Predictive equipment maintenance: the breakdown that warns before it breaks
A device rarely breaks down all at once. It usually gives signals for days or weeks —it draws more than usual, starts with difficulty, runs hot— that no one watches because no one has a reference to compare them against. Predictive maintenance does not read the future: it turns those signals into a timely warning to call the technician before, not after, the room goes down.
The problem: the breakdown is never "sudden"
Picture the fully booked Thursday when the star device refuses to switch on. Twelve appointments cancelled, an emergency technician, and a room down for a week. Everyone experiences it as a sudden misfortune. But it almost never is: over the previous three weeks, the device had been drawing more than usual on the same sessions. The warning was there from day one. No one saw it because no one had anything to compare it to.
Reactive maintenance —fixing when something breaks— is among the most expensive. You do not just pay for the emergency repair: you pay for the lost appointments, the idle room, and the client who goes elsewhere because her session "could not happen". Getting ahead of it cuts much of that cost, and it does not require factory technology: it requires looking.
The cost of waiting for it to complain
An unplanned breakdown does not just cost the technician. It costs the day’s schedule, the idle room, and the trust of clients left without their appointment. The failure is paid for three times, and all three are avoidable if the signal is read in time.
Calendar versus real use: why the date lies
The root mistake is maintaining by calendar. "The lamp gets changed every March", "the service is due in January". But an equipment part does not expire by months: it expires by pulses and by working hours. Two identical devices, one at half load and one flat out, do not reach the same point on the same date.
Maintaining by date has two bad endings, and both cost money. If the device worked little, you swap parts at half life and throw money away. If it worked a lot, you arrive late: weak results, clients redoing "courtesy" sessions, and cascading breakdowns. The right criterion is the one aviation uses —real usage hours— not the pages of the calendar.
- By date, with light use: you replace components that still had life left. Money thrown away.
- By date, with heavy use: the part runs out before the service. Poor results and breakdown risk.
- By real use: each part is reviewed by what it has worked, against the threshold the manufacturer sets. Neither early nor late.
Consumption as an early signal
Beyond the hours on the parts, there is a second source of warning: how the device consumes. The same treatment, on the same device and with the same settings, should draw a similar amount of energy each time. When that amount starts creeping up session after session, something is changing inside the machine —a fan working harder, a tired capacitor— even if it still runs.
That pattern is the early signal. It is not a diagnosis: a different draw can come from a protocol change, the day’s temperature, or the electrical grid. But a sustained deviation, repeated across equivalent sessions, is exactly the kind of question worth taking to your technician before the device goes down.
The signal is not the diagnosis
A change in consumption does not say what is broken or guarantee it will break. It says something has moved relative to normal. Its value is not guessing the fault, but triggering the review in time; interpretation always belongs to the manufacturer or a qualified technician.
How to build a baseline
The piece missing in almost every center is a reference for normal. Without it, no "deviation" is possible: every session is a loose data point. Building that baseline is simpler than it sounds, and you only have to do it once per device and treatment.
- Build normal: with one or two weeks of measurement you have the typical consumption of each treatment on each device. That is your baseline.
- Compare only the comparable: match equivalent sessions —same treatment, same settings— and keep the context of each reading.
- Apply the manufacturer’s threshold: a sustained deviation is not an order to repair; it is a signal to consult. The thresholds and the decision come from the device documentation or a qualified technician, never a generic number.
- Act safely: if the device raises an alert or the manufacturer says to stop it, you stop. No baseline replaces a safety alarm.
From reaction to anticipation: what the center gains
Measuring all of this by hand is possible, but it does not scale. The baseline becomes truly useful when each machine’s use and consumption record themselves and are compared over time. That is where device integration turns a metering plug into an early-warning system, and analytics shows each device’s trend instead of an isolated data point.
What the center gains is not a promise of "zero breakdowns" —that does not exist— but reaction time. Seeing the deviation three weeks earlier means scheduling the review in a free slot, ordering the part calmly, and not cancelling a full schedule. The full method, with the exercises to set it up, is in the The Honest Machine resource.
Illustrative center (scenario, not measured)
Imagine a device that, before it stopped, had spent weeks drawing above its baseline on the same sessions. With the signal in view, the center would have swapped an emergency repair and an idle room for a review scheduled in a dead slot. It is a scenario to explain the mechanism, not a result measured in any center.
What predictive maintenance is not
It is worth saying plainly, so as not to sell smoke. Predictive maintenance does not diagnose breakdowns, does not avoid all of them, and does not replace the preventive maintenance the manufacturer sets. Some failures warn in advance; others do not. A baseline helps you ask good questions, not give technical answers.
Its role is humbler and more useful than the word "predictive" promises: to put a previously invisible signal in front of you, so the decision —review, wait, stop— is made by someone who knows, with the data in hand and in time. That shift, from repairing to anticipating, is what separates an idle room from a scheduled review.
What if your center stopped having invisible leaks?
We show you Qleven working on your center's real operations. No commitment, in 15 minutes.
See Qleven in your center · 15-min demoFrequently asked questions
Does predictive maintenance guarantee my equipment will not break down?
Why is maintaining by usage hours better than by date?
Does a rise in consumption mean the device is broken?
Do I need new or special equipment for this?
Keep digging deeper
The Qleven features that solve what you just read about.
