Abstract
The oil and gas drilling industry is amid a digital transformation, especially regarding the digital oilfield, smart operations, and predictive maintenance. Traditional scheduled maintenance approaches to equipment replacement have been proven as inefficient[EP(1] [YR(2], as not just time in service, but operational and system variables have a significant impact on useful life. (Temer and Pehl, 2017). Reactive approaches to maintenance by responding to failure, has increased risks for downtime and safety, especially in consideration of critical drilling and safety components at the well center critical path. This paper describes an approach, through an in-line blowout preventer valve (IBOP) predictive failure use case, of smart equipment monitoring and reliability centered maintenance (RCM), supported by modern digitalization technologies, and the industrial internet of things (IOT).
To achieve this, extraction, transformation, and loading various data into a more unified analysis platform for usage along with loading aggregations and forecasting analysis was performed. Analyzing the SCADA data from the drilling rig, a logic-based algorithm was applied to identify IBOP usage cycles, from several separate signals. This created a dense data set of time series data identifying equipment loading cycles allowing analysis of a reliability centered maintenance health assessment and threshold. Historical usage was analyzed and forecasted using autoregressive integrated moving average to identify statistical approximation for the date where the forecasted usage would meet the health threshold, with results visualized into an interactive dashboard for operators.
The identification of a forecasted date where equipment usage is expected to cross the reliability centered maintenance threshold can be used for rig maintenance preparation and planning. Decision layer content, in the form of an interactive dashboard and scheduled reporting, can be employed to keep maintenance crews aware of usage progression toward the threshold.
Acquisition of data for analysis from many different systems and formats is challenging, especially for technical and engineering disciplines not fully aligned with traditional information technology (IT) skillsets, techniques, or platforms. Enabling flexibility for experimentation for new analysis can be by code-free, but code-friendly, data acquisition and analysis platforms that can be utilized at the onset of these advancements toward RCM. A data-fusion approach, methodically using multiple digitalization platforms, is employed as a strategy for improving condition monitoring, health assessment, and awareness. The application of new feature engineering and machine learning (ML) offers iterative improvements to RCM based health prognostic signals.
