Data Analyses of Quarry Operations and Maintenance Schedules: A Production Optimization Study

Abstract

In this research, data analytics and machine learning were used to identify the performance metrics of loaders and haul trucks during mining operations. We used real-time collected data from loaders and haul trucks operating in multiple quarries to broaden the scope of the study and remove bias. Our model indicates relationships between multiple variables and their impacts on production in an operation. Data analysis was also applied to ground engagement tools (GET) to identify key preventative maintenance schedules to minimize production impact from capital equipment downtime. Through analysis of the loader’s data, it was found there is an efficient cycle time of around 35 s to 40 s, which yielded a higher payload. The decision tree classifier algorithm created a model that was 87.99% accurate in estimating the performance of a loader based on a full analysis of the data. Based on the distribution of production variables across each type of loader performing in a similar work environment, the Caterpillar 992K and 990K were the highest-yielding machines. Production efficiency was compared before and after maintenance periods of ground engaging tools on loader buckets. With the use of maintenance and production records for these tools, it was concluded that there was no distinguishable change in average production and percentage change in production value before and after maintenance days.