Your plant has 40 motors. You are changing bearings on all of them every 12 months because that is what the maintenance schedule says. But here is the reality: some of those bearings are perfectly fine at 12 months and have another 6-8 months of life left. Others are already in trouble at 9 months because of misalignment, overloading, or contamination.
Predictive maintenance (PdM) solves this problem. Instead of replacing parts on a calendar schedule, you monitor the equipment's actual condition and only intervene when the data tells you a failure is developing. You catch the bearing that is going bad at 9 months and leave alone the one that is still running strong at 14 months.
The result: fewer unnecessary part replacements, fewer surprise breakdowns, and maintenance work that happens exactly when it is needed. Plants with mature PdM programs report 35-45% less unplanned downtime and 8-12% lower total maintenance costs compared to time-based preventive maintenance alone.
How Predictive Maintenance Works
PdM follows a straightforward logic. Equipment does not fail without warning. Before every failure, there is a period of measurable deterioration. A bearing gets rougher. Vibration increases. Temperature rises. Electrical insulation degrades. Oil contamination levels climb.
PdM uses sensors and instruments to detect these changes early, while there is still time to plan and schedule the repair. The process has three steps:
1. Collect condition data
Use sensors or instruments to measure the equipment's current condition. This can be continuous (permanently installed sensors sending data 24/7) or periodic (a technician takes readings with a handheld instrument on a monthly route).
2. Analyze the data
Compare current readings against baseline values and alert thresholds. Look for trends: is vibration gradually increasing? Is temperature climbing? Has the oil contamination level changed since the last sample? The analysis can be done manually by trained analysts or automatically by software.
3. Predict and act
When the data shows a developing problem, estimate how long before it becomes a functional failure. Schedule the repair during a planned window before the failure occurs. The technician knows what is wrong (the diagnosis came from the data), has the right parts ready, and performs a controlled repair instead of an emergency scramble.
PdM Technologies
There are four primary PdM technologies used in industrial settings. Each one detects different types of developing failures.
Vibration Analysis
This is the most widely used PdM technique for rotating equipment: motors, pumps, fans, compressors, gearboxes, and turbines. A vibration sensor (accelerometer) measures how much the machine is shaking and at what frequencies.
Different failure modes produce different vibration signatures. A bad bearing creates high-frequency vibration at specific bearing defect frequencies. Misalignment creates vibration at 1x and 2x the shaft rotation speed. Imbalance shows up at 1x rotation speed. A trained analyst or software can read these signatures and tell you exactly what is failing.
What it catches: Bearing defects (3-6 months before failure), misalignment, imbalance, looseness, gear wear, belt defects.
Cost: Handheld vibration analyzer: $5,000-$25,000. Permanent wireless sensors: $200-$800 per point. Monthly route-based program for 50 machines: $2,000-$5,000/month including analysis.
ROI example: One detected bearing defect on a critical 200 HP motor avoids a $40,000 unplanned downtime event. The entire vibration program pays for itself with one save.
Thermal Imaging (Infrared Thermography)
An infrared camera shows you the temperature distribution across equipment surfaces. Hot spots indicate problems: overloaded electrical connections, failing bearings, blocked cooling passages, refractory damage, steam trap failures, and insulation breakdown.
What it catches: Electrical hot spots (loose connections, overloaded circuits), mechanical friction (bearings, couplings), heat transfer problems (blocked fins, fouled heat exchangers), insulation gaps in buildings and piping.
Cost: Industrial IR camera: $3,000-$30,000 depending on resolution. Quarterly survey of electrical panels and rotating equipment: $1,500-$4,000 per survey.
ROI example: A thermal survey finds a loose bus bar connection in an MCC drawing 300 amps. Left unchecked, that connection overheats and causes an arc flash incident. The repair cost: $200 and 30 minutes of downtime. The avoided arc flash: potentially $500,000+ in equipment damage, facility repair, and safety consequences.
Oil Analysis
You send a sample of lubricating oil from a gearbox, hydraulic system, or engine to a lab. The lab tests for particle count, viscosity, water content, elemental analysis (iron, copper, chromium, silicon), and acid number. The results tell you what is happening inside the machine without taking it apart.
