Condition Monitoring

Condition monitoring is the practice of tracking physical parameters in machinery, such as vibration, temperature, lubrication quality, and acoustics, to identify developing faults before equipment fails. By establishing baseline operating norms and comparing continuous or periodic measurements against historical trends, maintenance teams can diagnose specific failure modes early on the degradation curve. This approach underpins condition-based and predictive maintenance strategies, enabling plants to schedule targeted repairs during planned downtime, prevent secondary component damage, and maximize overall equipment effectiveness.

Loading the machine…

Key facts

Primary techniques: Vibration analysis, thermography, oil analysis, ultrasound, and electrical signature analysis
Standard vibration metric: Root mean square (RMS) velocity in mm/s
Ultrasound testing range: 20 kHz to 100 kHz
Maintenance strategies supported: Condition-based maintenance, predictive maintenance, and TPM
Core reliability metric: Mean time between failures (MTBF)

By Matthew Savas — Founder of Kaizumi. Reviewed 29 August 2026.

Condition monitoring is the practice of tracking physical parameters in machinery, such as vibration, temperature, and lubrication quality, to identify developing faults before equipment fails. By establishing baseline operating norms and analyzing trends over time, maintenance teams can schedule targeted interventions during planned downtime rather than reacting to catastrophic breakdowns. Within industrial operations, condition monitoring forms the technical foundation of condition-based maintenance and predictive maintenance strategies. Unlike reactive maintenance, which repairs machines after failure occurs, or time-based preventive maintenance, which replaces components according to fixed schedules regardless of actual wear, condition monitoring bases maintenance activities on real-time and historical machine health data.

Core principles of condition monitoring

Condition monitoring relies on the principle that mechanical and electrical components exhibit measurable physical indicators before they suffer functional failure. When an internal component begins to degrade, such as a bearing race spalling, an impeller eroding, or an electrical winding overheating, the degradation generates physical symptoms. These symptoms include elevated kinetic energy dissipation in the form of vibration, increased thermal generation due to friction or electrical resistance, particulate contamination in lubricants, and high-frequency acoustic emissions.

The foundation of effective monitoring is the establishment of baseline operating norms. A baseline represents the measured physical behavior of an asset operating in a known healthy state under standard load, speed, and ambient conditions. Maintenance engineers record these baseline values immediately following commissioning, major overhauls, or validated component replacements.

Once baselines are recorded, periodic or continuous measurements are compared against historical trends and predetermined engineering limits. Machine degradation typically progresses along an identifiable degradation curve, moving from initial subsurface damage to detectable ultrasonic emissions, followed by measurable vibration anomalies, elevated temperature, audible sound, and finally functional failure. By detecting deviations early on this timeline, maintenance teams can identify the specific failure mode, track the rate of deterioration, procure replacement parts, and plan corrective tasks before secondary damage spreads to adjacent components.

Key parameters and monitoring techniques

Condition monitoring employs several distinct measurement techniques, each suited to specific failure modes and component types. A comprehensive program often combines multiple methods to obtain a complete assessment of asset integrity.

Vibration analysis

Vibration analysis is the most widely applied condition monitoring technique for rotating equipment, such as motors, pumps, fans, compressors, and gearboxes. Rotating machines generate vibration patterns that correlate directly with their mechanical structure and rotational frequency. Sensors, typically accelerometers or eddy-current proximity probes, convert mechanical movement into electrical signals.

Analysis of vibration involves measuring overall vibration levels, expressed as acceleration, velocity, or displacement, and evaluating the vibration frequency spectrum through fast Fourier transform algorithms. Root mean square velocity, measured in millimeters per second, is standard for evaluating general machine health in industrial machinery. Velocity indicates the destructive fatigue energy impacting the system. Spectral analysis breaks the overall signal into individual frequency components, allowing technicians to distinguish between distinct mechanical anomalies:

Imbalance generates high vibration at the fundamental rotational frequency of the shaft.
Misalignment between shafts generates characteristic peaks at two, three, and four times the rotational frequency.
Mechanical looseness generates multiple harmonics and sub-harmonics of the running speed.
Rolling element bearing faults generate non-synchronous impact frequencies determined by the geometry of the bearing races, cage, and rolling elements.
Gear defects generate distinct gear-mesh frequencies and related sidebands.

