Engineering Equipment Failures and Inventory Management Shifts. Statistical Insights from Papua New Guinea’s Mining Sector

Abstract

The operation of surface mines in Papua New Guinea (PNG) is associated with the intense mechanical loading, remote logistics, and the tendency of failure of the significant quantity of heavy engineering equipment. These ailments lead to huge downtime and ineffective provisioning of spare parts. In this proposed study, a statistically integrated system is designed and proposed which integrates predictive analytics of maintenance with inventory management techniques in order to solve the issues of making operations vulnerable to certain surface mining settings. A real time multi-sensor network containing pressure, temperature, vibration, flow rate and oil contamination sensors was employed to collect data. Within six months, this multivariate data was trained to classify four operational states of a device, normal, degrading, impending failure, and critical, where the classification was performed by a Random Forest model. The predictive model expressed great reliability with an overall accuracy of 94.8 percent and the recall of 91.3 percent on the impending failure events. Results show it could reduce delays by 37 percent fewer delays in provisioning, and a 23 percent increase in an alignment of inventory.