Introduction to Predictive Maintenance: Concepts, Program Structure, and the P-F Curve

What Is Predictive Maintenance?

Every industrial and marine facility must keep equipment running safely and cost-effectively. To do that, maintenance teams choose a strategy. There are three main strategies, and understanding how they differ is the foundation of this entire course.

Reactive maintenance means waiting until something breaks before fixing it. It requires no planning, but the costs are severe: unplanned downtime, emergency parts procurement, collateral damage to nearby equipment, and—most importantly—a high risk of injury. A pump that seizes unexpectedly can rupture lines, release hazardous fluids, and injure workers nearby.

Preventive maintenance (PM) replaces or services components on a fixed schedule, regardless of actual condition. Changing bearing grease every 2,000 hours whether the bearing needs it or not is a classic PM task. PM reduces surprise failures, but it wastes resources when components are replaced too early and can still miss failures that develop between service intervals.

Predictive maintenance (PdM) monitors the actual condition of equipment using technology—vibration analysis, infrared thermography, oil analysis, ultrasound, and others—to detect the earliest signs of degradation. Maintenance is performed only when evidence indicates it is needed. PdM combines the safety of PM with far less wasted labor and material.

The goal of a well-run PdM program is not simply to collect data. It is to make the right maintenance decision, at the right time, on the right asset.

The P-F Curve: Your Roadmap for Condition Monitoring

The P-F curve is the single most important concept in predictive maintenance. It describes the life of a failure from its first detectable sign to the point of functional failure.

• Potential-failure point (P): The earliest moment a monitoring technology can detect that something is wrong. At this point, the equipment still works, but degradation has begun.
• Functional-failure point (F): The moment the asset can no longer perform its required function—it has failed.
• Net P-F interval: The time between P and F. This interval is the window a maintenance team has to plan and execute a repair before a functional failure occurs.

The shape of the curve matters. The asset performs at full function for a period, then condition begins declining. Detection tools can pick up a signal at P; without action, the asset reaches F. Between P and F, the curve often steepens as damage accelerates.

Why does this guide technology selection? Different technologies detect failures at different points on the curve. Ultrasound and advanced vibration analysis can detect very early-stage bearing defects—far left on the curve, giving a long P-F interval. Simple vibration amplitude checks may not catch a problem until much closer to F. For a high-criticality pump on a marine vessel with no redundancy, you want the technology that gives the longest interval and the most lead time.

Why does this guide inspection frequency? If your P-F interval for a particular failure mode is 30 days, you must inspect at least every 15 days to have a reasonable chance of catching the failure in time to act. A rule of thumb: inspection interval should be no more than half the P-F interval.

Structure of a PdM Program

A sound PdM program has several components:

1. Asset criticality ranking: Identify which equipment failures would cause the greatest safety risk, production loss, or environmental harm. High-criticality assets get the most monitoring attention.

2. Failure mode identification: For each asset, list the ways it can fail and the physical change that accompanies each mode (heat, vibration, contamination, electrical noise).

3. Technology selection: Match a monitoring technology to each failure mode based on what physical change it detects and where on the P-F curve it gives detection.

4. Inspection intervals: Set intervals based on the P-F interval for each failure mode.

5. Data collection and analysis: Trained technicians collect readings on schedule, compare them to baselines and alarm limits, and trend data over time.

6. Work-order generation: When a reading exceeds a threshold, a work order is issued for planned corrective action before failure.

7. Continuous improvement: Track program metrics—failures prevented, cost avoidance, mean time between failures—and adjust the program.

Safety in PdM Work

Predictive maintenance often involves working around live, energized equipment—a critical safety consideration.

• Lockout/tagout (LOTO): When a PdM task requires physical contact with or access to hazardous energy sources, follow your facility's written LOTO procedure (OSHA 29 CFR 1910.147 is the governing standard). PdM does not exempt you from LOTO.
• PPE: Wear appropriate personal protective equipment for each task. Infrared thermography near electrical panels requires arc-flash-rated PPE. Vibration routes near rotating machinery require safety glasses and no loose clothing or jewelry.
• Route safety: Never reach over or across unguarded rotating parts to place a sensor. Use remote or permanently mounted sensors when access is unsafe during operation.
• Hazardous atmospheres: On marine vessels and in chemical plants, confined spaces and flammable atmospheres may require gas testing and a work permit before any monitoring route begins.

PdM improves safety not only because it prevents sudden failures but also because it replaces frequent intrusive PM tasks—many of which expose workers to hazards—with non-intrusive monitoring.