Thermography and temperature monitoring

Thermal monitoring tracks the heat generated by mechanical friction, electrical resistance, fluid flow anomalies, and thermal insulation breakdown. Surface temperature measurements can be collected continuously using permanently installed resistance temperature detectors and thermocouples, or periodically using handheld infrared pyrometers and thermal imaging cameras.

In mechanical applications, elevated temperatures indicate lubrication breakdown, excessive bearing pre-load, misalignment, or overload. In electrical systems, infrared thermography identifies high-resistance connections, unbalanced phases, overloaded circuits, and deteriorating insulation inside motor control centers, switchgear, and transformers without requiring physical contact or circuit de-energization.

Lubricant and oil analysis

Oil and fluid analysis assesses the condition of the lubricant itself and the wear state of the lubricated machinery. Lubricating oil carries debris away from contact zones, making it an information-rich medium for condition assessment. Oil analysis protocols generally evaluate three categories:

Fluid properties: Viscosity, acid number, base number, and additive depletion indicate whether the lubricant remains chemically viable.
Contamination: Ingress of water, soot, process chemicals, coolant, and particulate dirt accelerate wear and shorten oil life.
Wear debris: Spectrometric analysis detects microscopic metal particles suspended in the oil, identifying specific worn metallurgy such as iron from gears, copper from bushings, or aluminum from housings. Analytical ferrography examines larger wear particles under a microscope to determine the wear mechanism, such as abrasive cutting, sliding wear, or fatigue spalling.

Acoustic emission and ultrasonic testing

Ultrasonic monitoring detects high-frequency sound waves, typically between 20 kilohertz and 100 kilohertz, generated by friction, turbulence, and electrical discharge. Because ultrasound monitors frequencies well above the human audible range and standard mechanical background noise, it detects microscopic anomalies very early in the degradation cycle.

Airborne ultrasound locates compressed air leaks, vacuum leaks, steam trap failures, and electrical arcing or corona discharge. Structure-borne ultrasound detects early friction inside bearings before measurable heat or low-frequency vibration develops. This technique is also widely used to guide precise bearing re-lubrication, alerting technicians when friction drops to baseline during grease injection.

Motor circuit analysis and electrical signature analysis

Electrical monitoring methods assess the condition of electric motors and their associated cabling. Motor circuit analysis is a de-energized test method that evaluates resistance, inductance, capacitance, and insulation resistance to ground to identify winding faults, phase imbalances, and cable degradation. Electrical signature analysis evaluates energized motors by measuring current and voltage waveforms. Current analysis detects broken rotor bars, rotor eccentricity, power quality disturbances, and mechanical load variations reflected back into the motor supply.

Integration with lean manufacturing systems

In lean manufacturing environments, condition monitoring serves as a core technical enabler for total productive maintenance. Total productive maintenance seeks to maximize overall equipment effectiveness by eliminating breakdowns, speed losses, and quality defects.

Condition monitoring directly supports autonomous maintenance and planned maintenance pillars. Operating teams performing autonomous maintenance can use simple condition monitoring tools, such as portable infrared thermometers, vibration pens, and visual level indicators, to support standard inspection routines. When operators detect parameter drift, they trigger structured escalation protocols to the maintenance department.

The implementation of continuous parameter tracking significantly improves asset reliability metrics, specifically extending MTBF (mean time between failures). By identifying mechanical degradation while it is isolated to single expendable parts, such as a single bearing, maintenance teams prevent secondary damage to shafts, housings, and stators. This practice eliminates catastrophic failures that require long, expensive rebuilds.

Furthermore, condition monitoring eliminates recurring minor stops on high-speed production lines. Minor stops are often caused by subtle mechanical binding, thermal expansion, or intermittent sensor misalignment that do not trigger hard machine interlocks. Monitoring these conditions provides diagnostic insight into the root causes of brief line disruptions.

