The fault that keeps coming back: finding the root cause on a busy shift

It's Tuesday night. The case packer stops with a jam at the infeed. The fitter clears it, resets and runs. Twenty minutes, job done.

It did the same on Friday. And twice last week. Each time someone cleared it, reset and wrote "jam cleared" on the sheet. Each fix took 20 minutes. Add them up and the case packer has lost most of a shift this month to a fault nobody has looked at properly.

Everyone knows it's coming back. Nobody has the time to find out why. The shift wants it running, and running is what they get.

Repeat faults survive because every fix is quick enough to feel fine. The cost only shows when you add them up.

Running and fixed are two different things

Getting the line running is the right call at 2am. Nobody's arguing with that. The trouble starts when running gets written up as fixed.

HSE's guide to investigating incidents puts it well. It says "simply dealing with the immediate causes of an adverse event may provide a short-term fix" (HSG245). The guide is written for accidents and near misses. But the idea carries straight over to a breakdown. If you only clear the jam, the cause is still there, waiting for the next shift.

The same guide splits causes into three layers. The immediate cause is what you can see: the jam. The underlying cause is the less obvious reason it happened. The root cause is "an initiating event or failing from which all other causes or failings spring", and it's often a management, planning or organisational failing (HSG245).

On the case packer, that might look like this. The jam is the immediate cause. A worn guide rail that lets cases skew is the underlying cause. And the root cause is that nobody owns the guide rails, so they've never been on the PPM.

Get the evidence while it's fresh

You can't find a pattern in faults nobody recorded. "Jam cleared" tells you nothing.

So the first job is to get each shift to write down a bit more, every time it happens:

1. Date and time, and how long it was down
2. The exact fault code or alarm text, with a photo of the HMI before reset
3. What product, pack size or format was running
4. What was found, in plain words: "case skewed at the infeed guide"
5. What was done to get it running
6. Anything odd: a changeover just before, a new operator, a hot day, a new batch of cases

It takes a minute more per fault. After two weeks you'll have something to work with. A good breakdown log is how you see the pattern.

The law backs this up, in a small way. PUWER requires that where any machinery has a maintenance log, "the log is kept up to date" (PUWER, regulation 5). The code of practice adds that a detailed log can "inform maintenance personnel what actions have been taken previously" (L22, paragraph 78). That's the point. The next fitter shouldn't have to start from scratch.

Look for what the faults have in common

Lay the records side by side. Look for patterns.

Does it happen on one product and not others? Then look at that format's settings and parts. Is it always the first hour after a changeover? Then look at the changeover. Is it nights only? Then ask what's different on nights: the crew, the temperature, the speed. Is it getting more frequent? Then something is wearing.

Then ask what changed. Faults that start suddenly nearly always start after something changed. A new supplier of cases or film. A replacement part that wasn't quite like for like. A speed increase. A software change. Nine times out of ten, in our view, the answer is in that list.

Some faults won't show while you're watching. If the fault is intermittent, our guide to catching the fault that won't show has ways to trap it.

Change one thing at a time

When you've got a theory, test it. Change one thing, then watch. If you change three things at once and the fault stops, you won't know which one fixed it. If it comes back, you won't know which one to undo.

And when you change a part, ask why the old one failed. A burnt contactor might mean a loose terminal, a sticking coil or a motor drawing too much. Fit a new one without looking and you'll be fitting another next month. The same goes for a motor that's burnt out.

Keep blame out of it

Repeat faults often point at people. The operator who loads cases badly. The fitter who set it up wrong last time. It's easy to stop there.

Don't. HSG245 is clear that investigations should be done with prevention in mind, "not placing blame" (HSG245). If the operator keeps loading cases badly, ask why. Is the magazine hard to load? Is there no training on the new format? That's where the root cause is. Our post on five whys without a blame session shows how to run it.

Our view: give it to someone with time

Here's what we think. Your shift team can't find the root cause of a repeat fault in the gaps between other breakdowns. They're too busy keeping things running, and that says nothing about their skill. There just aren't enough hours on a shift.

So give the repeat fault to one person, with time set aside to look at it. That might be your best fitter on a day shift with no other jobs. It might be a controls engineer, if the fault looks like a sensor, timing or PLC issue. If you don't have that person, bring one in for a day. Their day costs less than another month of 20-minute stops, and a fresh pair of eyes often sees what the team has stopped noticing.

For more on the methods, see root cause analysis for maintenance teams.

What to do on Monday

1. Pick your worst repeat fault from the last month. Count how many times it stopped the line and add up the hours.
2. Put the six-point fault record on a card at that machine, and ask every shift to fill it in.
3. Give the fault to one named person, with time booked to look at it.
4. Ban "reset and run" as a write-up. Every entry says what was found.

* Regulation and paragraph references correct as of 2 September 2026. Check PUWER on legislation.gov.uk and HSE's L22 for the latest. Paragraph numbers refer to L22, fourth edition (2014, amended 2018).