Predictive maintenance sensors help industrial teams detect developing faults in motors, pumps, fans, compressors, gearboxes, and conveyors before those faults become unplanned downtime. The most practical starting point is usually vibration and temperature monitoring on critical rotating equipment, supported by clear alarms, reliable data collection, and a maintenance process that turns alerts into completed work orders.
This article explains how to select, install, commission, and operate predictive maintenance sensors in a commercial industrial environment. It is written for maintenance managers, reliability engineers, plant engineers, controls technicians, and procurement teams evaluating sensors for factories, processing plants, utilities, warehouses, and heavy industry.
What Predictive Maintenance Sensors Do
A predictive maintenance sensor measures a physical condition that changes as equipment health changes. Instead of waiting for a bearing to seize, a motor to trip, or a pump to lose performance, the maintenance team receives an early warning based on measurable evidence.
Common sensor inputs include:
Vibration
Surface or process temperature
Motor current and power
Speed or RPM
Pressure
Flow
Acoustic or ultrasonic emissions
Oil condition
Humidity
Position or displacement
Magnetic flux
Wireless vibration sensors commonly measure multi-axis vibration and temperature, then transmit data to a gateway or software platform for trending, alarms, and diagnostics. The purpose is not simply to collect data; it is to produce timely, trustworthy maintenance decisions.bakerhughes+1
Predictive maintenance differs from preventive maintenance because it uses actual asset condition rather than a fixed calendar. Condition-based maintenance triggers work when measured data indicates a developing problem, which can reduce unnecessary intervention while still protecting critical assets.
Why Sensors Matter for Maintenance Reliability
Unplanned equipment failure is expensive because it often arrives at the worst possible time: during production, outside normal working hours, or when spare parts and technicians are unavailable.
Sensors help maintenance teams move from “respond after failure” to “plan before failure.” A vibration sensor may reveal a bearing defect weeks before failure. A temperature sensor may identify an overloaded motor or failing coupling. A current sensor may show a pump operating away from its best efficiency point.
The business case is strongest when sensors are applied to assets where failure causes:
Production loss
Safety risk
Environmental release
Quality defects
High repair cost
Long lead-time spare parts
Regulatory exposure
Repeated manual inspections
Wireless condition monitoring is especially useful for inaccessible, hazardous, or remote equipment because it reduces the need for technicians to physically approach running machinery.
Types of Predictive Maintenance Sensors
Vibration Sensors
Vibration sensors are the foundation of most rotating-equipment monitoring programmes. They detect mechanical motion and can reveal imbalance, misalignment, looseness, bearing defects, gear problems, resonance, cavitation, and lubrication issues.
Typical vibration measurements include:
Overall RMS velocity
RMS acceleration
Peak acceleration
Crest factor
Envelope spectrum
FFT spectrum
Time waveform
Bearing defect indicators
Many industrial wireless sensors measure vibration in three axes and include surface temperature, battery status, and sensor-health information. These measurements are useful because different fault types create different vibration signatures.
For example:
Vibration sensors are particularly valuable on motors, pumps, fans, blowers, compressors, gearboxes, conveyors, agitators, and centrifuges.
Temperature Sensors
Temperature sensors detect overheating in bearings, motors, electrical connections, hydraulic systems, gearboxes, and process equipment.
There are two broad approaches:
Contact measurement using RTDs or thermocouples
Non-contact measurement using infrared sensors
RTDs and thermocouples are commonly used for direct temperature measurement, while infrared sensors are used when direct contact is impractical.
Temperature is most powerful when combined with another measurement. A rising bearing temperature alone may indicate lubrication problems, load changes, ambient conditions, or sensor placement issues. Combined with vibration, it becomes much more diagnostic.
Current and Power Sensors
Motor-current sensors monitor electrical load, current imbalance, start-up behaviour, and operating changes. They are useful for pumps, fans, conveyors, crushers, compressors, and other motor-driven equipment.
Current monitoring can identify:
Overload
Underload
Rotor-bar problems
Voltage imbalance
Mechanical binding
Cavitation or process restriction
Belt slipping
Coupling problems
Current sensors are especially valuable when vibration sensors cannot easily be mounted close to the machine or when electrical data is already available from a drive or motor-control centre.
Pressure and Flow Sensors
Pressure and flow sensors support predictive maintenance on hydraulic systems, compressed-air systems, pumps, filters, and process equipment.
