You have a 100 HP cooling water pump with a bearing that the manufacturer says to replace every 12 months. So every year, you shut the pump down, pull the bearing, and install a new one. The old bearing looks fine every time. Your technician holds it up and says, "This thing had another year in it."
Meanwhile, another identical pump in a dirtier, hotter part of the plant chews through the same bearing in 7 months because of contaminated water and higher operating temperatures. But it is on the same 12-month schedule, so it fails catastrophically at month 9 before the PM is due.
Condition-based maintenance (CBM) fixes both problems. Instead of replacing parts on a calendar, you monitor the equipment's actual condition and only act when the data says it is time. The clean pump keeps its bearing for 20 months. The dirty pump gets a replacement at 6 months before the damage gets serious.
What is Condition-Based Maintenance?
Condition-based maintenance is a maintenance strategy where you monitor the real-time condition of equipment and perform maintenance only when indicators show that performance is declining or a failure is approaching.
The core idea is simple: equipment tells you when it needs attention. A bearing gets hotter and louder before it fails. A filter shows increasing pressure drop before it clogs. A motor draws more current before it burns out. CBM uses measurements to detect these changes and trigger maintenance at the right time.
CBM sits between time-based preventive maintenance (fixed schedule, regardless of condition) and fully predictive maintenance (using trend analysis and algorithms to forecast future failures). In practice, the boundaries between CBM and PdM are blurry, and many people use the terms interchangeably.
The practical distinction: CBM acts when a measurement crosses a threshold right now. PdM uses historical trend data to predict when that threshold will be crossed in the future. CBM says "the bearing temperature is 195F, replace it now." PdM says "the bearing temperature is trending upward at 3F per week, it will hit the 200F limit in 4 weeks, schedule replacement for next weekend."
A Real Example: Bearing Temperature Monitoring
Let's walk through a concrete example of CBM in action. You have a 75 HP motor driving a process fan in a cement plant. The motor has two bearings: drive end (DE) and non-drive end (NDE).
You install a temperature sensor on each bearing housing. The sensor sends a reading to your monitoring system every 5 minutes. Here is what the data looks like over 6 months:
| Month | DE Bearing Temp (F) | NDE Bearing Temp (F) | Status |
|---|---|---|---|
| Month 1 (baseline) | 145 | 142 | Normal. Both bearings running cool. |
| Month 2 | 147 | 143 | Normal. Slight increase on DE, within range. |
| Month 3 | 155 | 144 | Watch. DE bearing trending up. NDE stable. |
| Month 4 | 168 | 145 | Alert. DE exceeds warning threshold (165F). Schedule repair. |
| Month 4.5 (repair) | 146 | 144 | Bearing replaced during planned shutdown. Back to normal. |
What happened here: the DE bearing started degrading around Month 2, probably due to contamination or lubrication breakdown. By Month 4, the temperature crossed the warning threshold. The maintenance team scheduled a planned replacement during the next available window, two weeks later. They replaced the bearing in 2 hours of planned downtime.
Without CBM, this motor would have been on a 12-month PM schedule. The bearing would have failed catastrophically around Month 5 or 6, causing an unplanned outage of 6-8 hours, potential damage to the shaft and seal, and $15,000-$25,000 in emergency repair and production loss. The CBM approach caught it early and the planned repair cost under $800.
CBM vs Time-Based vs Predictive: Three Strategies Compared
| Factor | Time-Based PM | Condition-Based (CBM) | Predictive (PdM) |
|---|---|---|---|
| Trigger for action | Calendar or usage counter | Threshold exceeded now | Predicted future threshold crossing |
| Monitoring required | None | Periodic or continuous | Continuous with trend analysis |
| Planning lead time | Known in advance (fixed schedule) | Days to weeks | Weeks to months |
| Unnecessary maintenance | High (parts replaced early) | Low | Very low |
| Technology cost | None (just scheduling) | Moderate (sensors, instruments) | Higher (sensors + analytics software) |
| Analysis complexity | None | Basic (compare to threshold) | Advanced (trend modeling, ML) |
| Typical cost savings vs reactive | 20-25% | 25-35% | 35-45% |
The key takeaway: CBM is the natural next step after time-based PM. It does not require the advanced analytics of full PdM, which makes it accessible for plants that are not ready for machine learning and AI-powered analytics. If you can read a temperature gauge or a vibration meter and compare it to a threshold, you can do CBM.
