Condition monitoring
From first signal to teardown: the evidence chain for a bearing decision
A repeatable way to turn a weak vibration signal into a maintenance decision that people trust.
Why a chain
A single high reading rarely gets a machine stopped. A chain of evidence does: the fault is named, followed, judged and then confirmed. Each link makes the next decision easier.
The five links
| Link | Question | What to record |
|---|---|---|
| 1. Identify | Which frequency, which part? | The frequency in hertz and in orders, the harmonics, the sensor location |
| 2. Repeat | Does it change? | Date, speed, load and the same amplitude measure every time |
| 3. Judge | How fast is it getting worse? | The ratio between readings and the time between them |
| 4. Recommend | When should the bearing be changed? | A date, the spare, the stop needed |
| 5. Confirm | Was the diagnosis right? | Photos of the damage, which ring, how large |
Rules of thumb
- More than double in under two weeks is fast. Plan the change.
- A clean series of harmonics on a heavy, slow machine is worth acting on even at a low level.
- Change in speed or load can change the amplitude. Compare in orders and note the conditions.
- Use the same sensor position and the same band each time.
Common mistakes
- Alarming on the overall level and ignoring the pattern.
- Comparing readings taken at different speeds without correcting for it.
- Changing the bearing and not opening it, so the diagnosis is never checked.
- Leaving the amplitude measure undefined, so two readings cannot be compared.
Takeaway. Name the fault, follow it, judge its speed, recommend a date, then open the bearing. That chain is what makes monitoring believed.
Put it to work in the tool