The Debate That Shouldn’t Be a Debate
Walk into any maintenance conference and you’ll hear people argue about predictive versus preventive maintenance as if they’re opposing philosophies. They’re not. They’re complementary tools in a maintenance strategy toolkit. The right answer for any given piece of equipment depends on its failure patterns, criticality, and the economics of monitoring versus scheduled intervention.
Getting this mix right can reduce total maintenance costs by 25-40% while simultaneously improving equipment availability. Getting it wrong means you’re either spending money on unnecessary scheduled maintenance or missing failures that could have been detected and planned.
Preventive Maintenance: The Calendar and Hour-Based Approach
Preventive maintenance (PM) performs maintenance activities at fixed intervals — every 3 months, every 2,000 hours, every 500,000 cycles. The assumption is that equipment wears out predictably over time and intervention at regular intervals prevents failure.
When PM Works Well
- Age-related failure modes. Rubber components (seals, hoses, belts, diaphragms) degrade predictably based on time, temperature, and chemical exposure. Scheduled replacement before the expected end of life works because the failure pattern is strongly age-related.
- Regulatory and compliance requirements. Safety relief valves, fire suppression systems, pressure vessels, and emissions equipment often have mandated inspection intervals. These are non-negotiable regardless of condition.
- Consumables. Filters, lubricant changes (when condition monitoring isn’t applied), and wear items with known service lives are straightforward PM candidates.
- Simple, inexpensive equipment. Small motors, fractional-HP pumps, solenoid valves — the cost of condition monitoring exceeds the cost of scheduled replacement. PM or run-to-failure makes more sense than investing in monitoring technology.
When PM Falls Short
The fundamental problem with PM is that it assumes failure probability increases with age. For many industrial components, this isn’t true. The Nowlan and Heap study commissioned by United Airlines in 1978 — the research that led to RCM — found that only 11% of component failure patterns were age-related. The remaining 89% showed either random failure probability or infant mortality patterns.
This means that scheduled overhauls on complex equipment (turbines, compressors, gearboxes) at fixed intervals may actually introduce more failures than they prevent. Reassembly errors, installation damage, and infant mortality of new components create a spike in failure probability right after the PM is performed. The industry term for this is “maintenance-induced failure,” and it accounts for a larger share of failures than most maintenance managers realize.
Predictive Maintenance: The Condition-Based Approach
Predictive maintenance (PdM) monitors equipment condition and performs maintenance only when the data indicates a developing problem. The equipment tells you when it needs attention, rather than a calendar or hour meter.
When PdM Works Well
- Random failure patterns. Bearings, gears, and most rotating equipment components fail based on operating conditions, contamination events, and manufacturing variations — not age. Condition monitoring catches the onset of degradation regardless of when it occurs.
- Expensive equipment with long lead times. A critical gearbox with a 16-week lead time on replacement gears needs early warning. PdM provides months of lead time for planned repair. PM provides scheduled downtime that may or may not coincide with actual need.
- Equipment where unnecessary maintenance is risky. Opening a gearbox for inspection introduces contamination risk. Replacing a mechanical seal that still has life remaining wastes money and introduces the risk of installation error. PdM avoids unnecessary interventions.
- High-consequence failures. Equipment whose failure causes safety incidents, environmental releases, or major production losses justifies the investment in monitoring technology and trained analysts.
When PdM Falls Short
- Failure modes without detectable precursors. Some failures happen without warning. Electronic component failures, sudden fractures from manufacturing defects, and instantaneous seal blowouts don’t provide the gradual degradation signal that PdM technologies need to detect.
- Very slow degradation. Corrosion-under-insulation on piping progresses over years. Annual thickness measurements might suffice, but calling this PdM overstates its sophistication. Scheduled inspection at appropriate intervals handles this adequately.
- Low-cost, high-population equipment. Monitoring 500 small motors individually with vibration analysis costs more than replacing them when they fail. Statistical approaches (tracking failure rates across the population) make more sense than individual monitoring.
The Right Mix: How to Decide for Each Asset
Use this decision framework for each significant piece of equipment:
Step 1: Identify the Dominant Failure Modes
For each asset, list the 3-5 most likely failure modes based on historical data and engineering knowledge. Determine whether each failure mode is predominantly age-related or random.
Step 2: Evaluate Detectability
For each failure mode, ask: is there a measurable parameter that changes as the failure develops? Vibration increases before bearing failure. Temperature rises before insulation breakdown. Oil contamination increases before gear wear becomes critical. If a detectable precursor exists and the P-F interval (time between detectable onset and functional failure) is long enough to plan a repair, PdM is viable.
Step 3: Compare Economics
Calculate the cost of the PdM program for this asset (monitoring equipment amortized across all monitored assets, technician time for data collection and analysis, software costs) versus the cost of the PM program (parts, labor, downtime for scheduled maintenance whether needed or not). Factor in the cost of failure under each strategy — PdM catches most developing failures but not all; PM prevents some failures but induces others.
Step 4: Consider the Hybrid Approach
Most equipment benefits from a combination of both strategies. A centrifugal pump might have:
- PdM: monthly vibration monitoring for bearings (random failure pattern)
- PdM: quarterly oil analysis for lubrication system (contamination-driven)
- PM: annual mechanical seal replacement on a critical service (age-related degradation plus high consequence)
- PM: semi-annual coupling element replacement (rubber element with known life)
- RTF: drain plugs, gaskets, paint — replace when they fail or during planned outages
This hybrid approach applies the right strategy to each failure mode rather than treating the entire pump with a single maintenance philosophy.
Shifting the Balance Over Time
Plants starting their reliability journey typically run 60-70% reactive, 25-30% PM, and less than 5% PdM. The natural progression is to first convert reactive work to PM (stopping the bleeding), then gradually shift appropriate PM tasks to PdM (optimizing the strategy).
Target state for a mature maintenance organization:
- Less than 10% reactive/emergency work
- 25-35% preventive maintenance (age-related failure modes and compliance)
- 45-55% predictive/condition-based maintenance (random failure modes on critical equipment)
- 5-10% run-to-failure (deliberate, documented decisions for low-consequence equipment)
This transition takes 3-5 years and requires sustained investment in technology, training, and cultural change. The payoff — lower total maintenance costs, higher equipment availability, and fewer surprises — makes it one of the best investments a manufacturing or process facility can make.
Stop arguing about PdM versus PM. Start analyzing your equipment failure modes and applying the right strategy to each one. That’s where the value lives.