Data Quality of Fleet Management Systems in Open Pit Mining: Issues and Impacts on Key Performance Indicators for Haul Truck Fleets
MetadataShow full item record
Open pit mining operations typically rely upon data from a Fleet Management Systems (FMS) in order to calculate Key Performance Indicators (KPI’s). For production and maintenance planning and reporting purposes, these KPI’s typically include Mechanical Availability, Physical Availability, Utilization, Production Utilization, Effective Utilization, and Capital Effectiveness, as well as Mean Time Between Failure (MTBF) and Mean Time To Repair (MTTR). This thesis examined the datasets from FMS’s from two different software vendors. For each FMS, haul truck fleet data from a separate mine site was analyzed. Both mine sites had similar haul trucks, and similar fleet sizes. From a qualitative perspective, it was observed that inconsistent labelling (assignment) of activities to time categories is a major impediment to FMS data quality. From a quantitative perspective, it was observed that the datasets from both FMS vendors contained a surprisingly high proportion of very short duration states, which are indicative of either data corruption (software / hardware issues) or human error (operator input issues) – which further compromised data quality. In addition, the datasets exhibited a mismatch (i.e. lack of one-to-one correspondence) between Repair events and Unscheduled Maintenance Down Time states, as well between Functional-Failure events and Production states. A technique for processing FMS data, to yield valid Functional Failure events and valid Repair events was developed, to enable accurate calculation of MTBF and MTTR. A concept for identifying data quality issues in FMS data, based upon an examination of feasible durations for Production states (TBF’s) and Unscheduled Maintenance states (TTR’s), was developed and implemented through the duration-based filtering of both these categories of state. The sensitivity of the KPI’s in question to duration based filtering was thoroughly investigated, for both TBF and TTR filtering, and the consistent trends in the behavior of these KPI’s in response to the filtering were demonstrated. These results have direct relevance to continuous improvement processes applied to haul truck fleets.