IMC-2019 Presentation 37:29 Minutes

by Randy Yount, Noble Reliability and Craig Churchman, Capstone Pinto Valley Mine

Using oil analysis to predict future equipment failures has been used for many decades. It is generally considered reliable but is placed far down the predictive – failure timeline because of the long intervals between oil samples. By automatically doing real time oil sampling every 5 minutes and having the data interpreted by an Internet of Things (IoT) cloud server, the predictability of failures is greatly increased. Adding specialized algorithms and Artificial Intelligence (AI) to the analysis can also determine the type of wear occurring, such as sliding vs abrasion vs adhesion. In addition, the source of contaminants can often be identified, especially when external contaminants are present. Seal failures can be identified in minutes and alerts sent out. Water ingress into the oil can also be quickly identified. Overall machine health can also be monitored much more closely. Oil analysis at 5 minute intervals provides significant data that can be overlaid onto contextual data such as normal runs, starts, stops and maintenance events. The software learns from the additional context and can provide early warning when similar events begin in the future. Data will be presented from actual field experiences in multiple locations with different types of equipment.