A Systematic Approach to Bearing Health Monitoring

This paper presents recent advances at Honeywell's Condition Based Maintenance Center of Excellence (CBM COE) in developing a bearing health monitoring system for a gas turbine engine. Our systematic approach characterizes bearing damage progression by identifying damage milestones based on bearing geometry. For each damage milestone, we derive appropriate condition indicators and model the damage progression as a function of these indicators. Using an oil debris monitoring (ODM) sensor, we developed a condition indicator based on particle count and size to detect the onset of a spallation while ignoring debris "fuzz." An additional condition indicator based on particle accumulation rate quantifies bursts of particles after spall initiation. A third condition indicator based on accumulated particulate mass is used throughout the failure progression and is tuned to the specific stage of damage evolution. The ODM-based condition indicators are complemented with vibration-based condition indicators. More than 300 condition indicators (CI) are defined based on vibration signals and bearing geometry. These CIs are grouped and processed hierarchically to produce second level indicators for bearing damage isolation. A two-stage fusion method, based on fuzzy logic and using ODM and vibration CIs, provides robust detection and isolation. The resulting bearing health indicators are mapped onto on-board and on-ground notices for pilots and maintainers. These notices support confirmation of impending failure by external evidence from oil filter analysis. The approach is demonstrated using rig and engine test data and scenarios describing how this approach facilitates scheduling and coordinating ground logistics are discussed.