Robust estimation and protection of locked charge in grinding mills

One very important charge condition that can produce a catastrophic failure during the starting of grinding mills is the compacting of the load inside the mill, especially after long downtimes. This condition is usually named frozen load and the load remains locked to the mill shell and after a half turn drops damaging the mill body and bearings. Therefore, the mill protection system must give an early detection of this condition and aborts the start before the compacted mill charge falls hitting the bottom of the mill shell. Actual protection schemes against locked load use fix torque or current trip values and do not consider the effect of the specific charge inside the mill on each startup. This paper presents the development of a robust estimation and protection algorithm of locked load condition based on drive and process signals. The proposed algorithm includes a locked load model used to define torque, rms current and power thresholds for the existing starting charge conditions, and compares the locked load signals to the online estimation of drive variables. This comparison is used to detect the beginning of charge cascading, and the threshold values are used to command the starting abortion. Main advantage of the proposed protection algorithm is that considers the effect of the mill charge and charge position, and therefore gives an accurate and safe detection of locked load for any charge condition during the mill starting. Proposed algorithm was evaluated using field records of an 8.2 MW semi-autogenous (SAG) mill drive confirming its ability to handle startups with different charge conditions.