Centrifugal Casting Machines (CCMs) are used for casting ductile iron pipes at Tata Metaliks. These machines are operated using hydraulic systems. Main traversing cylinder (MTC) of these machines is a very critical component and is directly affected by any issue in the hydraulic system, especially pump failure. There are 5 hydraulic power packs at the CCMs having a cumulative of 22 swash plate operated pressure compensated piston pumps of which 13 pumps are used for MTC operation.

Recurrent failure of hydraulic pumps at the corresponding power packs used to be a critical problem at Tata Metaliks. This abstract outlines activities undertaken under the Pump Condition Monitoring project and insights generated from the same.

These pumps were fitted with case drain ultrasonic flowmeters to gauge the amount of wear on the shoe retainer plates. Oil contamination sensors measuring the TAN delta value of the oil in these power packs were installed to gauge oil quality and monitor its degradation over time. Resistance Temperature Detectors (RTDs) were installed in order to record the trend of oil temperature for actionable insights, as high oil temperature can lead to seal failures and therefore oil leakages. Laser based oil level sensors were installed in place of the float based type in order to have a reliable alert when oil level dips below the appropriate level thereby avoiding tripping and damaging of the pumps. Pressure transmitters were installed in order to gauge the system pressure and for effective monitoring of pump performance.

This data was pushed onto a SCADA system to generate actionable insights. Basic descriptive analytics were built using the SCADA platform wherein the entire power pack was recreated as in real life. All major KPIs were mapped onto real time alerting mechanisms. On basis of inputs received from these alerts and dashboards, the pumps were changed over adaptively. This helped avoid critical pump damage, which used to render the pumps useless earlier. Upon changeover, these pumps were removed and overhauled, wherein lapping was done on the retainer plates. Thereafter, the pumps could be taken back on line.

This initiative enabled us to extend the pumps’ usable life beyond the earlier observed life. Till Dec’20, the Business Analytics Group approved year on year savings resulting from the project was estimated to be around ₹ 18 lakhs.

To increase the effectiveness of the system, all the values derived were fed into an unsupervised machine learning algorithm to derive the overall power pack health, any deviation is marked with a possible anomaly and alerts over mail.To increase the effectiveness of the system, all the values derived were fed into an unsupervised machine learning algorithm to derive the overall power pack health, any deviation is marked with a possible anomaly and alerts over mail.