Georgia Pacific (GP) paper mills must maintain 100% uptime for six weeks. Paper travel is controlled by dozens of rollers turning on bearings. The bearings that are used in the dryer rolls of paper making machines operate in hot environments because high temperature steam is passed through the rolls. They are prone to catastrophic failure which can force replacement of a costly roller and many hours of downtime. GP was making bearing replacement at 60% of total life as a preventative failure method. Emerson wired vibration sensors are connected to each bearing. But they were only analyzed once per month, and unexpected catastrophic failures can occur within that time window.
Use Prognostics to Track RUL
DecisionIQ prepared a diagnostic analysis of rotating bearings using prognostics methodology and machine learning to determined remaining useful life (RUL). We ran the bearing waveform data through our platform to extract features that indicated degradation and failure. High-confidence estimates of remaining useful life were possible given the low sample rates of the wired sensor data. There were also many gaps in data, and imputation methods were used to fill in gaps.
Significant Reduction in Replacements
Bearing degradation models were able to accurately predict