Machine Condition Monitoring: Definition

What Is Machine Condition Monitoring?

Every machine in operation generates information about its own health through the physical signals it produces: the vibration signature of a bearing, the temperature of a motor winding, the current drawn by a compressor, the particle content of a gearbox oil sample. Most of this information is invisible to the human senses under normal operating conditions and requires instruments and trained interpretation to extract. Machine condition monitoring is the discipline of systematically collecting and analyzing these signals to detect developing faults while they are still in early stages, when they are inexpensive to correct and before they produce the secondary damage, extended downtime, and emergency procurement costs that characterize unplanned failures.

The core insight underlying condition monitoring is that most mechanical failure modes are not instantaneous events. They develop over time through recognizable stages: a bearing does not go from perfectly healthy to completely failed without passing through weeks or months of measurable surface fatigue, progressive spalling, and increasing roughness that produce detectable vibration signatures. A motor winding does not fail without first degrading through insulation deterioration that generates measurable temperature rise and current anomalies. The window between when a developing fault first becomes detectable and when it becomes a functional failure is called the P-F interval, and condition monitoring's purpose is to detect faults within that window and allow maintenance to respond before the functional failure occurs.

Machine condition monitoring is related to the broader concept of condition monitoring applied to any asset type, but specifically focuses on rotating and reciprocating machinery: motors, pumps, fans, compressors, gearboxes, turbines, and similar equipment where vibration, temperature, and oil analysis are the primary diagnostic signals.

The P-F Curve: Why Condition Monitoring Works

The P-F curve is a foundational concept from reliability-centered maintenance that describes the degradation progression of a component. The curve plots the detectable condition of the component from a healthy baseline through progressive degradation to functional failure:

• Point P (Potential Failure): The point at which a developing fault first becomes detectable by a specific monitoring technology. Point P is different for each technology: a developing bearing defect may be detectable by oil analysis (elevated iron and chromium from wear particles) weeks before it is detectable by vibration analysis, and by vibration analysis weeks before it is detectable by the human ear or a temperature rise at the bearing housing.
• Point F (Functional Failure): The point at which the component can no longer perform its required function at the required standard. For a bearing, this might be when vibration exceeds a level that causes unacceptable secondary effects, when temperature rises to the point of lubricant breakdown, or when the bearing seizes.
• The P-F interval: The time available between initial detection and functional failure. This interval determines how quickly maintenance must respond once a fault is detected. A long P-F interval allows scheduled intervention at the next convenient maintenance window. A short P-F interval requires rapid response to avoid operational impact.

The practical implication: the monitoring technology that detects a fault earliest provides the longest P-F interval and the most time to plan a response. This is why multi-technology monitoring programs combine oil analysis (early), ultrasound (early to intermediate), vibration analysis (intermediate), and temperature (intermediate to late), providing detection at the earliest possible stage for each fault type.

Condition Monitoring Technologies

Vibration analysis

Vibration analysis is the most widely applied condition monitoring technique for rotating machinery. Every rotating component produces a vibration signature that reflects its condition. A rolling element bearing with a developing inner race defect generates a repeating impact at a frequency determined by the bearing geometry and shaft speed (Ball Pass Frequency Inner, BPFI). Gear tooth damage generates impacts at the gear mesh frequency. Imbalance generates vibration at 1x shaft speed. Misalignment generates elevated vibration at 1x and 2x shaft speed. Each fault type produces a characteristic signature in the vibration frequency spectrum that can be identified and tracked.

Vibration monitoring as a program tracks both overall vibration level (total energy in the signal) and the frequency spectrum over time. Alert thresholds are set at levels that indicate significant departure from baseline. When the overall level or a specific frequency band crosses a threshold, an alert is generated for investigation. Continuous monitoring detects faults in real time; periodic monitoring with portable analyzers provides spectrum data at defined intervals from the same measurement point on the machine each time, enabling valid trend comparison.

Temperature monitoring

Temperature rises ahead of most mechanical failure modes: friction from lubrication breakdown, electrical resistance increases from winding or connection degradation, overloading beyond design limits, and cooling system failures all manifest as elevated temperature before functional failure. Continuously installed temperature sensors on bearing housings, motor windings, and other critical points provide real-time detection of these conditions. Infrared thermography provides a broader view, allowing technicians to scan electrical panels, heat exchangers, and large equipment surfaces during operation to identify thermal anomalies that point sensors would miss.

Current signature analysis (CSA)

Every mechanical fault in a motor-driven machine modulates the motor's current draw in a characteristic way. A developing bearing defect in a motor produces sidebands at specific frequencies around the supply frequency in the current spectrum. A broken rotor bar produces sidebands at twice the slip frequency. An eccentric rotor produces current modulations at twice the supply frequency. Mechanical load variations from attached equipment (pump cavitation, gear tooth damage, misalignment) also appear in the current signal. CSA can be performed with non-contact current clamps without shutting down or accessing the motor, making it a convenient monitoring method for installed equipment.

