QR codes on industrial equipment: identification, and its honest limits

The problem tags solve

Walk any pump house and you'll find the setup for the error: two identical pumps a metre apart, twin filters in parallel, four identical transmitters on one rack, the same valve at both ends of a line. Now ask a new starter — or a tired veteran at 03:00 — to record a reading against “Feed pump B”.

Recording against the wrong equipment is the quiet data disaster of operator rounds. It doesn't look like an error; it looks like a normal reading. But it poisons everything downstream: pump A's history now contains pump B's numbers, the trend an operator might have caught is smeared across two assets, and when B fails, the record says it was healthy. Paper has no defence against this at all. An equipment identity tag — a printed QR label fixed at the point of reading — is the cheap, boring, effective one.

What a scan proves

Precision matters here more than anywhere else in this field, so here is the exact claim a QR scan supports: this device recorded this entry while physically reading this specific label, at this moment.

That single fact eliminates the wrong-equipment error — the record is bound to the label in front of the device, not to a line on a form. It also makes wholesale fabrication awkward in a way paper never can: filling in a round from the mess room requires the labels, and the labels are out on the plant. Modest claims, honestly kept, are what make the record defensible later.

What a scan never proves

The industry oversells here, so the list is worth stating baldly. A scan does not prove:

• Who held the device. Accountability comes from sign-in and assignment, the ordinary way — not from the tag.
• That anyone inspected anything.A scan places the device at the label; the looking, listening and judging remain the operator's work, and no marker technology changes that.
• That the person stayed. Scan-then-walk-away is physically possible in every tagging system ever deployed.
• That the equipment is safe. A tag is an address, not an assessment.

QR vs NFC, RFID and engraved tags

Every marker technology makes the same identification claim with different trade-offs:

• Engraved tags and manual entry — indestructible, readable by eye, and transcribed by hand, which reintroduces the exact error the tag was meant to kill. Fine as the human-readable backup on the same plate; weak as the primary mechanism.
• Barcodes— QR's older sibling. Works, but QR adds error correction (scans through grime and scuffs) and higher data density in a smaller square. There's rarely a reason to prefer 1D today.
• NFC— no line of sight needed, reads through dirt and in the dark, pleasant with gloves. Costs more per tag, needs NFC-capable readers, and is invisible to the eye — you can't tell a dead tag from a missing one without a reader. A good complement where grime or darkness defeats cameras; not a replacement.
• RFID (longer range)— reads from metres away, which is precisely the problem: “the reader was near this rack” is a much weaker identification claim than “the camera was pointed at this label”. Excellent for inventory; wrong semantics for point-of-reading identity.
• GPS / indoor positioning— useless under steel and concrete, and at best proves area, never equipment. Treat any “GPS-verified inspection” claim with suspicion.

QR's winning combination is mundane: printable on site for pennies, readable by any camera, visibly inspectable, error-corrected, and replaceable in minutes. For point identification on a round, mundane wins.

Deploying tags that survive a real plant

• Fix the tag to the point of reading, not just the asset. The operator scans where they stand. A tag on the far side of the skid gets scanned from memory — which defeats it.
• Never label removable covers. Covers, guards and cassettes get swapped; the tag walks away with them and the identity lie begins. Fix tags to the permanent structure.
• Choose materials for the environment — laminated stock, UV-stable print outdoors, placement out of jet-wash lines and away from surfaces that get repainted.
• Carry the human-readable identity on the same label. The equipment reference in plain text beside the code, so a dead phone doesn't strand the operator and a human can sanity-check what the code claims.
• Make damaged tags a finding. Torn, painted-over or missing labels get reported through the round itself and replaced promptly — a tagging system nobody maintains decays into scenery within a year.

The register behind the tag

A QR code is just an address. Everything it's worth depends on the register that answers “address of what?” — and that register needs the same discipline as any other operational record. The binding between a tag and its equipment is itself a record: who attached it, to what, when; and when a damaged tag is replaced, the history must show the succession rather than silently rewriting it. Otherwise the tag's testimony is only as good as an uncontrolled spreadsheet — which is to say, as good as the paper it replaced.

Done properly, the pieces sum to something quietly powerful: a scan binds the entry to a label, the register binds the label to equipment, and the record preserves both bindings untampered. None of it authenticates a person; all of it makes the record answer the auditor's real question — “how do I know this reading is about this equipment?” — with something better than trust.