What it catches: Abnormal wear (metal particles indicate which component is wearing), contamination (water, dirt, wrong oil), oil degradation (oxidation, viscosity breakdown), and coolant leaks.
Cost: $25-$75 per sample for standard analysis. Quarterly sampling of 20 machines: $2,000-$6,000/year including sample kits and lab fees.
ROI example: Oil analysis on a gearbox shows elevated iron and chromium particles. This indicates gear tooth wear. Scheduled repair during a planned shutdown costs $8,000. Catastrophic gearbox failure would cost $45,000 plus 3 days of lost production.
Ultrasonic Testing
Ultrasonic instruments detect high-frequency sounds that humans cannot hear. Failing bearings, leaking valves, compressed air leaks, and electrical arcing all produce ultrasonic emissions. A technician with an ultrasonic detector can scan equipment and pinpoint problems quickly.
What it catches: Early-stage bearing defects (detects problems before vibration analysis), compressed air leaks (a typical plant wastes 20-30% of compressed air through leaks), steam trap failures, electrical arcing and corona discharge.
Cost: Handheld ultrasonic detector: $2,000-$8,000. Compressed air leak survey for a plant: $3,000-$8,000 (often pays for itself immediately through energy savings).
ROI example: An ultrasonic air leak survey in a plant with a $200,000/year compressed air energy bill finds leaks responsible for $45,000/year in wasted energy. Fixing the leaks costs $3,000 in labor and fittings. Payback: 24 days.
PdM vs Preventive Maintenance: When Each Makes Sense
PdM does not replace preventive maintenance. It supplements it. Here is a direct comparison to help you decide which approach fits each piece of equipment.
| Factor | Preventive Maintenance | Predictive Maintenance |
|---|---|---|
| When maintenance happens | Fixed schedule (time or usage) | When data indicates need |
| Risk of over-maintenance | High (replacing good parts on schedule) | Low (only act on actual condition) |
| Risk of under-maintenance | Moderate (if interval is wrong) | Low (continuous or periodic monitoring) |
| Setup cost | Low (procedures and schedules) | Medium to high (sensors, software, training) |
| Skill requirements | Standard maintenance skills | Specialized analysis skills or software |
| Best for | Consumables, filters, lubrication, safety checks | Critical rotating equipment, high-value assets |
| Typical cost reduction vs reactive | 20-25% | 30-40% |
The practical approach: use PM for routine tasks (lubrication, filter changes, safety inspections) and PdM for condition-sensitive components on critical equipment (bearings, gears, electrical connections). Most mature plants run about 60% PM, 30% PdM, and 10% reactive.
When PdM Pays Off: ROI Calculation Example
Let's work through a real ROI calculation for a vibration monitoring program on 25 critical motors in a food processing plant.
| Item | Annual Cost |
|---|---|
| PdM Program Costs | |
| Wireless vibration sensors (25 motors x 2 points x $400) | $20,000 (Year 1 only) |
| Software platform license | $8,000/year |
| Analyst time (internal, 4 hours/week) | $12,000/year |
| Total Year 1 cost | $40,000 |
| Total Year 2+ cost | $20,000/year |
| Avoided Costs (based on historical data) | |
| Avoided unplanned motor failures (was 8/year, now 2/year) | $120,000 |
| Avoided production loss (6 events x 4 hours x $3,000/hour) | $72,000 |
| Reduced overtime labor (fewer emergency calls) | $18,000 |
| Extended bearing life (fewer unnecessary replacements) | $15,000 |
| Total annual savings | $225,000 |
Year 1 ROI: ($225,000 - $40,000) / $40,000 = 462%
Year 2+ ROI: ($225,000 - $20,000) / $20,000 = 1,025%
These numbers are aggressive but realistic for a food processing plant where production loss is $3,000+ per hour. In lower-value production environments, the ROI is smaller but still strongly positive. As a rule of thumb: if you have more than 10 critical rotating machines and your unplanned downtime costs more than $1,000/hour, PdM will pay for itself within the first year.
PdM Implementation Roadmap
Do not try to implement PdM across your entire plant at once. Start small, prove value, and expand. Here is a proven roadmap.
Stage 1: Pilot (Months 1-3)
Pick 5-10 of your most failure-prone critical machines. These are machines where you know failures are costing you real money. Start with vibration analysis if they are rotating equipment, or thermal imaging if they are electrical systems.