Condition Monitoring Techniques
CBM requires measuring something about the equipment's condition. Here are the most common monitoring techniques, what they detect, and what equipment they work best on.
| Technique | What It Measures | Best Equipment Types | Typical Cost Per Point |
|---|---|---|---|
| Temperature monitoring | Bearing temp, motor temp, process temp | Motors, pumps, compressors, bearings | $50-$300 (wireless sensor) |
| Vibration monitoring | Vibration amplitude and frequency | All rotating equipment | $200-$800 (wireless sensor) |
| Current monitoring | Motor current draw (amps) | Electric motors, drives | $100-$400 (CT sensor) |
| Pressure monitoring | Differential pressure, system pressure | Filters, heat exchangers, hydraulic systems | $100-$500 (pressure transmitter) |
| Oil analysis | Particle count, viscosity, water, metals | Gearboxes, hydraulic systems, engines | $25-$75 (per sample, lab analysis) |
| Ultrasonic detection | High-frequency sound emissions | Bearings, valves, steam traps, air leaks | $2,000-$8,000 (handheld instrument) |
| Thermal imaging | Surface temperature distribution | Electrical panels, motors, insulation, piping | $3,000-$30,000 (IR camera) |
For most plants starting a CBM program, the combination of temperature monitoring and vibration monitoring covers 70-80% of the failure modes on rotating equipment. These two techniques together give you the best return for the investment.
Setting Threshold Alerts
The effectiveness of CBM depends on setting the right alert thresholds. Set them too low, and you get constant false alarms that your team learns to ignore. Set them too high, and you miss developing failures.
Most CBM programs use a three-tier alert system:
Normal (Green)
Readings are within the expected operating range. No action needed. For bearing temperature on a standard motor, normal is typically 110-160F depending on ambient conditions and load.
Warning (Amber)
Readings have moved outside the normal range but have not reached a critical level. Action: increase monitoring frequency, investigate the cause, and plan a repair. For bearing temperature, warning is typically 165-185F.
Alarm (Red)
Readings indicate a failure is imminent if action is not taken soon. Action: schedule repair at the earliest available opportunity. For bearing temperature, alarm is typically above 185F. Above 200F, consider emergency shutdown to prevent catastrophic failure and shaft damage.
Where do these threshold numbers come from? Three sources, in order of reliability:
- Your own historical data. If you have been taking readings on this equipment for 6+ months, you know what normal looks like. Set your warning threshold at 2-3 standard deviations above the mean. This is the most accurate approach because it accounts for your specific equipment and operating conditions.
- Industry standards. ISO 10816 provides vibration severity standards for different machine classes. ISO 7919 covers shaft vibration. These are widely accepted starting points.
- Manufacturer specifications. The equipment manual often lists maximum operating temperatures, vibration limits, and pressure ranges. Use these as your alarm thresholds and set warning thresholds 10-20% below them.
The P-F Curve: Understanding the Window of Opportunity
The P-F curve is the most important concept in CBM. It shows the progression from the point where a failure first becomes detectable (Point P, for "potential failure") to the point where the equipment actually fails (Point F, for "functional failure").
The time between P and F is your intervention window. CBM works by detecting the problem at Point P and scheduling the repair before you reach Point F.
The P-F interval varies by failure mode and equipment type. For a bearing defect on a motor, the P-F interval might be 3-6 months: vibration analysis detects the early defect months before the bearing actually seizes. For an electrical insulation breakdown, the P-F interval might be only days to weeks.
This is why your CBM monitoring frequency must be shorter than the P-F interval. If your P-F interval for a bearing is 3 months, monthly vibration readings will catch it. Weekly readings give you even more lead time. But if you only take readings quarterly, you might miss a fast-developing defect.
A good rule of thumb: set your monitoring interval to no more than half the P-F interval. If the P-F is 3 months, monitor monthly. If the P-F is 2 weeks, monitor daily or continuously.
