When the Experienced Operator Leaves, So Does Your Reliability Program

The Invisible Reliability Program

A plant's formal reliability program lives in documents, PM schedules, and predictive maintenance intervals. But the real reliability program—the one that catches failures before they happen, that prevents cascading breakdowns, that keeps equipment running smoothly—often exists nowhere on paper. It lives in the body knowledge of operators who've spent years on the same equipment.

This operator knowledge isn't arcane or mysterious. It's precise and reproducible. An experienced operator knows the bearing temperature on a centrifugal pump should be between 140 and 155 degrees. They know what bearing noise sounds like at 50% load versus full load. They can tell if a seal is degrading weeks before failure because they've felt the subtle changes in vibration from month to month. They've learned how pressure fluctuations in the discharge line indicate upstream issues. They know which warning signs matter and which are normal for that particular equipment configuration.

This knowledge is built through repeated exposure—thousands of hours observing equipment behavior, learning what normal is, recognizing deviations from normal, and understanding what those deviations mean. It's not memorized from a manual. It's developed through sensory learning: seeing abnormal vibration patterns, hearing changes in motor pitch, feeling bearing heat through equipment housings, smelling early seal degradation. It's learned through trial, error, and correction over years. It's developed by being responsible for keeping equipment running and experiencing the consequences of missed early warnings.

This is the reliability expertise that walks out the door when experienced operators leave. And it's typically the moment plants discover just how much of their failure detection capability was never codified, never trained to others, and never systematized.

Why Documentation Misses the Point

Most plants try to capture this knowledge through documentation: equipment manuals, operating procedures, PM checklists, even video walkthroughs of experienced operators. The assumption is logical—if you document what experienced operators do, new operators can follow the same steps.

This fails at a fundamental level. A procedure can document actions. It cannot document perception. A procedure says "measure bearing temperature every shift." It doesn't teach someone to recognize that a bearing running at 152 degrees in February—which was normal five years ago—now indicates seal degradation because the ambient temperature profile has changed. A checklist says "listen for unusual noise." But it doesn't give someone the reference library of normal sounds across dozens of different load conditions, ambient temperatures, and operational modes.

Documentation attempts to codify pattern recognition without acknowledging that pattern recognition requires pattern exposure. You cannot teach someone to recognize abnormal if they don't have a deep internalization of what normal is. This is especially true on older equipment or equipment with design quirks. A 20-year-old pump is different from its manufacturing twin because it's worn differently, been operated through different thermal cycles, and accumulated specific vibration signatures. The pattern an operator recognizes is specific to that equipment instance, not a generic pump class.

When a new operator reads the documentation, they follow procedures correctly. But they're following steps without understanding context. They measure temperature but don't know if it's trending up or down relative to baseline. They listen for noise but don't have the auditory reference library to distinguish normal operation from early degradation. They report data, but they don't detect the subtle signal in the noise. The result is compliance without capability.

The Reliability Regression Curve

The impact is measurable and predictable. In the first weeks after an experienced operator leaves, plant failure detection rates drop sharply. New operators follow procedures, but they miss signals that experienced operators would have caught immediately. Anomalies that should trigger escalation or maintenance get passed through as "normal variation." Small deviations develop into larger problems before anyone escalates them.

On equipment with high failure frequency, this becomes obvious quickly. A bearing that should last six months begins failing every three months. A process that should run 95% uptime drops to 85%. But on equipment with lower failure frequency—systems that normally run months without intervention—the regression can remain hidden for much longer. Problems that experienced operators would have prevented develop slowly and manifest as "reliability decline" months later, long after the connection to workforce change has faded from organizational memory.

Some plants recover this capability over time. If new operators get 2-3 years of supervised experience on the same equipment, they develop pattern recognition capability, albeit usually less reliable than experienced operators had. But many plants don't have that luxury—they have equipment turnover in operators long before any operator can build deep expertise. The result is a plant that never builds institutional pattern recognition capability and always operates at a degraded reliability state.

Building Transferable Capability Before the Knowledge Walks Out

Preventing this requires a different approach to knowledge transfer. Instead of trying to document what experienced operators know, the goal is to accelerate the development of pattern recognition capability in new operators.

This starts with structured observation. Before an experienced operator leaves, they should be actively teaching by describing what they're observing and why it matters. Not by dictating procedures, but by explaining the pattern recognition process: "This bearing is running at 148 degrees. For this pump at full load in summer, that's normal. But three weeks ago when it was running at 152, that indicated early seal wear because the pump was at part load. The temperature drop tells me the seal wear progressed and we need to schedule replacement." This explanation teaches the decision-making framework, not just the action.

It continues with direct pattern exposure. New operators need to see, hear, and feel normal equipment behavior across different operating conditions. They need baseline data and trending context. They need to observe consequences—when they miss a signal, what happens? This can be accelerated by having experienced operators explicitly walk through different equipment states and explain what normal looks like at each state. It can be augmented by using sensor data and historical trending to help new operators develop data-informed pattern recognition, not just sensory pattern recognition.

Most critically, new operators need to have overlapping tenure with experienced operators. This isn't a two-week overlap. It's months of working together, with progressively increasing responsibility and decision-making authority. It's the experienced operator catching failures the new operator missed and explaining why, in real time. It's the new operator building the reference library through repeated exposure to actual equipment behavior, guided by someone who already has expertise.

The window for this transfer is finite. Once an experienced operator leaves, the opportunity to build that capability in others is gone. Plants that wait for an operator to announce retirement, then try to accelerate knowledge transfer in the final weeks, find that structured capability transfer takes months, not weeks. The message is clear: if you have equipment-critical knowledge concentrated in one or two people, your transfer timeline needs to start years in advance of their departure.

The Immediate Action

Identify your critical equipment and the operators who run it. If an operator has been running the same equipment for more than 10 years and their departure would cause an immediate capability gap, start planning the knowledge transfer now. Don't wait for retirement announcements. Structure observation sessions, implement baseline data collection, and begin building pattern recognition capability in the next tier of operators. The goal isn't to replace their knowledge—it's to preserve the failure detection capability that knowledge enables.