Useful indicators include:
Filter differential pressure
Pump suction and discharge pressure
Hydraulic system pressure
Compressed-air pressure drop
Flow deviation
Pressure pulsation
Leak-related pressure loss
A rising filter differential pressure, for example, can indicate clogging before the system loses performance or trips on a high-pressure alarm.
Ultrasonic and Acoustic Sensors
Ultrasonic sensors detect high-frequency sound associated with compressed-air leaks, steam leaks, electrical arcing, valve leakage, and early bearing distress.
They are particularly useful for:
Compressed-air leak surveys
Steam-trap testing
Electrical inspection
Valve and trap diagnostics
Hard-to-access machinery
Early-stage bearing defects
Ultrasonic monitoring complements vibration analysis because it can detect phenomena that conventional vibration sensors may miss, especially in low-speed equipment or electrical systems.
Oil-Condition Sensors
Oil sensors measure contamination, moisture, particle count, viscosity changes, dielectric constant, or wear debris in lubrication and hydraulic systems.
They are valuable for:
Gearboxes
Hydraulic power units
Large bearings
Turbines
Compressors
Circulating-oil systems
Oil analysis can identify water ingress, particle contamination, additive depletion, and abnormal wear metals. Inline sensors provide continuous visibility, while laboratory oil analysis provides deeper diagnostic detail.
Displacement and Position Sensors
Displacement sensors measure shaft movement, clearance, expansion, or position. They are common on large turbines, compressors, hydroelectric equipment, and high-value rotating machinery.
Typical measurements include:
Shaft radial displacement
Axial displacement
Thrust position
Expansion
Valve position
Actuator position
These sensors are usually more specialised and expensive than general-purpose vibration sensors, so they are normally reserved for critical or high-energy assets.
Wired Versus Wireless Sensors
Choosing between wired and wireless sensors is one of the most important early decisions.
Wireless systems are attractive because they remove cabling, speed deployment, and allow monitoring of hard-to-reach equipment. They typically use battery-powered sensors that wake at set intervals, capture vibration data, and transmit it to a gateway or cloud platform.
Wired systems remain preferable for machines where high-frequency data, continuous waveform capture, fast shutdown response, or integration with protection systems is required.
How to Select the Right Sensors
Do not begin by choosing a sensor brand. Begin by defining the asset, failure mode, measurement objective, and required action.
Define the Asset Criticality
Classify assets by consequence of failure. A critical asset may justify permanent monitoring, while a lower-risk asset may only need periodic route-based measurement.
Ask:
Does failure stop production?
Is there a safety or environmental consequence?
Are spare parts readily available?
How long does repair take?
Has this asset failed before?
Is the failure mode detectable by sensors?
What is the cost of a false alarm?
Match the Sensor to the Failure Mode
A sensor must be capable of detecting the failure mode you want to prevent.
Examples:
Consider the Operating Environment
Industrial sensors must survive the environment where they are installed.
Check:
Temperature range
Moisture and washdown exposure
Dust and corrosive chemicals
Vibration levels
Electromagnetic interference
Hazardous-area certification
IP ingress protection
Impact and abrasion exposure
UV exposure
Cleaning chemicals
A sensor suitable for a clean indoor motor room may fail quickly in a food-processing washdown area, outdoor pumping station, or mining environment.
Evaluate Battery Life and Connectivity
For wireless sensors, battery life is not just a specification. It depends on measurement frequency, transmission interval, temperature, radio conditions, and alarm behaviour.
Ask vendors:
What is the expected battery life at the intended sampling rate?
Can sampling be adjusted by asset criticality?
Does the sensor support event-triggered measurement?
How is low-battery status reported?
Can batteries be replaced in the field?
What happens when the gateway loses connection?
Is local data buffering available?
Some wireless sensors support different operating modes, including scheduled wake-up intervals, continuous monitoring, and event-triggered measurements.
Review Data Output and Diagnostics
A good system should produce decisions, not only raw data.
Look for:
Baseline creation
Trend charts
Alarm thresholds
Fault classification
Spectral analysis
Envelope analysis
Temperature correlation
Work-order integration
Mobile alerts
Reporting
API or Modbus integration
Export options
Wireless sensors can transmit overall vibration, temperature, and diagnostic data to central software, where anomalies can be detected before failure. However, the platform must present information in a way that maintenance planners and technicians can act on.
Where to Install Sensors
Sensor placement is often more important than sensor brand. A high-quality sensor installed in the wrong location can miss the fault entirely.
