Can switchgear be tested without a shutdown?

6 min read · Updated 30 July 2026

There is a quiet irony in how electrical testing is usually scheduled. The plant waits for a shutdown window so the boards can be checked "properly" — and the moment the supply goes off, the two most consequential failure mechanisms disappear from view. The de-energised test day measures a system in a state it never operates in.

The physics does the arguing.

A failing connection announces itself as heat — but only while current flows through it. Kill the load and the joint cools to ambient within the hour, looking exactly like its healthy neighbours. Failing insulation announces itself as partial discharge — tiny sparks inside or across the insulation system — but only at operating voltage. De-energise the board and the discharge stops with it.

So "can we test without a shutdown?" has the logic backwards, at least for the screening layer. The running plant is not an obstacle to the measurement. It is the test condition.

What runs live.

Between them, the two methods cover the two ways a switchboard actually fails — the current path and the insulation system — with the plant running and the panels closed. One survey day, no outage request, findings rated in writing.

What a live screen proves — and what it does not.

Honesty about the boundary keeps the rest credible. A live screen detects and prioritises: it tells you which assets carry indications, how severe they read, and what deserves action on what timescale. It does not pinpoint the exact defect inside an insulation system, and it does not certify a board healthy in the way a full diagnostic work-up can. The screening fee should never quietly become a repair bill — a finding justifies investigation, and the investigation justifies whatever comes next. Each step should stand on its own evidence.

What still needs an outage — and how the screen shrinks it.

Some measurements are physically impossible on a live board: insulation resistance testing puts test voltage on de-energised conductors; protection relay verification injects secondary currents; contact resistance needs the circuit isolated; and mechanical servicing needs dead equipment by definition. None of that disappears because the screening ran live — it moves to where it belongs, inside a planned maintenance window.

The sequencing is the point. A facility that screens live first walks into its outage window with a ranked list: these two boards carry the serious findings, this relay protects the asset that matters, this joint gets re-checked after repair. The shutdown work gets shorter, targeted, and justified line by line — instead of a blanket "test everything while it's off" that consumes the window and still misses what only shows up under load. That escalation logic — screen live, deepen where the evidence justifies it — is how our assessment programme is built, declared interest and all. The order holds whoever does the work: measure the running system first, and spend the outage on what it found.