What predictive maintenance can and cannot do
Predictive maintenance is often described as knowing when a machine will fail. In practice, no sensor can give a reliable failure date for a compressor. What it can do is show that something has changed, how fast it is changing and whether there is enough time to act before the change becomes a breakdown.
That time between the first credible warning and functional failure is the lead time. Condition-based maintenance (CBM) is worthwhile when the lead time is long enough to plan parts, labour and a shutdown. If a failure mode gives no warning, or the warning arrives minutes before the trip, monitoring alone will not help and another strategy, such as redundancy or scheduled replacement, is needed.
CBM does not replace the manufacturer’s service schedule. It sits alongside preventive maintenance and refines it, showing where intervals can safely stay as they are and where conditions on site are wearing parts faster.
From failure mechanism to maintenance decision
Every useful monitoring point answers a chain of questions:
- Failure mechanism: what physically degrades, for example rolling-element bearing fatigue in the airend.
- Indicator: what measurable signal changes, for example vibration at bearing defect frequencies or rising bearing temperature.
- Healthy baseline: the reading when the machine was known to be in good condition, at a recorded load and ambient temperature.
- Method and frequency: how and how often to measure, set so that the trend is caught well inside the lead time.
- Warning and action limits: when to increase monitoring and when to plan intervention.
- Response: the inspection or repair that follows, with parts and shutdown planned.
- Verification: as-found and as-left readings that confirm the fix and reset the baseline.
An alert is only useful if it can be verified, acted on within the available lead time and closed with evidence.
Which indicators matter on a compressed-air system?
| Component | Failure mechanism | Condition indicator | Typical method |
|---|---|---|---|
| Airend bearings | Fatigue, lubrication breakdown, contamination | Vibration at bearing frequencies, bearing temperature, wear metals in oil | Vibration route and spectrum, oil analysis |
| Airend rotors | Wear, increasing internal clearances | Falling delivered air at same speed, rising loaded time, higher discharge temperature | Performance trend, loaded-hour ratio |
| Oil coolers and aftercoolers | External dust fouling, internal varnish or scale | Rising discharge temperature, larger approach temperature | Temperature trending against ambient |
| Oil separator | Element loading or rupture | Separator differential pressure, oil carry-over, oil consumption | Pressure-drop trend, oil top-up record |
| Lubricant | Oxidation, varnish, water ingress | Viscosity, acidity, water content, colour | Oil sampling and laboratory analysis |
| Motor | Bearing wear, insulation ageing, overheating | Vibration, winding temperature, insulation resistance, current imbalance | Vibration, thermography, electrical tests |
| Belt and coupling drive | Wear, misalignment, slip | Vibration at running and belt frequencies, temperature, visual condition | Vibration, inspection |
| VSD | Capacitor ageing, cooling fan or heatsink fouling | Drive temperature, fault history, DC bus ripple | Controller data, thermography |
| Dryer | Condenser fouling, refrigerant loss, desiccant degradation | Pressure dew point trend, inlet temperature | Dew-point sensor, temperature logging |
| Filters and drains | Element loading, drain failure | Differential pressure, drain cycling | Gauges, operator checks |
Vibration monitoring
Vibration is the most informative indicator for rotating machinery, but only when measured consistently. Readings should be taken at fixed, marked points on bearing housings, in the same directions (horizontal, vertical and axial), with the same sensor mounting each time. A magnet on a clean, flat spot gives repeatable data; a hand-held probe on a painted cover does not.
- Overall level: a simple trend that shows when condition is changing.
- Frequency spectrum: separates running-speed effects (imbalance, misalignment, looseness) from bearing defect frequencies, rotor lobe-pass frequencies, gear mesh and belt frequencies.
- Load and speed context: a screw compressor vibrates differently loaded and unloaded, and a VSD machine differently at each speed. Compare readings in the same state.
- Trend rate: a slow, steady rise allows planning; a sudden step change calls for prompt specialist review.
Severity limits are machine-specific. General vibration standards give guidance, but a reliable action limit comes from the machine’s own baseline, its manufacturer’s guidance and specialist interpretation. After any repair, take a new baseline spectrum.
Temperature and thermal performance
Temperature is the simplest predictive indicator to collect, because almost every controller records it. On an oil-injected screw compressor, normal discharge temperature is typically about 75–95 °C. The important signal is not one reading but the trend relative to ambient.
- Record ambient, compressor inlet and discharge temperatures together.
- Track the cooler approach: the difference between cooling-air inlet temperature and the cooled oil or air outlet. A widening approach at similar load means fouling.
- In Indian summers, with ambient temperatures of 40–45 °C, and above 45 °C on open sites, and dusty air, coolers foul faster and margins to high-temperature shutdown shrink. Thermal trends warn of this weeks before trips begin.
- Thermography of electrical panels, motor terminal boxes and bearings reveals hot connections and failing components before they cause a trip.
Oil analysis and lubricant condition
Oil analysis looks inside the machine without opening it. A typical programme checks viscosity, oxidation or acidity, water content, particle count and wear metals such as iron and copper. A rising iron trend may point to bearing or rotor wear; water may point to condensation from running too cool or high humidity; rapid oxidation points to high temperature.
Oil analysis is not a standalone diagnosis. Read the result with the machine’s temperature history, oil type, top-up record and hours since change. Take samples from the same point, while the machine is warm and running, using clean bottles, and send them to the same laboratory so results are comparable.