CBM Implementation Steps
Here is a practical step-by-step approach to implementing CBM at your plant.
Step 1: Identify candidate equipment
Not every machine needs CBM. Focus on equipment that meets at least two of these criteria: high production impact when it fails, repair cost greater than $5,000, known failure modes with measurable indicators, and failure P-F intervals longer than 2 weeks.
Start with 5-10 machines. Trying to instrument your entire plant in Month 1 is a recipe for data overload and incomplete implementation.
Step 2: Select monitoring techniques
Match the monitoring technique to the failure mode you are trying to detect. Use the decision tree above as a starting guide. For most rotating equipment, temperature and vibration monitoring cover the majority of common failure modes.
Step 3: Establish baselines
Before you can detect abnormal conditions, you need to know what normal looks like. Take readings on each machine when it is running well and in its typical operating state. Collect baseline data over 2-4 weeks to account for normal variation from load changes, ambient temperature shifts, and production cycles.
Step 4: Set alert thresholds
Use the three-tier system (normal, warning, alarm) described above. Start with manufacturer specs or industry standards, then refine based on your own data as you accumulate operating history.
Step 5: Define the response process
An alert without a clear response process is useless. Define what happens when each alert level is triggered:
- Warning alert: Increase monitoring frequency. Investigate during the next available maintenance window. Create a work order in your CMMS for planned repair within 2-4 weeks.
- Alarm alert: Schedule repair at the next production break. Order parts if not in stock. Create a priority work order. Target repair within 1-2 weeks.
- Critical alarm: Notify maintenance supervisor immediately. Evaluate whether the machine can safely continue running. Schedule emergency repair or controlled shutdown.
Step 6: Train your team
Technicians need to understand what the sensors are measuring, what the alert levels mean, and what actions to take. This does not require deep analytical expertise. A 4-hour training session covering sensor basics, threshold meanings, and the response process is sufficient for most technicians.
For the person reviewing the data (often the reliability engineer or maintenance planner), deeper training on trend analysis and diagnostic interpretation is valuable.
Step 7: Review and optimize
After 6 months, review your CBM data. Adjust thresholds based on actual experience. Add monitoring to additional equipment if the pilot proved value. Look for patterns: if the same type of equipment keeps triggering alerts, there may be a systemic issue (wrong lubricant, incorrect installation procedure, design limitation) worth investigating with a root cause analysis.
CBM and Your Maintenance Metrics
CBM directly improves the key maintenance metrics your management cares about:
- MTBF increases because you catch and fix problems before they become full failures. A bearing replacement during a planned stoppage is a PM event, not a failure event. It does not reset your MTBF clock.
- MTTR decreases because when you do have a repair, you already know what is wrong (the CBM data told you), the parts are pre-staged, and the work is planned. Diagnosis time drops from hours to minutes.
- OEE improves because both Availability (fewer unplanned stops) and Performance (equipment running at designed speed because it is in good condition) go up.
Where Dovient Fits
Dovient adds a critical layer to your CBM program: the knowledge that connects sensor data to practical repair actions.
- From alert to action plan. When a CBM sensor triggers an alert, Dovient's AI diagnostic engine immediately provides context: past repairs on this equipment for similar symptoms, the most likely root cause based on your plant's history, and the recommended repair procedure. Your technician does not just know something is wrong. They know exactly what to do about it.
- Repair knowledge that builds over time. Every CBM-triggered repair is documented in Dovient with the sensor data that triggered it, what the technician found, and how it was fixed. Over time, this builds a knowledge base that makes your CBM program smarter. Patterns emerge: "Every time vibration on this pump exceeds 0.4 in/s at 2x RPM, it is a coupling alignment issue, not a bearing problem."
- Threshold refinement from actual data. Dovient's repair history helps you refine your CBM thresholds. If your warning threshold triggered at 165F and the actual failure investigation showed the bearing was fine at that temperature, you can safely raise the threshold. If you set an alarm at 185F but the bearing was already badly damaged, the threshold was too high.
Ready to see where your equipment stands today? Try our free OEE Calculator to measure your current performance, or schedule a conversation with our team to discuss implementing CBM at your plant.