Squinting at a Decade-Old Scanned Drawing Until I Found the 100 Millimeters Nobody Caught

The MRT station on our current project was built over a decade ago, station and tracks fully operational, tunnels running underneath the city like they always had. What most commuters walking through it have no idea about is that a section was deliberately left as a bulkhead back then, a dead end built into the original structure specifically so a future tunnel could connect into it someday. That someday started in 2019, and it's still going, and I've been on it long enough now that I sometimes forget the tunnel wasn't always half-finished.

Connecting a brand new structure to something built ten-plus years ago sounds straightforward until you actually try to do it. The new tunnel's roof slab and base slab need to physically align with the old bulkhead's levels, millimeter for millimeter in theory. In practice, the only record we had of the original structure was a scanned drawing set from over a decade ago, low resolution, text barely legible in places, the kind of document where you zoom in at 400 percent and still can't tell if a number is a 3 or an 8, and you just sit there squinting until your eyes hurt.

Squinting at a Number for the Fourth Time

I built out a model from what little clean information those old scans actually gave us, cross-referencing every dimension I could pull against whatever metadata had survived a decade of file transfers and format changes. Somewhere in that process, a gap showed up between the design team's assumed levels and what the old bulkhead's actual roof and base slab sat at. About 100 millimeters off.

I didn't trust it at first. Assumed I'd misread the scan, checked it again, then a third time, then pulled up the original PDF on my second monitor next to the model just to triple-confirm I wasn't the one making the mistake. It was late afternoon by the time I was sure, the office mostly quiet, everyone else already thinking about heading home, and I remember just sitting there for a second not saying anything to anyone, half annoyed nobody had caught it earlier and half pleased it was me who finally did.

A hundred millimeters doesn't sound dramatic written down like that. On a tunnel connection, it's the difference between a clean structural tie-in and a section that physically doesn't meet where the drawings say it should, and there's a real difference between catching that on a screen at five in the afternoon versus an excavator operator finding out the hard way three months later with concrete already poured on one side.

The Boring Numbers That Actually Matter Most

Old, imprecise records bleed into other parts of the job too, the parts that never show up in any progress report anyone reads. The Final Excavation Level has to be exact, not approximate, the depth an excavator stops at before blinding concrete, waterproofing, and the tunnel foundation go in. Dig too shallow and the structure sits wrong. Too deep and the foundation bearing is already compromised before a single thing's been poured. There's no rounding error you can quietly absorb there.

Rebar in the base and roof slabs has to land at exact levels where it ties into the D-wall couplers too, which meant pulling precise values out of haunch details across drawings that didn't always agree with each other, and handing that over before anyone ordered steel. Same story with cut-off levels for the bored piles, and the king posts holding up the roof slab mid-construction. None of it is interesting to talk about at a dinner party. All of it is the kind of thing that, done wrong, doesn't show up as a line item anywhere, it shows up as someone hacking concrete back out of a slab that's already cured.

A Utility That Kept Moving on Paper

A different problem came up around a pedestrian overhead bridge going in at one of the tunnel portals. Its foundation and lift shaft needed a specific footprint, except by the time I checked, the construction sequence had already diverted the utilities that used to run through that exact spot, diverted them somewhere that now sat right where the new foundation needed to go.

Going back through the drawing history to figure out how that happened, I found something that bothered me more than a simple mistake would have. In a couple of revisions, that utility's position had been nudged, just enough to clear the new footprint on paper, no note explaining why. In other versions it was missing entirely. I genuinely don't know if someone shifted it on purpose to make a problem quietly disappear on screen, or if it was just careless revision control by someone in a hurry. Either way, it made me go back and check three more utility positions near that same section out of pure paranoia, and one of those turned out fine, but I didn't trust "fine" until I'd actually confirmed it myself.

A Pipe, a Roof Slab, and a Deadline That Almost Won

One more, smaller but still real. A 900mm water main sat directly above where our Tunnel Jet Fan section's roof slab needed to go, and the construction sequence hadn't accounted for diverting it before that slab was due to be poured. We were maybe two weeks out from that pour date when this surfaced, which is not a comfortable amount of runway. Instead of forcing a rushed diversion, we worked with the design team on extending the roof slab section instead, routing around the pipe rather than moving it. The client accepted it. The pipe stayed put. Nobody had to explain a schedule slip to anyone above them.

Looking back at that whole stretch, I don't think any single one of these felt heroic while it was happening. Mostly it felt like a lot of squinting, a lot of double-checking things I'd already checked once, printing paper I probably didn't strictly need to print. But a decade-old scan, a nudged utility, a pipe nobody had scheduled around, none of those turned into an actual site problem, and that's really just because someone was tired enough, paranoid enough, and stubborn enough to keep checking past the point where it would've been easier to assume everything was fine.

Related reading:

The Duplicate Family That Cost Us Two Weeks on a Tunnel Ventilation Shaft

The Coordinate Mistake That Sent Our Tunnel Two Kilometers Off Course