FMEA & maintenance planning
Every FMEA has a column your team is guessing. Measure it instead.
Severity, you know — it’s your line. Detection, you know — they’re your controls. But Occurrence — how often does this failure actually happen? — gets scored from the memory of whoever’s in the room. Acervas counts it from the maintenance history of every plant running your machine, and builds the worksheet for you.
FMEA is the right discipline. Rank every failure mode by Occurrence × Severity × Detection, work the worst first — no reliability engineer argues with the method. What the method can’t survive is bad inputs and a dead document. That’s not a failing of your team. It’s what happens when the analysis runs on memory and lives in a spreadsheet.
How FMEA gets done today — and where it leaks
The workshop
A meeting room, a spreadsheet, and the memory of whoever could make the session. Occurrence gets scored one-to-ten by feel.
The frozen sheet
Scored once, saved to a shared drive, reopened at the next audit. The plant keeps changing; the RPNs don’t.
The single-plant blind spot
Your history only holds the failures you’ve already survived. The modes you haven’t met yet aren’t on the sheet at all.
Occurrence is a measurement, not an opinion
Acervas counts failures per failure mode across every plant running the same make and model — anonymised, normalised per machine-year — and scores Occurrence on a disclosed one-to-ten ladder. The receipts sit right next to the score: how many failures, across how many machines, at how many companies, over the last 24 months. When someone asks where the eight came from, the answer is on the sheet.
From CMMS history to a ranked worksheet
- 1
Connect your CMMS — read-only
Work orders and PM history sync from the system you already run. AI labels each record’s failure mode — labels only. Every count on every worksheet comes from deterministic code, never from a model.
- 2
Findings surface, with receipts
Repeat faults with nothing scheduled against them. Overhauls followed too often by the same failure they exist to prevent. PMs targeting a failure your machine has never once had. Each finding states its evidence — how many, how often, over what window — and waits for your verdict: confirm it into the plan, or dismiss it.
- 3
Score what only you know
Occurrence is already on the sheet, measured. You add the two columns no fleet can score for you — Severity, because only you know what this failure does to your line, and Detection, because only you know whether you’d catch it. RPN ranks every sheet worst-first, and re-ranks after every sync.
The result isn’t a report — it’s a register. Verdicts survive every re-sync, confirmed findings stay tracked until they’re actioned in your own CMMS, and the worst risk in the plant is always at the top of the page when Monday’s maintenance meeting starts.
The failure modes you haven’t met yet
A single-plant FMEA can only rank the failures a plant has already had. On the network, your worksheet also carries the modes other plants have logged on the same machine — flagged fleet only, with a measured frequency — before the first one ever costs you a shift. The warning arrives ahead of the failure, which is the whole point of the exercise.
And it travels the way the rest of the network does: the fix and the failure count cross, the fingerprint doesn’t. Make and model travel; your company, plant, line and people never do. You opt in, you both give and get, and you can leave. How the knowledge layer works →
Built for lines where one machine stops everything
Ranked risk matters most where a single machine going down stops the chain behind it and the meter runs to thousands an hour — continuous, cascading lines. That’s the plant Acervas is built for, and the plant where a re-ranking FMEA earns its keep. See the industries →
Questions reliability engineers ask first
- Where does the Occurrence score come from?
- It’s counted, not estimated — anonymised failure records from every plant in the network running the same make and model, normalised per machine-year, mapped to a disclosed 1–10 ladder. The receipts sit next to the score: how many failures, across how many machines, at how many companies, over the last 24 months.
- Do we have to run FMEA workshops first?
- No. The worksheet builds itself from your synced maintenance history. You bring the judgement — Severity and Detection — not the data entry.
- Is our failure data shared with competitors?
- Failure counts are pooled by machine and anonymised. The make and model travel; your company, plant, line and people don’t. Joining the network is your choice, and reversible.
- Are the numbers AI-generated?
- No. AI labels each record’s failure mode — that’s all. Deterministic code does every count, and Severity and Detection are yours. No number on your worksheet was produced by a model.
- Does Acervas write to our CMMS?
- No — the connection is read-only. You action findings in your own system, then mark them actioned so the register keeps tracking them.
See how Acervas works with your CMMS, check the pricing, or talk to us about your line.
Stop losing knowledge. Start building on it.
Tell us about your plants. We'll onboard your first one free for 90 days — your engineers are capturing knowledge inside a week.