The objective of this study was to accurately measure the degradation of equipment indirectly (without the use of sensor or machine data). This was done in a semi-autonomous manner by leveraging the bulk data in the organization's maintenance system and its underlying metadata relationships. This study included a novel combination of two traditional techniques for asset management: The P-F curve from Condition Based Maintenance (CBM), and the "Crow – United States Army Material Systems Analysis Activity (Crow-AMSAA)". The P-F curve is a widely accepted tool to represent the degradation and eventual failure of assets. It is used to approximate the lead time to a specific failure occurring, where "P" represents the point where a failure can first be detected, and "F" represents the point where the actual functional failure occurs. Traditionally this curve is derived in a CBM system using knowledge-based or data-driven approaches that depend heavily on field sensor data. This study instead uses a statistical based method (Crow-AMSAA) to approximate the curve based on properties of maintenance record data and the intensity of data generation within the maintenance system. As theorized in previous studies, we can indeed observe changes in maintenance record properties prior to rig stoppage events (equipment failures) and approximate the P-F curve. The underlying input variable from the maintenance system was shown to be the Mean Time Between Corrective Maintenance (MTBCM), which is derived from qualitative and quantitative properties of maintenance records (work orders or jobs). When this value is visually modelled over a period of time based on the Crow-AMSAA method, a visual representation of the P-F curve can be easily identified. It was also shown as possible to determine a threshold value for taking action prior to failure. To represent this, the Crow-AMSAA output graphics were augmented with additional metadata and features that underly the maintence records. With these proper visual aids and regression techniques implemented, it is possible to identify threshold values in units of MTBCM, as well as days (unit of time), for specific assets. This allows for actions to be implemented prior to imminent failures caused by degradation growth.

This content is only available via PDF.
You can access this article if you purchase or spend a download.