Semantic web technologies for event analysis in reservoir engineering

UnrestrictedAn event refers to an action or a change of state of an object in a specific domain. Efficient integration and analysis of event information is critical for timely reaction to certain events in event driven decision making processes. The challenges to such event analysis in an industrial domain like the oilfield industry include: 1) The event information is scattered among various unstructured or semi-structured legacy files. These have to be manually screened by experienced workers to infer useful information; 2) Personnel working on different areas must coordinate in analyzing event relationships. This is challenging due to various formats and conventions employed in different organizations.; To address these challenges, we apply semantic web technologies to the integration and analysis of event information. Specifically, we design an extensible data model to uniformly represent events using OWL. Based on the event model, we define rules to represent the logic of event relationships. As a result, event information can be stored in a uniform and structured manner that is more easily accessed and shared. We also propose an event management framework to leverage a rule engine to automatically correlate events using the rules.; We apply our event management framework to real use cases in the oilfield domain. We design and develop an application for integrating event information from legacy files in reservoir simulation and visualizing the events on top of forecast curves. The application assists domain experts to analyze simulation results more efficiently by automatically aggregating and correlating event information. We also design and develop a well surveillance prototype which detects underperforming wells based on event information collected from the reservoir. The prototype provides similar functionality as a state-of-the-art expert system used in the domain. The extensibility of the event model and the event management framework is demonstrated by these applications. The performance of both the applications is evaluated to meet the requirements of a reasonably large number of wells and events to be processed