The first hour of a breakdown, and why it's never about knowing what to do

It's 2am. The palletiser's stopped with a fault on the HMI nobody recognises. Your shift fitter is on it. The shift manager is stood behind them asking how long. Trailers are booked for six.

Nobody on that floor needs telling how to lock off a machine. They've done it a thousand times. But it's 2am, someone's asking how long, and the guard's already half off.

Breakdowns go wrong because of pressure. Your team knows what to do. In the first hour, with the line stood and the phone going, they stop doing it.

Don't trust the e-stop

The oldest shortcut on any shop floor starts with "it's on the e-stop, it's fine."

It isn't. HSE's code of practice for PUWER is clear that an emergency stop or an interlocked guard is not isolation (L22, paragraph 195). An e-stop is a control. It can fail, be reset by someone who doesn't know you're in there, or leave stored energy in the machine.

Isolation means a secure break in the supply. That's the isolator off, a lock on it, and a tag with a name. If two of you are in the machine, you each fit your own lock. The code of practice suggests a hasp that takes several, so nobody can switch back on until the last lock comes off (L22, paragraph 197).

Then think about what's still in the machine after the power's gone. Air in the receiver. Pressure in a hydraulic accumulator. A lift table that's up. A spring that's loaded. Caps in the drive that haven't discharged yet. The code of practice gives the example of moulding machines that can move on stored hydraulic energy under fault conditions (L22, paragraph 173). One more trap: closed isn't always safe. Some drains and vents need locking open, not shut (L22, paragraph 196).

None of this is new to your team. The point is that at 2am, with someone asking how long, it's the first thing to slip.

Photograph the HMI before anyone touches reset

The fault history on the screen is the best clue you'll get. The first thing anyone does is press reset to "see if it clears". Sometimes it does. Then it trips again at 4am and the evidence is gone.

So make it a rule: photo of the HMI and the alarm list before reset. Then get the operator's version while it's fresh. What was it doing? Any noise, smell or smoke before it stopped? What's been tried already?

While you're there, ask one question. Could this be reportable? Most breakdowns aren't. But a lifting equipment failure, such as a load-bearing part letting go, is a RIDDOR dangerous occurrence even if nobody was hurt. So is a pressure system failure that could have killed someone (HSE, dangerous occurrences). If it might be one, keep the scene as it is until you know. The report has to reach HSE within 10 days* (HSE).

"Just a quick look with it live"

This is where good fitters get hurt.

The panel's open, the fault's intermittent, and the fastest way to find it is with the power on. We understand the temptation. But the Electricity at Work Regulations only allow live work when it's unreasonable for the kit to be dead, reasonable to work on it live, and proper precautions are in place (regulation 14). "Production needs it back" doesn't pass the first test.

There's a difference between live testing and live working. Proving a supply with the right meter, the right probes and the right person is planned work. Poking around a live panel to see what happens isn't. If your team needs to fault-find live, plan it and brief it. HSE's HSG85 is the guide to work from.

Make the call by minute 30

Here's our view. Most sites wait too long to get outside help.

Nobody decides to wait. There's ten minutes of "one more thing to try", then another ten, and suddenly it's 4am. Nobody writes down the hour spent hoping.

Fix it before the night it happens. Work out your cost per hour of downtime on a quiet afternoon. Lost output times margin, plus idle labour, overtime to catch up, scrap on restart, and any late delivery penalties in your contracts. Once you have that number, the 2am question gets easier. Is another hour of trying cheaper than a callout? Put your own figures in and the answer is usually obvious.

Then set the trigger. If it's not diagnosed in 30 minutes, the shift calls for help. You need outside help when:

• it's a PLC, drive or servo fault and nobody on shift is confident with it
• it's the same failure as last month and nobody found the root cause
• your fitter is already spread across two other jobs
• the kit is hired, and the hire firm may be responsible for its maintenance. That should be agreed in writing up front (L22, paragraph 77)

HSE expects maintenance to be done by people competent for that job (L22, paragraph 79). A good mechanical fitter isn't automatically competent on a servo drive. Saying that out loud protects them, and your shift managers need to hear it from you.

The contractor is on site. You're still responsible

Your duties stay with you when someone else picks up the spanners. You have to give the contractor's employer clear information about the risks on your site and how you control them (Management Regulations 1999, regulation 12). When two employers share a site, even for one shift, you're both expected to co-operate (regulation 11).

In practice that means a proper brief, sent before they set off rather than shouted at the gate:

1. Machine make, model and serial, and where the manuals are
2. What it was doing when it stopped, with the HMI photos
3. What's been tried and what changed
4. Isolation points, and who holds the locks
5. Site rules: induction, permits, PPE, hot work
6. Who signs off the restart

HSE's short guide Using contractors (INDG368) has a good checklist of what to ask them.

Back running is not the same as fixed

Locks come off when the last person confirms they're clear. Guards go back on. Test run with everyone stood back.

Then comes the bit most sites skip: write it up while you still remember it. Under PUWER, a machine that has a maintenance log must have that log kept up to date (regulation 5). More usefully, the write-up is your evidence. One more point from the code of practice is worth keeping for budget meetings. When PUWER says equipment must be kept "efficient", it means efficient for safety, not for output (L22, paragraph 70). So a machine that keeps failing is a safety problem as well as a production one. Put that in front of the finance director and see how fast planned work gets funded.

What to do on Monday

1. Work out your cost per hour of downtime for your three worst machines.
2. Agree the 30-minute trigger with your shift managers, and write it down.
3. Make the HMI photo before reset a rule on every shift.
4. Have an outside engineer lined up before you need one.

* RIDDOR reporting deadline correct as of 25 September 2026. Check HSE's RIDDOR pages for the latest. Paragraph numbers refer to L22, fourth edition (2014, amended 2018).