The Problem:
Unplanned equipment downtime was occurring due to trip limits on grinding mill motors being triggered.
The solution:
The Electrical reliability team was set up with email and SMS alerts before the trip limit was reached to allow sufficient time for corrective actions to be undertaken.
Three levels of alert were deployed to avert trip limits being exceeded
- Level 1 alert – Weekly anomaly reports identified trends and anomalies in anomalies to predict downtime events days or weeks in advance.
- Level 2 alert – Real time rate of change or skewing of normalised data. This alert indicated to the area engineer that closer monitoring of key parameters and process conditions for this equipment was required.
- Level 3 alert – Tag approaching trip limit or significant rate of change detected. Immediate investigation required.
FORESTALL was both supervised and unsupervised machine learning to detect anomalies in the condition and performance of the motors. FORESTALL automatically identified normal operation by analysing trends in performance data and noted when these trends and relations deviate from normal operation, reflected as anomalies in the data.
The Results:
Level 1 FORESTALL automatically analysed mill motors data weekly, where a ‘data sweep’ was scheduled each week to identify anomalies, skewing or significant rates of change. Following the data sweep, the appropriate alerts are to be sent via email and SMS to the designated personnel.
The chart below shows accuracy of Level 3 motor stator current prediction using FORESTALL. The prediction was two hours in advance and was used to accurately predict trips before they occurred.
Downtime was prevented by providing advanced warning of trip limits 1-2 hours before they were exceeded.
FORESTALL is available now for demonstration, please contact us to arrange a walk through.