The data gathered through long-term condition monitoring feeds into early equipment management processes. When design and procurement teams specify new machinery, historical vibration, thermal, and reliability profiles inform engineering choices. This ensures that new assets incorporate better materials, improved access points for inspection, and factory-installed sensors that simplify ongoing health tracking.

Step-by-step implementation process

Effective condition monitoring requires a systematic process rather than merely gathering raw sensor data:

Asset criticality assessment and failure mode identification: Evaluate all manufacturing assets to establish criticality based on safety impact, production throughput constraints, repair costs, and replacement lead times. For critical assets, conduct failure mode and effects analysis to identify specific mechanical, electrical, and thermal failure modes that can be detected through physical parameters.
Sensor selection and baseline data establishment: Select appropriate sensor types, measurement locations, and sampling frequencies based on the target failure modes. Mount sensors securely at load-bearing points as close to the mechanical interface as possible. Commission the sensors and record baseline operating norms across standard operating speeds, loads, and environmental conditions.
Threshold setting and alert configuration: Establish warning and alarm levels using international technical standards (such as ISO standards for mechanical vibration), original equipment manufacturer specifications, and statistical process control limits calculated from baseline data. Configure automated notifications to route alerts to relevant personnel.
Data analysis, trend tracking, and root cause diagnosis: Continuously or periodically collect measurement data, plot trends over time, and compare current values against baselines. When a parameter exceeds an alert threshold, perform detailed diagnostic analysis, such as vibration spectrum evaluation, oil laboratory testing, or thermographic inspection, to confirm the failure mechanism and estimate remaining useful life.
Work order integration and continuous review: Link monitoring alerts directly into the computerized maintenance management system to generate work orders automatically. Schedule corrective maintenance during planned production breaks. After completing the repair, take post-maintenance baseline measurements to verify proper installation and close the diagnostic loop.

Practical application and worked example

To demonstrate the application of condition monitoring, consider a standard pumping system consisting of a continuous-duty, horizontal 75-kilowatt electric motor coupled to a centrifugal water pump.

During initial commissioning, maintenance engineers establish the healthy operating baseline for the motor drive-end bearing under full hydraulic load:

Baseline vibration velocity: 1.1 mm/s RMS
Baseline bearing housing temperature: 40°C

The asset is placed on an automated condition monitoring routine that logs overall vibration and temperature once every hour, while collecting a full high-resolution vibration spectrum once per day.

Detection and trending

Over a four-month period of continuous operation, the logged data shows a progressive upward drift in both kinetic and thermal parameters:

Month 1: 1.2 mm/s RMS and 41°C
Month 2: 1.8 mm/s RMS and 44°C
Month 3: 3.2 mm/s RMS and 52°C
Month 4: 6.4 mm/s RMS and 67°C

At the end of Month 4, the automated system triggers a warning alert when the vibration crosses the 4.5 mm/s RMS operational warning threshold and temperature exceeds 65°C.

Evidence gathering

The maintenance engineer gathers multi-parameter evidence to diagnose the exact root cause:

Vibration evidence: Spectral analysis of the 6.4 mm/s RMS signal reveals non-synchronous peaks matching the calculated ball pass frequency of the outer race of the motor drive-end bearing, accompanied by sidebands indicating modulating impacts.
Heat evidence: Thermal imaging shows localized heat concentration directly at the drive-end bearing housing (67°C), while the motor non-drive-end housing remains at 42°C and the pump casing remains at 35°C.
Oil and grease evidence: A physical sample of the bearing grease shows oxidation, discoloration, and metallic particles visible under microscopic inspection.
Sound evidence: High-frequency ultrasonic readings rise significantly above baseline levels, showing distinct repetitive peaking that corresponds to the rotational frequency.