Take baseline readings on each machine when it is running normally. These baselines are your reference point. All future analysis compares against these baselines.
Stage 2: Prove Value (Months 3-6)
The goal of this stage is to demonstrate ROI. Every time the PdM program catches a developing problem, document it: what was found, what would have happened without PdM, and what the planned repair cost versus the avoided unplanned repair cost.
After 3-6 months, you should have 3-5 documented saves. These real examples are worth more than any vendor ROI projection when making the case for program expansion.
Stage 3: Expand (Months 6-12)
Extend monitoring to 25-50 machines. Add a second PdM technology. Set up a formal alert and action process: when a PdM alert fires, a work order gets created in your CMMS with the diagnosis and recommended action.
Stage 4: Optimize (Year 2+)
By now you have enough PdM data to start optimizing your PM schedules. If vibration monitoring shows that bearings on Motor A consistently last 18 months instead of the 12-month PM schedule, extend the PM interval. This reduces unnecessary work and frees up technician time for higher-value activities.
Common PdM Mistakes
These are the mistakes that cause PdM programs to fail or underperform:
- Starting too big. Trying to monitor 200 machines in Month 1 overwhelms your team with data they are not ready to handle. Start with 5-10 machines, learn the process, then expand.
- Collecting data without acting on it. If PdM data sits in a software dashboard and nobody creates work orders from it, you have an expensive data collection program, not a maintenance program. Every PdM alert needs a clear action path.
- No baseline readings. Without baselines, you cannot tell if a reading is normal or abnormal. Always take baseline readings when equipment is in good condition.
- Expecting instant results. PdM detects developing failures. If a bearing is going to fail in 4 months, PdM catches it at month 1 and gives you 3 months to plan the repair. But you have to be monitoring for that month to catch it. Results build over time.
- Skipping the analysis step. Raw vibration data is meaningless without analysis. Investing in sensors but not in the expertise to interpret the data is like buying a blood pressure cuff but not knowing what the numbers mean. Either train an internal analyst or hire a service provider.
- Ignoring the organizational change. PdM changes how your maintenance team works. Technicians need to trust and act on PdM recommendations. Planners need to schedule PdM-driven work orders. Production needs to release equipment for condition-based repairs. If you do not manage this change, the technology alone will not deliver results.
PdM and Related Maintenance Strategies
PdM is one piece of a complete maintenance strategy. Here is how it connects to other approaches:
- Preventive Maintenance (PM) handles routine, consumable-based tasks (lubrication, filter changes, calibration). PdM handles condition-sensitive components. Use both.
- Condition-Based Maintenance (CBM) is closely related to PdM. The distinction: CBM monitors current condition and acts when a threshold is crossed. PdM goes further by using trend data to predict when a threshold will be crossed in the future.
- MTBF improves as PdM reduces the number of unexpected failures. Track MTBF before and after PdM implementation to measure reliability improvement.
- OEE improves because PdM reduces unplanned stops (Availability) and prevents the speed reductions that come from running degraded equipment (Performance).
Where Dovient Fits
Dovient adds intelligence to your PdM program by connecting sensor data with the maintenance knowledge your team has built over years of operating your equipment.
- Contextual diagnostics. When a vibration sensor flags a developing bearing defect on Pump 7, Dovient's AI diagnostic engine immediately shows the technician: how this failure was handled last time on similar equipment, which parts were needed, and how long the repair took. This turns a PdM alert into an actionable repair plan.
- Knowledge capture from every repair. Every PdM-driven repair is a learning event. Dovient captures what the technician found, what the actual root cause was, and how it was fixed. Over time, this builds a knowledge base that makes your entire team smarter about your specific equipment.
- Repair history patterns. Dovient tracks whether PdM-detected issues are recurring on specific equipment. If the same motor keeps developing bearing defects every 9 months, that points to a systemic issue (misalignment, overloading, wrong bearing specification) that PdM alone will not solve. Dovient surfaces these patterns for root cause analysis.
Want to start measuring how your equipment is performing today? Try our free OEE Calculator to establish your baseline. For a deeper conversation about PdM implementation at your plant, schedule a conversation with our team.