The machine tells you before it stops
Most failures are not sudden. A bearing gets louder for weeks; a filter clogs a little every day. Put a threshold on the reading and the work raises itself while the fix is still cheap.
Condition monitoring
Live readingsPUMP-07Vibration
3.9 mm/sWarn at4 mm/sAlarm at4.5 mm/sTrendRising since TuesdayCH-02Approach temp
1.4 °CWarn at2 °CAlarm at2.8 °CTrendStableAHU-14Filter Δp
210 PaWarn at180 PaAlarm at250 PaTrendPast the change pointGEN-01Oil pressure
3.1 barWarn at2.5 barAlarm at2 barTrendNormal
Watch the reading, not the date
Condition thresholds raise work when a machine actually needs attention, not when a calendar says so.
Warning before alarm
Two levels: one that schedules an inspection in normal hours, one that says stop and deal with it now.
Rules you can explain
Plain thresholds an auditor can read, not a score from a black box nobody can question.
Rules you set once
You know your plant better than any model does. The system just watches the numbers tirelessly.
- When a reading crosses a level, raise a specific job
- Different levels raise different priorities
- Parts are reserved at the same moment the job is raised
Rules you set once
4 activeWHEN Vibration above 4.0 mm/s
THEN Raise an inspection
Raised asMedium priorityWhy this levelEarly enough to schedule in normal hours.WHEN Vibration above 4.5 mm/s
THEN Raise a repair
Raised asHigh priorityWhy this levelBearing wear is now likely; do not wait for the next PM.WHEN Filter Δp above 180 Pa
THEN Raise a filter change
Raised asMedium priorityWhy this levelReserves the filter from stock at the same time.WHEN Runtime past 500 hours
THEN Raise the service PM
Raised asPlannedWhy this levelCounts hours actually run, not days elapsed.
The signal was always there
A trend shows what a single reading cannot: that this started weeks ago.
- Every reading kept against the asset, not just the last one
- The trend travels with the work order to the technician
- A single spike does not raise a job — a sustained move does
PUMP-07 vibration
12 weeksCrossed alarm in week 12
The bearing did not fail on week 12. It had been telling you since week 5 — nobody was reading it.
It works while nobody is watching
The value is in what happens at two in the morning, when no one is looking at a screen.
- Threshold crossed, work order raised, parts reserved — unattended
- It lands on the morning list with the trend attached
- A planned change at 07:00 instead of a failure at 14:00
What happened overnight
No one was awakeReading crossed 4.5 mm/s
02:14Sensor
Fifth consecutive reading above alarm — not a single spike.
Work order raised
02:14Automatic
WO-1043, High, against PUMP-07, with the trend attached.
Parts reserved
02:15Automatic
Bearing 6204-2RS held from bin C7.
Assigned on the morning list
07:02S. Mehta
First job of the day, before the line started.
A planned bearing change at 07:00 instead of a seized pump at 14:00.
Ready to simplify your maintenance operations?
Book a personalized demo and see how NeoGen CMMS fits the way your team already works.
Predictive Maintenance questions
Do we need sensors on everything?
No. Start with the handful of assets whose failure actually hurts. Manual readings taken during a PM work exactly the same way as an automatic feed — the threshold does not care where the number came from.
Is this machine learning?
The thresholds described here are rules you set and can read. That is deliberate: a maintenance decision you cannot explain is one you cannot defend to an auditor or a finance director.
What if we do not know the right threshold?
Start conservative and watch. Because every reading is kept against the asset, after a few months you can see where the normal band actually sits for your plant, and tighten it.
Will it raise a job every time a reading blips?
No. A rule can require a sustained move rather than a single reading, so a one-off spike from a bad measurement does not fill your queue with noise.