Oil analysis

Oil analysis samples the lubricant from gearboxes, hydraulic systems, compressors, and other oil-wetted equipment to measure wear metal concentration, lubricant condition, and contamination levels. Wear metal analysis (spectrometric oil analysis, SOAP) detects increasing concentrations of iron, copper, aluminum, and other metals that indicate developing wear of specific internal components before the wear generates detectable vibration or temperature changes. Particle counting and morphology analysis detects the size and shape of wear particles, distinguishing normal rubbing wear from abnormal fatigue or adhesive wear modes. Oil analysis is particularly valuable for enclosed gear drives, hydraulic systems, and other equipment where internal wear would not be detectable by external vibration or temperature measurement.

Ultrasonic monitoring

Ultrasonic instruments detect high-frequency sound (typically 20 kHz to 100 kHz) that is inaudible to the human ear. Early-stage bearing defects, compressed air and steam leaks, cavitation in pumps, electrical arcing and corona discharge, and valve leakage all produce characteristic ultrasonic signatures. Ultrasonic bearing monitoring is particularly valuable for very low-speed bearings (below approximately 100 RPM) where conventional vibration analysis is less sensitive; these bearings produce ultrasonic impacts from surface contact that are detectable well before the bearing reaches a state of vibration detectable by standard accelerometers.

Condition Monitoring Technology Comparison

Continuous vs. Periodic Monitoring

The choice between continuous and periodic monitoring affects both the capital cost of the program and the minimum P-F interval it can exploit.

Continuous monitoring uses permanently installed sensors that transmit data in real time to a monitoring platform applying automatic alert thresholds and trend analysis. Any anomaly triggers an alert without requiring a technician to be present. Continuous monitoring is essential for machines with short P-F intervals (where a fault can develop rapidly to failure) and for machines where the consequences of unexpected failure are severe. The capital cost is higher than periodic monitoring, but the operational cost per machine is lower once deployed, and the detection-to-response interval is the shortest achievable.

Periodic monitoring uses portable instruments carried by trained technicians on regular routes. Measurement points must be consistent (same location, same sensor orientation, same machine operating conditions) to enable valid trend comparison between visits. The monitoring interval must be shorter than the shortest expected P-F interval for the monitored fault types, otherwise a fault could develop to failure between visits without detection. Periodic monitoring provides a practical and cost-effective solution for the broader asset population where continuous monitoring is not warranted by criticality or consequence.

Implementing a Machine Condition Monitoring Program

A practical implementation sequence for a condition monitoring program follows four phases:

1. Asset criticality assessment: Determine which machines have the highest failure consequences (production impact, safety, cost) and direct monitoring resources to these assets first. Not every machine warrants the same investment in monitoring.
2. Technology selection by fault type: For each prioritized machine, identify the most likely failure modes and select the monitoring technology that provides the earliest detection of each mode. A high-speed pump requires vibration monitoring for bearing and impeller faults; it also benefits from temperature monitoring and periodic oil analysis if it has a mechanical seal with lubricant.
3. Deployment and baseline: Install sensors or establish measurement routes. Take baseline measurements from the machine in known good condition. Set alert thresholds based on the baseline and the alarm severity standards appropriate for the machine criticality. Document measurement points, conditions, and intervals.
4. Alert response process: Define what happens when an alert is generated. The most common failure in condition monitoring programs is not the technology; it is the absence of a response process. An alert that goes uninvestigated for weeks, or is cleared without root cause analysis, does not contribute to reliability. The response process should connect monitoring alerts directly to work order generation in the CMMS.

Machine Condition Monitoring and Asset Health

Asset health monitoring integrates condition data from multiple technologies into a unified view of machine status, presenting a health score or trend for each asset that synthesizes all available monitoring signals. When any parameter triggers an alert, the health monitoring platform generates a notification linked to the asset record in the CMMS, creating a maintenance work order for investigation. This connection between monitoring and maintenance execution is what converts condition data into reliability improvement.

Applied systematically, machine condition monitoring directly improves overall equipment effectiveness: catching developing faults before they cause unplanned downtime improves the availability component of OEE; enabling planned rather than emergency repair reduces repair time and cost; and detecting performance degradation (reduced pump flow, compressor efficiency decline) allows early correction that maintains throughput.

Frequently Asked Questions

The Bottom Line

Machine condition monitoring converts the information that every operating machine generates about its own health into actionable intelligence for the maintenance team. By detecting developing faults while they are still minor, condition monitoring replaces emergency repairs with planned interventions, reduces secondary damage from run-to-failure events, and extends the life of components that would otherwise be replaced prematurely on calendar-based schedules.

The return on condition monitoring investment is realized through avoided failures, reduced emergency maintenance costs, extended component life, and improved equipment availability. The greatest barrier to realizing this return is not technology cost or complexity: modern wireless sensors and cloud monitoring platforms have made continuous monitoring accessible to any size facility. The barrier is organizational: ensuring that the alerts the monitoring system generates are consistently acted on within the available P-F interval. A monitoring program with a well-defined response process outperforms a more sophisticated program where alerts go uninvestigated.