Corrective action

Rather than waiting for the bearing to seize, which would score the motor shaft, overheat the stator windings, damage the mechanical seal, and force an unplanned production stoppage, the maintenance team takes controlled action:

The maintenance planner reviews production schedules and identifies a scheduled sanitation changeover window occurring four days later.
The planner generates a work order to replace the motor drive-end bearing and orders standard replacement parts from the plant storeroom.
During the four-hour scheduled changeover window, technicians uncouple the motor, replace the bearing, re-pack the cavity with clean grease, and align the motor-to-pump coupling using laser alignment tools.
Upon restart, technicians record a new baseline of 1.0 mm/s RMS and 39°C, verifying that the assembly has returned to a fully healthy operating condition without causing unscheduled downtime.

Limitations and operational challenges

While condition monitoring provides substantial reliability benefits, implementing and sustaining a program presents specific operational challenges:

Initial capital investment: Outfitting an entire facility with continuous online sensors, wireless transmitters, data acquisition hardware, and diagnostic software requires substantial upfront capital. Organizations often manage this by prioritizing critical assets first and using portable data collectors for lower-tier machinery.
Technical expertise requirements: Extracting actionable conclusions from complex vibration spectra, oil ferrograms, or electrical current signatures requires specialized training and certification. Without skilled analytical personnel, plants risk misinterpreting data, leading to false alarms or undetected failures.
Data management and sensor maintenance: Sensors themselves can fail, drift out of calibration, or lose physical adhesion to machine surfaces. Loose wiring, high ambient electromagnetic interference, and process temperature swings can introduce signal noise. Organizations must establish maintenance routines for the monitoring infrastructure itself to maintain data integrity.
Dynamic operating conditions: Variable-speed drives, fluctuating production loads, and intermittent operating cycles alter baseline physical parameters significantly. A machine operating at low speed naturally generates lower vibration and heat than the same machine running at full capacity. Condition monitoring systems must account for operational states to avoid false positive alarms during normal process fluctuations.

Frequently asked questions

How does condition monitoring differ from time-based preventive maintenance?: Time-based preventive maintenance replaces components according to fixed calendar or operating schedules regardless of actual wear. Condition monitoring instead tracks physical operational data to detect active mechanical degradation before failure occurs. This approach prevents unnecessary part replacements while eliminating unexpected breakdowns between scheduled maintenance intervals.
Which mechanical defects can be identified through vibration analysis?: Vibration spectral analysis isolates specific frequencies to diagnose shaft imbalance at the fundamental rotational speed and shaft misalignment across multiple running-speed harmonics. It also detects mechanical looseness, gear-mesh faults, and rolling element bearing defects characterized by non-synchronous impact frequencies. These discrete frequency patterns allow maintenance teams to pinpoint component degradation without dismantling the machine.
What machine conditions are revealed by lubricating oil analysis?: Lubricating oil analysis evaluates fluid chemical viability, external contamination, and microscopic wear debris suspended in the fluid. Tracking viscosity, acid numbers, and additive depletion determines whether the lubricant must be replaced. Analyzing the metallurgy and geometry of wear particles reveals active mechanical wear mechanisms, such as abrasive cutting, sliding wear, or component fatigue spalling.
What is the difference between airborne and structure-borne ultrasonic monitoring?: Airborne ultrasound detects high-frequency acoustic emissions traveling through the atmosphere to locate compressed air leaks, vacuum leaks, steam trap failures, and electrical discharge. Structure-borne ultrasound monitors high-frequency stress waves traveling through solid machine housings, identifying early friction inside bearings before measurable heat or low-frequency vibration appears. Structure-borne readings are also used during maintenance to ensure bearing friction returns to baseline levels during lubrication.
How do motor circuit analysis and electrical signature analysis differ?: Motor circuit analysis is performed on de-energized motors to evaluate resistance, inductance, capacitance, and ground insulation for winding and cabling defects. Electrical signature analysis evaluates energized motors by capturing current and voltage waveforms under operational load. This energized monitoring identifies broken rotor bars, rotor eccentricity, power supply disturbances, and mechanical load anomalies reflected back through the motor.

Terms

MTBF OEE Total Productive Maintenance