查看文章

This paper presents a predictive maintenance-planning solution in the oil and gas industry. The solution focuses on the mechanical components of bottomhole assemblies, which are the equipment that goes underground in well-construction activities. Two components were successfully modelled, namely crossovers and collars. They are tubular structures that have to be repaired (“cut”) after performing a drilling job to remove worn material. A linear behaviour was observed for the degradation of crossovers. The subsequent linear model predicted the remaining jobs of a fleet of crossovers with a mean absolute percentage error of 6.5%. With these predictions and two thresholds for the minimal number of remaining jobs in a fleet and for its minimal number of crossovers, the purchase of new crossovers could be automatically triggered. Collars, due to low data quality, were modelled just by using the median of the …