Fusion of Operations, Event-log and Maintenance Data: A Case Study for Optimizing Availability of Mining Shovels

Abstract: The modern mining industry is highly mechanized and relies on massive, multimillion-dollar pieces of equipment for achieving production targets. In an increasingly challenging international economic climate, mining operations are reliant on economies of scale to remain competitive. To maximize revenue, it is imperative that at each stage of the mining process, equipment is operating optimally without preventable and unnecessary interruptions. As a result, the focus of all mining operations is to increase equipment uptime and utilization.
The Aitik open pit copper mine is located approximately 15 km south-east of the town of Gällivare in Northern Sweden. It is the largest copper mine in Sweden and uses a conventional truck-shovel operation. The loading fleet at Aitik consists of several power shovels to load trucks with blasted material for transport, either to the crushers or to the waste dumps. The maintenance activities for mining shovels are critical when it comes to keeping them available for production. Hence, any attempt to achieve improvements in the maintenance function for mining shovels, is a worthwhile endeavour.
The Aitik mine employs various computer-aided applications to track and maintain mobile mining equipment. These applications also serve as chronological operational and maintenance databases for the equipment. The work presented in this thesis comprises six mining shovels and is based on the analysis of three data types; historical maintenance data from CMMS Maximo, operational data from mine management system Cat® MineStar™, and event-log data from individual shovels. The data sets span over two years from March 2010 to March 2012 and, were sourced from the maintenance department at the Aitik mine. As part of this investigation, the three distinct data sets were analysed separately and further research was undertaken to integrate them. For a standalone investigation of individual data types, separate methods of analysis were required. Historical maintenance data consisted of over sixteen and a half thousand individual work orders. These were analyzed to identify shovel subsystems and components with respect to the number of associated corrective maintenance work orders. From MineStar data, maintenance key performance indicators such as availability, mean time between failures, and mean time to restore were calculated, using European standard EN15341:2007. Such KPIs are useful for identifying underperforming assets within the shovel fleet, and provide a quantifiable means for benchmarking maintenance performance. Finally, frequently recurring events were identified through a Pareto analysis and were correlated with historical maintenance data.
The results from this study indicate towards viable prospects for such a synthesis. A regular time-lapse integration of the diverse data types displays potential; and could prove to be helpful, for achieving overall improvements in the maintenance function. In addition, this integration could prove to be beneficial from an operational standpoint. For instance, such a data synthesis can aid in the development of a proactive and collaborative approach towards asset management.