Vibration Sensor Placement
For rotating equipment, mount vibration sensors on a rigid part of the machine structure as close as practical to the bearing or component being monitored. Mounting on a flexible cover, thin guard, or loose bracket can distort the signal.
General guidance:
Measure horizontally, vertically, and axially where possible
Place sensors close to the bearing housing
Avoid thin panels, covers, and unsupported structures
Use consistent mounting orientation
Use a repeatable mounting method
Avoid locations affected by structural vibration from other equipment
Document exact sensor coordinates
For a motor-driven pump, useful measurement points often include:
Motor drive-end bearing
Motor non-drive-end bearing
Pump bearing housing
Pump casing near the volute
Baseplate or skid for structural checks
Triaxial sensors can simplify installation by measuring horizontal, vertical, and axial vibration in one unit.
Temperature Sensor Placement
For bearings and gearboxes, place temperature sensors close to the load zone or bearing outer race where feasible. For motors, monitor winding temperature where available and bearing or housing temperature where appropriate.
Avoid placing temperature sensors where they will be affected by:
Direct sunlight
Process heat
Cooling fans
Washdown water
Airflow changes
Nearby hot piping
Current Sensor Placement
Current sensors are normally installed at the motor control centre, variable-frequency drive, starter, or cable entry point. Ensure the sensor is rated for the conductor size, current range, and installation category.
Pressure Sensor Placement
Install pressure sensors where the reading represents the condition you need to monitor. For filter monitoring, measure differential pressure across the filter rather than relying only on upstream pressure.
Baseline Data and Alarm Limits
A new sensor does not immediately know what “normal” means for your machine. Baseline data establishes the equipment’s healthy operating signature.
Create a Baseline
Collect data under defined operating conditions:
Normal production load
Stable speed
Known product or process condition
Similar ambient temperature
No recent maintenance disturbance
No abnormal process condition
Record:
Vibration levels by axis
Temperature
Motor current
Speed
Pressure or flow
Machine operating state
Date and time
Technician notes
A baseline should not be a single reading. Collect multiple readings across normal operating cycles before setting alarm limits.
Use Absolute and Relative Alarms
Absolute alarms use general standards or vendor guidance. Relative alarms compare current readings with the asset’s own historical baseline.
A practical approach uses both:
Alert level: investigate at the next planned opportunity
Alarm level: schedule diagnosis and corrective work
Critical level: consider operational action or shutdown
Rate-of-change alarm: investigate sudden changes even if the absolute value remains acceptable
Published guidance for general machinery often uses RMS velocity thresholds, but these must be adapted to machine type, speed, mounting, and application. One commonly cited threshold framework places general machinery alarm levels above 4.5 mm/s RMS and danger levels above 11.2 mm/s RMS, but site-specific baselines remain essential.
Avoid Alarm Fatigue
Too many alerts destroy confidence in the system. Every alarm should have an owner, a response procedure, and a defined priority.
Good alarm design includes:
Different severity levels
Alarm suppression during startup or shutdown
State-based thresholds for variable-speed equipment
Confirmation requirements for critical alarms
Clear instructions for first response
Feedback loop to improve thresholds
Monthly alarm review
Integrating Sensors With CMMS and Maintenance Workflow
Sensors create value only when alerts become planned work.
A practical workflow is:
Sensor detects abnormal trend or threshold breach.
System sends alert to reliability engineer or planner.
Engineer validates the signal and reviews history.
Work order is created with priority, scope, and required parts.
Technician performs inspection or corrective task.
Results and findings are recorded in the CMMS.
Sensor baseline and alarm settings are updated if needed.
Integration with a CMMS allows maintenance teams to connect condition data to work orders, asset history, spare parts, labour, and downtime records. This makes it easier to prove the value of the monitoring programme.
Avoid creating alerts that only appear in a dashboard nobody reviews. Assign clear responsibilities:
Reliability engineer: diagnoses trends and validates alerts
Maintenance planner: converts confirmed issues into work orders
Technician: performs inspection and repair
Operations: provides operating context and shutdown access
Storeroom: ensures critical parts are available
Management: reviews KPIs and investment decisions
Implementation Roadmap
Phase 1: Identify Pilot Assets
Select three to ten assets that are important but not so critical that a mistake would halt the entire plant.
Good pilot candidates include:
A critical process pump
A large ventilation fan
A conveyor drive
A compressor
A gearbox with a failure history
A motor in a difficult-access location
Choose assets with known failure modes and measurable production impact.
Phase 2: Define Success Criteria
Before installation, define what success means.
