How to improve switchgear build traceability

BS EN IEC 61439 is a verification standard. It does not just describe how an assembly should perform — it expects you to show that a specific assembly meets the design through design verification and routine verification. The routine checks on every unit (wiring, dielectric, protective-circuit continuity, operation) only count if you can point to the record for the panel that shipped.

So the useful trail ties the design to the physical build and to the checks. Each link answers a question a client, an inspector or your own QA might ask later:

The records usually exist. They are just not connected to the job. A torque sheet in a job bag, an inspection note on a clipboard, an approval given verbally at the bench, a test cert saved to a shared drive under a filename only one person understands. Each is fine on its own. None is retrievable against the panel months later without a hunt.

The second failure is revision drift. The panel is built to a drawing that was superseded mid-job, and nothing in the record shows which issue was in force when the wiring went in. When the as-built and the latest drawing disagree, you cannot tell whether it is an error or an undocumented change.

The third is reconstruction. Nobody logged the evidence as the work happened, so at handover someone walks the floor asking who tightened what and whether the ACB was function-tested. Answers written down after the fact are weaker evidence and sometimes wrong. If a panel is queried a year later — an intermittent fault, a warranty question, an insurer's request — the memory has gone and the trail with it.

A good trail is captured once, at the point of work, and never assembled afterwards. Each stage is signed off against the panel as it completes, so at any moment you can see where the job is and what has been proven.

The design revision is locked to the job, so everyone on the floor works from the same current issue and a change forces a visible re-release rather than a quiet overwrite. Torque, inspection and test evidence attach to the panel, not to a person's clipboard. A raised non-conformance blocks sign-off until it is closed, so a panel cannot quietly ship with an open defect.

The test is simple: pick any finished panel and ask what it was built to and whether it passed. If the answer is a lookup rather than an investigation, the trail is working.

Whatever system holds it, a defensible trail for an LV assembly ties these together against the job:

Take a shop turning out around 30 distribution and control panels a month across a dozen live jobs. A client rings about a panel delivered nine months ago: a breaker has tripped and they want to know the torque setting used on the main incomer terminations and whether the panel was function-tested before dispatch.

On the loose-paper method, someone opens the archived job bag, finds a torque sheet that may or may not name the specific terminations, checks the drawing but cannot be sure which issue was live when the panel was wired, and asks the tester — who left in the spring. Half a day gone, and the answer still carries a caveat.

With evidence captured at the point of work, the same query is a lookup. Open the panel's record: it was built to drawing issue C, the incomer terminations were signed off at the specified torque by a named operative on a given date, routine verification and FAT passed and were signed off, and no non-conformance was open at dispatch. Five minutes, and the answer is evidence rather than recollection. The shop that logs as it builds is not doing more work — it is doing the same checks once and keeping them, instead of reconstructing them under pressure later.

Software holds and organises the evidence so it is complete and findable. It does not replace your own inspection, verification or type-test duties under BS EN IEC 61439 and BS 7671 — it makes meeting them, and proving you met them, far less painful. The standards are the bar; the trail is how you show you cleared it.