A Proportional Hazards Models perspective on MTBF estimation for centrifugal pumps

Centrifugal pumps are critical assets in oil and gas operations, where failures can trigger severe production losses, safety hazards, and environmental incidents. This industry faces shrinking margins, rising operational costs, and growing regulatory and sustainability pressures, making reliability a strategic priority. To help in addressing these issues, predictive reliability has evolved beyond traditional maintenance practices to ensure operational integrity and risk reduction, particularly for rotating equipment.

While many studies focus on component‑level prognostics using high‑frequency sensor data, these approaches often require extensive real‑time monitoring that is not feasible in many plants. This keynote introduces a system‑level framework for MTBF prediction of API centrifugal pumps using Proportional Hazards Models, leveraging operational data from an oil refinery. The methodology integrates design parameters, operating conditions, vibration severity, and maintenance records to deliver robust reliability estimates under data constraints.

Results highlight the dominant influence of maintenance practices and operating conditions, with vibration severity emerging as a critical factor aligned with ISO standards. By quantifying these effects, the framework provides actionable insights for optimizing maintenance strategies, improving asset performance, and reinforcing process safety and environmental protection in oil and gas facilities.