Vibration Monitoring for Predictive Maintenance: Catch Failures Weeks Early

Last updated: 13 July 2026

TL;DR: Vibration monitoring tracks velocity RMS and acceleration on motors, pumps, fans and gearboxes so you spot bearing wear, imbalance and misalignment weeks before breakdown. ISO 10816 and ISO 20816 severity zones tell you exactly when to act. Predictive maintenance built on this data saves 8 to 12 per cent over preventive schedules, and far more over run-to-failure. Wireless LoRaWAN-class sensors now make it affordable on ordinary assets, not just turbines.

What is vibration monitoring and why does it matter for predictive maintenance?

Vibration monitoring is the continuous measurement of vibration on rotating machinery: motors, pumps, fans, compressors and gearboxes. Because worn bearings, imbalance, misalignment and looseness all change a machine’s vibration signature long before it fails, trending that signature lets you plan repairs weeks in advance instead of reacting to breakdowns at 3am.

Every rotating machine vibrates, and a healthy machine vibrates in a stable, repeatable way. When a fault develops, energy that should be doing work starts shaking the structure instead, and the level and frequency content of that shaking changes in predictable patterns. Vibration monitoring turns those patterns into an early-warning system.

The economics are blunt. The average large plant still loses 27 hours a month to unplanned downtime across roughly 25 separate incidents, according to the Siemens Senseye True Cost of Downtime 2024 report. If you run maintenance for a UK plant, every one of those hours has your name on it, and vibration monitoring exists to claw them back.

How does vibration monitoring catch failures weeks before breakdown?

Vibration monitoring works because most mechanical failures are gradual, not sudden. A rolling-element bearing typically deteriorates through distinct stages over weeks or months, and each stage is detectable by vibration monitoring long before you can hear, feel or smell anything wrong. Maintenance planners call the gap between first detectable symptom and functional failure the P-F interval.

In the earliest stage, microscopic subsurface fatigue shows up only as high-frequency acceleration, at levels no human inspection would ever notice. As spalling develops on raceways, distinct fault frequencies emerge and overall velocity levels begin to climb. By the final stage the bearing is audibly rough and running hot, and the machine may have days left, sometimes hours.

That sequencing is why temperature-only monitoring disappoints: heat is a late-stage symptom. By the time a bearing housing runs hot, most of your planning window is gone. Vibration gives you the longest usable warning for mechanical faults, and pairing it with temperature confirms the diagnosis as failure approaches.

The practical payoff of vibration monitoring is scheduling power. A pump that enters the alarm zone this week can be repaired at the next planned shutdown with parts ordered at normal prices, rather than stopping a line while someone couriers a bearing across the country and strips the machine under pressure.

What are the ISO 10816 and ISO 20816 vibration severity zones?

ISO 20816, which replaced the widely quoted ISO 10816 series, grades machine condition into four evaluation zones, A to D, based on broadband vibration velocity measured in mm/s RMS over 10 to 1000 Hz on non-rotating parts such as bearing housings. The zone boundaries depend on machine size and mounting stiffness.

Zone A describes the vibration of a newly commissioned machine. Zone B is acceptable for unrestricted long-term running. Zone C is not suitable for continuous long-term operation: the machine can run for a limited period while you plan remedial work. Zone D vibration is severe enough to cause damage.

For the two machine groups that cover most industrial plant on rigid mounts, ISO 20816-3 sets these velocity boundaries:

The zones give your vibration monitoring alarms defensible thresholds on day one, before you have any trend history. The full standard, ISO 20816-3:2022, covers industrial machinery above 15 kW between 120 and 30,000 rpm. Machines on flexible mounts get higher boundaries, and the best practice is to alarm on both the absolute zone boundary and a significant rise above the machine’s own baseline.

Should you measure velocity RMS or acceleration?

Measure both. Velocity RMS in mm/s is the general-purpose severity measure because it weights the 10 to 1000 Hz band where imbalance, misalignment and looseness live, and it is what the ISO 20816 zones are defined in. Acceleration, in g or m/s², emphasises higher frequencies and reveals early bearing and gear defects that velocity misses.

A useful way to remember it: acceleration tells you first, velocity tells you how bad. A developing bearing defect can multiply high-frequency acceleration while overall velocity barely moves, which is exactly why a sensor that reports both gives you the widest detection window on the same asset.

Displacement, the third quantity, matters mainly on very low-speed machinery and sleeve-bearing turbomachinery with proximity probes. For the motors, pumps and fans that make up most of a UK plant, vibration monitoring that trends velocity RMS against the ISO zones, with acceleration or an enveloped bearing-condition value alongside, covers the realistic failure modes.

What does run-to-failure vs preventive vs predictive maintenance cost?

Run-to-failure looks free because there is no monitoring spend, but it is consistently the most expensive strategy. The US Department of Energy’s FEMP best-practice research, maintained by Pacific Northwest National Laboratory, finds preventive maintenance saves 12 to 18 per cent versus reactive maintenance, and a functioning predictive programme, the kind vibration monitoring enables, saves a further 8 to 12 per cent over preventive alone. Facilities that lean heavily on reactive work can find savings of over 30 to 40 per cent.

The same research notes that 40 to 60 per cent of maintenance in typical plants is still reactive, while best-in-class operations keep it under 10 per cent. Every reactive repair carries the hidden costs: overtime, expedited freight on parts, collateral damage to couplings and seals, and production loss that dwarfs the repair invoice.