Examples:
Detect at least one developing fault before failure
Reduce manual inspection time
Improve response time to abnormal conditions
Increase planned maintenance percentage
Reduce unplanned downtime
Create reliable baseline data
Generate actionable work orders
Phase 3: Install and Commission
Install sensors according to manufacturer instructions and site safety procedures.
Commissioning should include:
Verification of sensor mounting
Communication and signal-strength testing
Battery and sensor-health checks
Correct asset mapping
Baseline data collection
Alarm threshold configuration
User training
Alert-response testing
Documentation
Phase 4: Monitor and Refine
Review data weekly during the first months. Look for false alarms, missed events, poor signal quality, and unclear ownership.
Refine:
Sensor locations
Sampling intervals
Alarm thresholds
Alert routing
Work-order templates
Reporting format
Spare-parts strategy
Phase 5: Scale With Discipline
Expand only after the pilot produces reliable, actionable results. Scale by asset class, criticality, and failure history—not by installing sensors everywhere at once.
Common Sensor Installation Mistakes
Mounting on Flexible Surfaces
A sensor mounted on a guard, cover, or thin sheet-metal panel may measure panel vibration rather than bearing vibration. Use rigid mounting points near the component being monitored.
Inconsistent Sensor Orientation
If one technician mounts a sensor vertically and another mounts it horizontally, trends become unreliable. Standardise orientation and document it.
Ignoring Operating State
A machine at low load cannot be compared directly with the same machine at full load. Use operating-state tags or separate baselines for each normal operating condition.
Setting Alarms Too Early
Default thresholds may generate false alarms. Use a baseline period and adjust thresholds based on actual machine behaviour.
Failing to Close the Loop
An alert without a work order is not predictive maintenance. Ensure every valid alert has a defined owner and corrective action.
Poor Cybersecurity Practices
Wireless sensors, gateways, and cloud platforms must be treated as part of the plant’s digital infrastructure. Use network segmentation, strong authentication, controlled remote access, firmware updates, and vendor security review.
No Training
Technicians need to understand what the data means, what actions to take, and when to escalate. Reliability engineers need diagnostic training. Planners need to understand how to convert alerts into effective work orders.
Cost and Return on Investment
The cost of a predictive maintenance programme includes more than sensor hardware.
Budget for:
Sensors
Gateways or receivers
Software licences
Installation labour
Conduit, mounts, and adhesives
Batteries
Calibration or verification
Training
IT and cybersecurity work
System integration
Ongoing support
Data management
The return comes from avoided failures, reduced emergency labour, better spare-parts planning, lower energy waste, improved safety, and longer asset life.
A simple way to estimate value is:
Then compare annual benefit with total annual cost of ownership.
For example, if a monitored pump prevents one major failure per year that would have caused eight hours of production loss, the avoided production loss alone may justify the monitoring investment.
Key Performance Indicators
Track both technical and business outcomes.
Technical KPIs
Sensor uptime
Data completeness
Battery health
Alert accuracy
False-alarm rate
Number of faults detected early
Time from alert to work order
Time from work order to repair
Baseline coverage
Percentage of critical assets monitored
Business KPIs
Unplanned downtime hours
Maintenance cost per asset
Emergency work-order percentage
Planned-maintenance percentage
Mean time between failures
Mean time to repair
Spare-parts stock accuracy
Maintenance backlog
Production availability
Safety incidents related to equipment failure
Sensor Selection Checklist
Before purchasing, confirm:
The specific failure modes to detect
Required measurement range and accuracy
Mounting method and location
Environmental rating
Hazardous-area requirements
Battery life at the intended sampling rate
Wireless range and gateway coverage
Data ownership and export options
Cybersecurity requirements
Integration with CMMS, SCADA, or ERP
Alarm configuration flexibility
Diagnostic capability
Vendor training and support
Total cost of ownership
Pilot plan and success criteria
Final Recommendations
Start small, focus on critical rotating equipment, and build a process that converts sensor data into completed maintenance work. Vibration and temperature sensors are usually the best first investment because they address many common mechanical failure modes on motors, pumps, fans, and compressors.
Use wireless sensors to scale monitoring across larger fleets and difficult-to-access assets, but use wired or higher-bandwidth systems where continuous high-resolution diagnostics or fast protective action is required.
Most importantly, treat predictive maintenance as a reliability programme, not a technology purchase. The sensor detects the problem; the maintenance system, trained people, spare parts, and planning process solve it.
Comments
Post a Comment