way they would have recognized at the time. The PM read the narrative, which is a reasonable thing to read. The lead looked at the PDF, which is more than many leads do with an FYI. The crew built exactly what their sheet said. The failure was structural: the shop had no step at which a revised sheet had to be printed, compared to the truck set, and swapped, and no step at which anyone read the revision block on the sheet in their hands before building. A procedure that depends on someone noticing will fail on the day the summary is misleading, and summaries are misleading often enough.

The three moments that would have stopped it, each cheap:

• When the PM received it, logging each listed sheet against a sheet log would have shown M-201 going from Rev 1 to Rev 2, a mechanical sheet whatever the narrative said.
• When the lead received it, a printed full-size Rev 2 in place of Rev 1 makes the cloud impossible to miss, because it surrounds the entire run the crew was about to hang.
• When the crew started the main, a two-minute check of the title block revision against a log, before layout, would have found the mismatch in week 7, before the first hanger.

The routine the shop adopted

The shop now runs The Drawing Revision Control on the Truck SOP on every job with drawings. In short: one controlled set per job held by the lead, no taped-up layout copies; every revision logged by sheet the day it arrives, with the PM reading every cloud rather than the narrative; superseded sheets pulled the day the new one is delivered and stamped; a daily revision check before layout; and no building from anything on a phone. The PM also changed the mail rule so portal notices arrive in the inbox, and the shop now treats the portal posting date, not the email date, as the date of receipt, because that is the date the GC will cite.

One change the shop considered and rejected: asking the GC to phone about every bulletin. It shifts the shop's control to someone else's memory, and the portal already proves delivery. The fix belongs inside the shop.

References

• See related: The Revision Block and Why It Decides Everything
• See related: The Drawing Revision Control on the Truck SOP
• See related: How to Confirm the Truck Set Is Current
• See related: How to Trace a Duct Run to the Last Diffuser