Real-time data analytics of drilling sensor streams for prediction of critical situations

Publication from Digital

Herwig Zeiner, Bernhard Jandl and Martin Winter and Roland Unterberger, Rudolf Fruhwirth and Christian Derler

Poster at European Data Forum 2012 , 1/2012


The aim of this real-time sensor data analytics system is to make better decisions and to predict and avoid upcoming critical situations in drilling operations. The purpose is to support the drilling engineers with the ongoing processes and give them deeper insights into the current situations. This should help to avoid upcoming critical situation. It is of utmost importance to prevent damages to equipment, protect the crew from injuries, and avoid environmental pollution in drilling operations nowadays. We present a real-time data analytics prototype. The proposed research prototype has been applied to real world data of oil or gas reservoirs in onshore regions as well as in offshore regions. The research prototype consists of a complete data processing chain including several modules for data acquisition from sensors on the drilling platform, adaptive sensor data analytics and problem specific visualization. While the data acquisition modules collect the data from sensors at the rig and produce a live data stream in an appropriate WITS like format, the data processing algorithms have to analyze the data streams in real-time and classify the drilling operations, detect potentially abnormal upcoming critical events and give appropriate advice to the drilling crew, if possible. Finally, all the raw sensor data streams as well as several adaptive online learning algorithms results and several sensor channel quality parameters of the rig are visualized in a novel user interface to support drilling employees at the rig. Following that the current drilling situation is presented in a comprehensive manner and in real-time.