A predictive maintenance (PdM) programme uses continuous or periodic condition monitoring to detect equipment degradation before failure occurs, enabling repairs to be scheduled at the lowest possible cost and disruption. Unlike preventive maintenance — which replaces parts on a fixed calendar regardless of actual condition — PdM acts on real equipment data, so you intervene only when necessary. For Singapore facilities operating under MOM workplace safety obligations and competitive cost pressures, a well-structured PdM programme is one of the highest-return investments a maintenance team can make.
This guide takes you through every stage: asset prioritisation, technology selection, data collection protocols, team training, and how calibrated instruments underpin the entire process. Whether you manage a pharmaceutical cold chain, a heavy manufacturing line, or a commercial building's HVAC plant, the principles are the same — measure accurately, trend consistently, and act before the machine tells you it has failed.
Why Predictive Maintenance Outperforms Reactive and Preventive Approaches
Reactive maintenance (fix it when it breaks) carries the highest total cost: unplanned downtime, emergency labour premiums, collateral damage to connected equipment, and potential safety incidents reportable to MOM under the Workplace Safety and Health Act. Studies by the US Department of Energy and the European Maintenance Society consistently show reactive maintenance costs two to five times more per repair event than the equivalent planned intervention.
Preventive maintenance (PM) improves on this by scheduling work ahead of failure, but it introduces a different inefficiency: parts are replaced based on time intervals derived from average failure statistics, not from the actual condition of your specific machine. In a humid, tropical environment like Singapore, where ambient temperature and humidity accelerate corrosion and bearing wear differently from temperate climates, generic OEM intervals may be too conservative in some cases and too liberal in others.
Predictive maintenance resolves both problems. By measuring the condition of each asset — vibration signature, temperature, insulation resistance, oil viscosity, ultrasonic emission — you know the actual health of that machine at that moment. Interventions happen when the data says they should, not according to a calendar.
Step 1 — Asset Criticality Assessment
Not every piece of equipment justifies the investment of continuous monitoring. The first step is to rank your assets by criticality using a structured framework. A simple criticality matrix scores each asset on two axes:
- Consequence of failure: safety risk, production loss, regulatory compliance breach (e.g., NEA environmental permit violations, HSA GDP requirements for pharmaceutical storage), cost of repair, and lead time for spare parts.
- Probability of failure: age, duty cycle, operating environment, maintenance history, and OEM failure-rate data.
Assets scoring high on both axes become Tier 1 — candidates for continuous online monitoring or at minimum frequent periodic inspection. Mid-range assets suit periodic PdM (monthly or quarterly rounds). Low-criticality assets may remain on scheduled PM or even run-to-failure strategies.
For a typical Singapore manufacturing plant, Tier 1 assets commonly include main production motors, cooling tower fans, air compressors, transformer banks, and refrigeration plant. For a data centre or pharmaceutical facility, UPS systems, precision CRAC units, and standby generators also rank Tier 1.
Step 2 — Selecting the Right PdM Technologies
Each PdM technology targets a specific failure mode. A mature programme uses several in combination. The table below summarises the most widely deployed technologies and their primary applications:
| Technology | Primary Failure Modes Detected | Typical Assets |
|---|---|---|
| Vibration analysis | Bearing defects, imbalance, misalignment, looseness | Motors, pumps, fans, compressors, gearboxes |
| Infrared thermography | Electrical hotspots, refractory failure, mechanical friction | Switchboards, motors, steam traps, bearings |
| Ultrasonic testing | Compressed air/gas leaks, bearing lubrication state, partial discharge | Compressed air systems, electrical HV equipment, bearings |
| Motor current analysis | Rotor bar defects, stator faults, load anomalies | Three-phase induction motors |
| Oil analysis | Contamination, viscosity breakdown, wear debris | Gearboxes, hydraulic systems, transformers |
| Insulation resistance testing | Insulation degradation, moisture ingress | Motors, cables, switchgear |
| Thermal imaging of pipework | Blockages, heat loss, steam trap condition | Steam distribution, chilled water, refrigerant lines |
Unitest Instruments supplies calibrated instruments for each of these technologies — from Fluke vibration analysers and thermal cameras to Coltraco ultrasonic instruments — and can advise on the right tool for your specific asset portfolio.
Step 3 — Establishing Baselines and Setting Alert Thresholds
PdM data is only actionable when compared to a baseline. Baselines must be collected when the equipment is in known good condition — ideally at commissioning or after a full overhaul — and under consistent operating conditions (load, speed, temperature). Without a reliable baseline, it is impossible to determine whether a reading represents normal operation or the early stages of degradation.
Alert thresholds are typically set in layers:
- Advisory (Yellow): parameter has shifted from baseline but remains within acceptable operating range. Increase monitoring frequency and investigate at the next planned opportunity.
