Before Buying Sensors, Know What Failure You’re Trying To Catch
Too many facilities decide to implement condition monitoring, evaluate sensor platforms, and select based on cost, brand, or capability pitch without first defining what failure modes they’re trying to detect on which specific assets. The result is sensors generating data nobody acts on, coverage gaps on assets that matter most, and a program that underdelivers relative to its cost.
The questions that should come first: what failure, on what machine, and with what detection window?
Vibration sensors
Vibration monitoring is the foundation of rotating equipment condition monitoring and the most widely deployed sensor type for a reason. It reliably catches bearing defect frequencies (BPFO, BPFI, BSF, FTF) weeks to months before failure, along with imbalance, misalignment, looseness, and gear mesh anomalies, all with clear frequency signatures that point to specific fault types.
Where it falls short: Equipment running below roughly 10 RPM, electrical motor faults, and assets where secure sensor mounting is impractical.
Temperature sensors and thermal imaging
Low-cost, broadly applicable, and easy to integrate, thermal imaging is the most reliable non-contact method for catching loose electrical connections, overloaded phases, and failing components before they become failures. Surface temperature monitoring catches overloaded motors, blocked cooling paths, and fluid system overheating.
Where it falls short: Temperature is a late indicator on most mechanical faults. Bearings frequently progress through significant damage before surface temperatures rise detectably. Thermal monitoring complements vibration monitoring; it doesn’t substitute for it on mechanical fault detection.
Motor current signature analysis
Motor current signature analysis (MCSA) monitors the motor supply current through a clamp-on current transformer. No physical access to the machine is required. It catches broken rotor bars, air gap eccentricity, developing winding faults, and load imbalance. MCSA also detects driven-equipment faults like pump cavitation and gear damage through the motor’s current draw. This is useful on equipment where sensor mounting is otherwise impractical.
Where it falls short: MCSA spectral patterns require expertise or purpose-built software to interpret. Subtle sideband patterns from developing faults are easy to miss without a trained eye or an analytics platform that flags them automatically.
Ultrasonic sensors
Ultrasonic sensors capture high-frequency acoustic energy that standard vibration analysis misses in early fault stages. They detect early-stage bearing distress, compressed air and gas leaks, partial discharge in electrical switchgear, and steam trap condition. On bearing fault detection specifically, ultrasonic monitoring typically surfaces developing faults 4 to 8 weeks before vibration analysis detects the same issue.
Where it falls short: Ultrasonic sensors require careful baseline establishment and are sensitive to background noise in high-ambient environments. They are the best early-stage bearing detection tool available but are not a general-purpose replacement for vibration analysis.
More sensors don’t improve reliability. The right sensors do.
Sensor selection should follow failure mode analysis, not precede it:
- Define which failures would cause the most downtime, safety risk, or repair cost.
- Confirm the sensor’s detection window is long enough to allow planned intervention.
- Verify physical access and mounting requirements to ensure sensor integrity.
- Consider whether the data requires specialist interpretation or can be interpreted in-house.
- Start with the highest-criticality assets and one sensor type before expanding.
Matching the sensor type to a specific failure mode on a specific asset with a detection window long enough to act on is what produces results. Starting with the failure and working backward to the sensor is the only sequence that makes sense.
