Reducing unplanned downtime using Predictive Maintenance (PdM)

Abstract

Abstract Maintenance is one of the essential aspects that contribute to keep production process continue to operate and reduce unplanned downtime. In the cement industry, especially kiln area, unplanned downtime can cause not only low availability but also high maintenance costs due to unplanned demand for spare parts and the need for labour to complete unplanned downtime, hence the right method of maintenance is needed. One of the strategic maintenances could be applied is Predictive Maintenance (PdM). Among others maintenance strategy, PdM provides more accurate task due to real time condition monitoring of equipment based on sensors obtained from server data. In this paper, Mahalanobis-Taguchi System (MTS) is applied to provide reliable multisensory analysis and real-time decision making. MTS includes 1) fuses data from multiple sensors into a single system level performance metric; 2) extends MTS by providing a single tool for fault detection/diagnostic, isolation, and prognostic; and 3) offers a systematic way to determine the key parameters following with fault reducing analysis. Multi-sensor data in the kiln used in this research are speed motor main drive, Pfister Coal feed, kiln feed, speed motor IDFAN1, speed motor IDFAN2, speed motor EPFAN. Based on the MTS calculation, the normal data values are obtained from each sensor data. This method is also used to detect, identify, and classify failures based on tests from scheme data. The proposed diagnostic and prognostics scheme from this paper are useful for guiding the company to predict the failure as well as the maintenance schedule then it will increasing the availability from 80% to 90% by reduce the unplanned downtime.