The downtime numbers put a hard edge on it. In the Siemens True Cost of Downtime 2024 study, one lost hour costs $36,000 in fast-moving consumer goods and $2.3 million in automotive, and unplanned downtime drains 11 per cent of revenues, about $1.4 trillion a year, from the world’s 500 biggest companies.

Preventive schedules are not free of failure either. Calendar-based overhauls replace healthy parts, and intrusive maintenance introduces its own defects. Deloitte’s analysis of predictive technologies estimates poor maintenance strategies cut a plant’s productive capacity by 5 to 20 per cent, while predictive approaches raise equipment uptime by 10 to 20 per cent and cut maintenance planning time by 20 to 50 per cent.

Condition-based work orders mean you intervene exactly when the machine tells you to: later than a cautious calendar would, and earlier than a breakdown would force you to.

How do wireless LoRaWAN sensors make vibration monitoring affordable?

Wired vibration monitoring systems cost so much per measurement point, in cabling, junction boxes, conduit and engineering time, that historically only turbines and other critical machines justified them. Battery-powered wireless sensors change that equation: a compact unit magnet-mounts or stud-mounts on a bearing housing, samples vibration and temperature on a schedule, and transmits over LoRaWAN or a similar long-range, low-power network.

Because a single gateway such as the Milesight UG65 can serve hundreds of sensors across a site, the marginal cost of adding one more monitored motor is close to the sensor price alone. No cable pulls, no isolations, no scaffolding. That is what finally makes continuous vibration monitoring viable on the balance-of-plant assets, the everyday motors, pumps and fans, that never justified a wired system.

In the Indiott range, the Urban.io Vibration Sensor is built exactly for this: it monitors rotating and reciprocating plant, pumps, motors, fans and gearboxes, streaming readings over Urban.io’s private long-range network into a cloud platform with REST and MQTT APIs, so you can trend baselines and alarm on remote or hard-to-reach machinery. Pair it with an EM500-PT industrial temperature sensor where a PT100 process reading adds context.

This wireless approach is now mainstream, not experimental. In the Siemens 2024 survey, nine out of ten organisations already do some form of condition monitoring and almost half run dedicated predictive maintenance teams, twice the 2019 proportion. Wireless sensing is how sites extend that coverage beyond the critical few, and it is a natural first project in a wider industrial IoT rollout for manufacturing.

How do you set up a vibration monitoring programme?

Start small, on the machines whose failure hurts most, and let results fund expansion. A first vibration monitoring project can be live within days using wireless sensors, and a disciplined seven-step sequence keeps it defensible:

1. Rank your assets. Score each machine by downtime cost, safety impact and spares lead time. Monitor the top of the list first.
2. Classify each machine. Note power rating and mounting so you can apply the right ISO 20816 group and zone boundaries.
3. Mount sensors properly. Fix them to bearing housings, drive end first, with stud or high-strength magnet mounts on clean metal.
4. Capture a baseline. Record at least two weeks of normal running across typical loads before trusting any alarm.
5. Set two-level alarms. Alert on entry to zone C or a step change above baseline; act urgently at zone D.
6. Diagnose before you strip. Confirm alarms with a spectrum or a specialist survey so you repair the actual fault.
7. Close the loop. Log every find in your CMMS and review thresholds quarterly as history accumulates.

Resist the temptation to instrument everything on day one. A focused pilot on ten difficult assets that catches one real failure will win more budget than a hundred sensors nobody has time to interpret.

Frequently asked questions about vibration monitoring

What is a good vibration level for an electric motor?

For a typical rigid-mounted motor between 15 and 300 kW, velocity below 1.4 mm/s RMS is new-machine condition and up to 2.8 mm/s is acceptable for unrestricted long-term operation under ISO 20816-3. Larger machines get more allowance. Trend direction matters as much as the absolute number: a fast rise inside zone B still deserves investigation.

What is the difference between vibration monitoring and vibration analysis?

Vibration monitoring is the continuous or scheduled trending of overall levels, velocity RMS, acceleration and temperature, to flag machines that are deteriorating. Vibration analysis is the diagnostic deep-dive that follows: examining frequency spectra and waveforms to identify the specific fault, such as imbalance, misalignment or a bearing outer-race defect, and its severity.

Can vibration monitoring predict bearing failure?

Yes. Rolling-element bearings fail through progressive stages, and high-frequency acceleration rises weeks or months before functional failure, followed by climbing velocity levels and finally heat and noise. Continuous vibration monitoring catches the early stages, giving you time to order parts and schedule the repair into planned downtime rather than suffering a breakdown.

How often should vibration be measured?

Continuously, if the asset justifies a fixed sensor. Wireless condition monitoring sensors typically sample every few minutes to hours, which catches faults that monthly handheld survey routes miss entirely, including those that develop between visits. Monthly or quarterly routes remain reasonable for genuinely non-critical machines with cheap spares and no production impact.

Do wireless vibration monitoring sensors replace wired systems?

For the vast majority of motors, pumps and fans, yes: battery-powered LoRaWAN-class sensors deliver the trend data predictive maintenance needs at a fraction of the installed cost. Ultra-critical, high-speed turbomachinery still warrants wired online protection systems with proximity probes and fast sampling. The two approaches complement each other across a site.

Where should you start?

Pick your five most painful machines and put wireless vibration monitoring on them this quarter. Browse our industrial sensors and LoRaWAN gateways, or tell us about your assets and request a quote: Indiott is a UK industrial IoT specialist and we will spec the sensors, gateway and platform for your site.