Producing assets face a pressing demand to minimize unplanned maintenance. The use of basic monitoring consoles such as DCS, are limited in scope, provide isolated views and reactive in nature. This paper presents a case study of using predictive analytics in an offshore oil production facility. The facility operations performance is constantly monitored by sensors sitting across an entire installation, measuring relevant information such as temperatures, pressure levels and flow rate at different time intervals.
Predictive analytics was applied to integrate all relevant data for proactive maintenance. The predictive model uses advanced analytics to estimate problem specific KPIs that is used to generate automated monitoring, alerting and event specific reports. These help to provide advance notice of impending failures and minimizing maintenance-related disruptions in operations.
The production facility consists of many integrated components as part of the production process. Continuous workings of these components are required to meet production targets. However, some critical components malfunction without warnings and, in the worst case, halt altogether resulting in short term production loss. The purpose of the predictive model was to use data from the entire process to find any measurable characteristics that may serve to warn that these detrimental situations are approaching. Moreover, the project also aimed at providing suitable diagnostics of the situation. In this technical paper, we described a successful application of an innovative statistical method for proactive asset management using predictive analytics.
The predictive models were able to successfully provide warnings of impending failures for the component of interest and identify root causes of such impending failures.