- Warning (Amber): parameter has deteriorated to a level that suggests active degradation. Schedule an inspection within a defined short window (e.g., two weeks).
- Alarm (Red): immediate action required. The asset is approaching or has exceeded failure thresholds defined in ISO standards or OEM specifications.
Industry standards that provide threshold guidance include ISO 10816/ISO 20816 for vibration severity, NEMA MG-1 and IEC 60034 for motor insulation, and IEEE C57 for transformers. Your calibration laboratory can help verify that the instruments generating these readings are traceable — a requirement if the data is to be used for regulatory compliance or insurance purposes. Learn more about Unitest's calibration services.
Step 4 — Building the Inspection Route and Data Collection Protocol
Consistency is as important as accuracy in PdM. A reading taken at a different location on the machine housing, under a different load condition, or with an improperly calibrated instrument will not be comparable to previous readings, destroying the trend data that makes PdM work.
Inspection routes should specify:
- Exact measurement point locations (use paint marks or adhesive pads for vibration, and document thermal image framing for thermography)
- Machine operating conditions required at time of measurement (e.g., full load, steady-state temperature)
- Instrument settings and measurement parameters (frequency range, averaging, integration)
- Data storage format and naming convention
- Responsible technician and supervisor sign-off
Digital route-based data collection — using handheld analysers that store readings directly to a CMMS or PdM software platform — eliminates transcription errors and automatically flags out-of-tolerance readings. Many Fluke instruments support Bluetooth or USB data transfer to computerised maintenance management systems.
Step 5 — Instrument Calibration and Traceability
Every instrument used in a PdM programme must be calibrated at defined intervals and the calibration must be traceable to national standards. In Singapore, this means traceability through SAC-SINGLAS accredited laboratories. Unitest Instruments holds SAC-SINGLAS accreditation LA-2023-0845-C covering eight measurement disciplines — electrical, temperature, pressure, humidity/moisture, dimensional, force/torque, flow, and chemical — and is an ILAC-MRA signatory, meaning calibration certificates issued are internationally recognised.
Why does calibration matter so much for PdM? Because a trend that appears to show degradation may simply reflect instrument drift. If your vibration analyser overreads by 15%, you may schedule an unnecessary bearing replacement. If it underreads by 15%, you may miss a developing fault until it causes a catastrophic failure. Calibrated instruments eliminate this ambiguity.
Calibration intervals for PdM instruments should be reviewed annually as a minimum. High-use instruments, or those that have been dropped or exposed to harsh conditions, should be returned for calibration sooner. Read our guide on calibration frequency.
Step 6 — Data Analysis and Work Order Generation
Raw PdM data has no value until it is analysed and acted upon. A mature programme connects data collection to a CMMS (Computerised Maintenance Management System) so that when a reading crosses an alert threshold, a work order is automatically generated with the appropriate priority, skill requirement, and parts list pre-populated.
For smaller organisations without a CMMS, a structured spreadsheet with automated conditional formatting and a manual review cadence can achieve the same outcome, albeit with more administrative overhead. The key discipline is that every anomalous reading must result in a documented decision — investigate, schedule repair, increase monitoring frequency — and that decision must be time-stamped and attributable to a named responsible person.
Trend analysis is more powerful than snapshot analysis. A single high vibration reading may indicate a one-off event (a transient overload, a foreign object in the airstream). A rising trend over six consecutive measurement cycles is a strong predictor of impending bearing failure regardless of whether any single reading has crossed an absolute alarm threshold.
Team Training and Competency Requirements
PdM is a skilled discipline. Vibration analysis in particular has a formal competency framework (ISO 18436) with four certification levels, from basic data collector through to advanced analyst. Thermography practitioners should ideally hold ISO 18436-7 (thermography) certification. For Singapore facilities subject to MOM scrutiny, demonstrated analyst competency strengthens your defence in any incident investigation.
Instrument suppliers including Unitest Instruments can provide guidance on instrument operation and connect customers with certified training providers. When purchasing a new PdM instrument, always ask about application training and ensure the analyst assigned to interpret the data has appropriate experience for the failure modes being monitored.
Continuous Improvement and Programme Maturity
A PdM programme should be reviewed at least annually. Review questions include: Which assets generated the most alerts? Were those alerts confirmed as real faults or false positives? Did any failures occur that the programme did not predict? What was the cost avoidance achieved through planned versus unplanned repairs?
As the programme matures, you will refine thresholds, adjust inspection frequencies, add or remove assets from the monitoring schedule, and potentially integrate online continuous monitoring for the highest-criticality assets. The goal is a closed loop: measure, analyse, act, learn, and improve.
For facilities pursuing ISO 55001 asset management certification — increasingly sought by Singapore operators in energy, water, and transport infrastructure — a documented, data-driven PdM programme is a core requirement. The Unitest team can advise on instrumentation strategies that align with your asset management system requirements.