Although used successfully in the defense industries, prognostics and health management techniques must be adapted to the unique challenges faced by nuclear plants.

As components age, advanced information processing capabilities can support detailed equipment health assessments, enabling nuclear plants to achieve equipment reliability goals. Prognostics and health management (PHM) techniques – used successfully in the defense industries – monitor equipment degradation over time and provide informed estimates of remaining useful life.

EPRI is evaluating PHM’s applicability to the nuclear industry. As a first step, EPRI is developing guidance for sensor requirements, monitoring, and prognostic algorithms that would provide health assessment data for a pump motor. Subsequent development plans include a 2009 PHM demonstration on a medium-voltage motor and horizontal pump.

To apply PHM in the nuclear power industry, improvements will be needed in several areas, followed by integration with maintenance management processes.

• Sensors and Data Processing: As diagnostics are performed in a more automated fashion, uncertainty can be reduced with sensors that address specific failure modes. Further, since wiring costs for new sensors can be significant, wireless sensors with on-board data processing may offer a low-power, low-maintenance, lower-cost solution. On-board processing can also reduce the amount of data transferred to the plant network, reducing the typical data flood experienced when new sensors are added to the plant.
• Failure Modes and Effects Analysis (FMEA): FMEA supports improved diagnostics, but advances are needed to align observable failure symptoms with early warning indications from sensors, predictive maintenance tasks, operator rounds, and component health assessments.
• Diagnostics: Diagnostics are typically performed once a failure mode has progressed to a level affecting equipment performance. Advanced diagnostic algorithms – including statistical processing, artificial intelligence, and model-based reasoning – could accelerate detection of performance degradation and increase equipment reliability.
• Prognostics: While diagnostics indicate when a failure either has occurred or is near, prognostics provide an estimated time to failure. Failure prediction before degradation, or without an indication, is based primarily on prior knowledge of failure modes. Failure prediction after degradation is based on useful life projections. Prognostic techniques that incorporate ongoing research and lessons learned related to the physics of degradation are essential for remaining useful life calculations.