Description
In 2010 Shell began investigating how to automate the initial response to a well control incident.
Since at least 25% of kicks in deep water GoM wells occur on connections it was quickly realized that robust kick detection during connections was important but especially challenging due to the associated transient flow and pit volume signatures. A work stream was kicked-off to develop new software based on pattern recognition technology and machine learning. The resulting IDAPS (Influx Detection at Pumps Stopped) software has now been implemented as a real-time monitoring application for all Shell operated GoM deep water wells. Presentation will include IDAPS development roadmap and implementation results, including adding a ballooning discriminator.