What is Context-aware Condition-based Maintenance (CBM) ?
The world is transitioning towards Condition-Based Maintenance (CBM), a strategy where real-time sensor data is used to predict potential failures in advance.
Various sensors such as vibration, temperature, ultrasound, and supply current are installed on the outside of industrial assets, making them non-invasive and quick to install. The data from these sensors are processed using algorithms, which include physics-based and data-driven approaches.
In both approaches, the algorithms establish a nominal condition and then monitor for any deviations that indicate abnormal conditions. This works well if the asset has one nominal condition; however, if the asset has multiple nominal conditions, this could reduce the effectiveness of the condition based maintenance algorithm.
Where does context-aware CBM outperform traditional asset monitoring?
In many industries, equipment power ramps up and down depending on operational condition. For example, in oil and gas drilling operations, assets such as mud pumps and top drives work harder as they drill deeper. The mud pump pressure would increase, causing more vibration, and the top drive torque would increase, drawing more supply current.
This can cause CBM algorithms to misdiagnose the increase in vibration or supply current as abnormal, while in fact they actually represent a different nominal condition for the assets. This challenge is addressed by context-aware condition based maintenance, where operational data such as pressure, speed, torque, RPM, and drilling state are used to add context to the sensor data.
This brings multiple benefits:
Increased prediction accuracy: By combining operational data with sensor data, prediction accuracy can be significantly improved, especially for assets with multiple nominal states.
Improved explainability: By incorporating operational data, changes in sensor data trends can be easily understood.
Improved automation: By applying operational data, analytics algorithms can be adjusted automatically based on the current nominal state, reducing the need for manual sensor data analytics.
The challenge with context-aware condition based maintenance is that operational data and sensor data come from different sources. For example, drilling operational data comes from an Electronic Data Recorder (EDR) or control system, while sensor data comes from a third-party sensor provider, so data integration is required to bring the two data sources together.
If you are interested in enhancing your sensor-based CBM solutions with context-aware operational data, contact us for more information.