Pressure, performance and energy indicators
Many degradation mechanisms show up first as lost performance. If the plant’s demand is unchanged but a compressor spends more time loaded, it may be delivering less air because of internal wear, an inlet valve fault, a slipping belt or a blocked intake filter. Rising specific power, in kW per m³/min, reveals the same thing in energy terms.
- Intake filter differential: rising restriction reduces delivered air.
- Separator differential: a rising trend increases energy; a sudden fall may indicate a ruptured element.
- Pressure band behaviour: frequent cycling or failure to reach unload pressure suggests capacity loss or leaks.
- Dew point: a slow rise in refrigerated dryer dew point above its typical +3 °C warns of condenser fouling or refrigerant issues.
Setting thresholds and lead times
Thresholds should be set in two stages. A warning limit means “look more closely”: increase measurement frequency, inspect or seek specialist review. An action limit means “plan the intervention now”. Both should be based on the machine’s baseline, manufacturer guidance and experience with that failure mode, and reviewed once real data accumulates.
Measurement frequency must be shorter than the lead time. A bearing defect that typically develops over weeks can be caught with monthly vibration readings on a non-critical machine, but a critical base-load compressor may justify weekly readings or continuous monitoring. Record the reason for each frequency so it can be reviewed.
Data quality and records
Poor data produces false alarms and missed failures. Each reading should carry the asset ID, measurement point, method, units, operating state, load, ambient temperature and who took it. Sensors drift: pressure transducers, temperature probes and dew-point sensors need periodic checks. When a sensor is replaced or moved, mark the change so a step in the trend is not mistaken for a fault.
Keep the history. A trend over several years, through repairs and overhauls, is the most valuable predictive tool a maintenance team can have, and it supports later decisions on airend overhaul or replacement.
When to call a specialist
CBM tells you something is changing. Diagnosing exactly what, and deciding whether the machine can safely run to the next planned stop, often needs specialist analysis: detailed vibration spectra, oil laboratory interpretation, motor electrical testing or an airend inspection. The cost of a specialist review is small compared with an unplanned airend failure.
How Reddy Compressors helps
Reddy Compressors, at Tarnaka in Secunderabad, has served industry since 2002 as an independent multi-brand compressor specialist, not an authorised dealer. We have tie-ups with all major and reputed brands. We service ELGi, Atlas Copco, Ingersoll Rand, KAESER, Chicago Pneumatic, Sullair, Doosan and many other brands, with over 20 service engineers and technicians covering all of Telangana and Andhra Pradesh.
We can help define what to monitor on each machine, take baseline readings, set up operator rounds and controller data reviews, arrange oil sampling, trend temperature and pressure data, investigate warnings, plan and carry out repairs and overhauls, and supply spares identified from the nameplate. Where a critical machine must be taken out of service, we can provide a rental compressor from our own fleet.
What to send us
- A list of compressors, dryers and VSDs with make, model, serial and nameplate photos.
- Running hours, loaded hours and recent controller alarm or trip history.
- Any existing vibration, oil analysis or thermography reports.
- Which machines are critical and whether standby capacity exists.
- Current symptoms: rising temperature, noise, oil consumption, water in air or loss of pressure.
- Compressor-room conditions: ventilation, ambient temperature and dust.
Frequently asked questions
What is predictive maintenance for air compressors?
Predictive maintenance uses measured trends, such as vibration, temperature, oil condition, pressure drop and power, to detect degradation early and plan repairs before a breakdown. It does not predict an exact failure date. Its value comes from giving enough warning to arrange parts, labour and a shutdown, so that repairs happen on your schedule rather than the machine’s.
What is the difference between preventive and predictive maintenance?
Preventive maintenance is done at fixed running hours or calendar intervals, such as oil, filter and separator changes. Predictive or condition-based maintenance acts when a measured indicator shows that a component is degrading. Most compressor programmes use both: manufacturer preventive tasks as the foundation, and condition monitoring to catch faults and refine intervals for the site.
How is vibration measured on a screw compressor?
Vibration is measured with an accelerometer at fixed, marked points on the airend and motor bearing housings, in horizontal, vertical and axial directions. Readings should be taken in the same load state each time. Overall levels are trended, and a frequency spectrum is analysed when levels change, to identify bearing defects, misalignment, looseness or rotor problems.
What does oil analysis show in a compressor?
Oil analysis shows lubricant health, including viscosity, oxidation, acidity and water content, and machine condition through wear metals and particle counts. Rising iron or copper can indicate bearing or component wear, water suggests condensation, and fast oxidation points to high temperature. Trends from regular samples are far more useful than a single result.
What temperature trend indicates a cooler problem?
Watch the difference between cooling-air or cooling-water inlet temperature and the cooled oil or air outlet. If that approach widens at similar load and ambient, the cooler is fouling. On an oil-injected screw, discharge temperature is typically about 75–95 °C; a steady rise towards the shutdown limit is a clear warning to clean or inspect.
Is condition monitoring worth it for small compressors?
For small, non-critical compressors, simple monitoring is usually enough: operator rounds recording pressure, temperature, hours and oil level, plus regular filter checks. Advanced methods such as continuous vibration monitoring are better suited to large or critical machines where an unplanned stop is costly. Match the monitoring effort to the consequence of failure.
Related pages
Technical content reviewed by D. Srinivas Reddy, Managing Partner and service engineer, Reddy Compressors. Last updated 17 October 2026. How our engineering content is reviewed.