Development and Testing of a Rig-Based Quick Event Detection System to Mitigate Drilling Risks

Abstract

Abstract High accuracy, speed, and a low false alarm rate are critical to the success of event detection systems and the reduction of non-productive time (NPT) due to unplanned events. The balance between them has been difficult to achieve; however, the application of a probabilistic approach with the use of Bayesian methods has shown significant promise. The results of a research prototype for kick detection using such methods were described in a previous paper (Hargeaves, Jardine, Jeffryes, 2001). The improvements in automated signal processing technology and increased computer power have meant that these new approaches to large scale data analysis in real time are now more widely possible. This paper describes the extension of this work to develop a rig-based quick event detection (QED) system for mitigating a number of common drilling risks, including kicks, circulation loss, and drillstring washout detection. The QED software application runs on a standard standalone PC at the wellsite and continuously calculates the probability that an event is occurring. The probability changes as an event occurs, which may trigger an alarm. The system continues to evaluate the event as the real-time data is updated and this evaluation provides the drilling team with a continuously evolving probability that the event is occurring. This system helps the team make timely, informed decisions. The system requires minimal hands-on interaction to set up and uses data input via WITS for flow-in and flow-out measurements from typical rig sensors. The detection of drilling problems in real time using the QED software on several Integrated Project Management well construction projects is presented, including the lessons learned from deploying the technology in field operations. The significance of this development is to mitigate drilling risks during well construction projects, where minimizing NPT due to unplanned events is critical to success. Introduction The main problems during drilling are related to events such as kicks, stuck pipe, wellbore collapse, lost circulation, and equipment failures (Nilsen, 2002). Analysis of the majority of drilling problems has shown that there were early indications, which if correctly interpreted, could have helped to avoid the problem. However, it is well documented that post analysis is of limited use to real-time drilling operations. In order to realize the true value it is necessary to perform analysis in real time to provide key information as early as possible to the drilling team while it is still possible to take corrective action to reduce the cost and severity of events (Havrevold, Hwien, Parigot, 1991). There are a number of technology advances such as the increasing availability of real-time drilling data, from both surface and downhole sensors, open standards for real-time data access (Cayeux et al, 2006), new techniques in signal processing, and the computing power to process the data in real time. In 2006, events within an Integrated Project Management company involving well control cost USD 3.3 million, and lost circulation USD 2.6 million. Application of the QED software on integrated well construction projects is now demonstrating the potential to significantly reduce these through early detection and improved risk mitigation. QED was initially developed by Schlumberger Cambridge Research as a proof of concept of Bayesian methods to real-time data analysis and event detection. The identification of kicks from the analysis of delta mud flow and tank volume changes was selected as the first application of this technique. This process has been described previously (Hargeaves, Jardine, Jeffryes, 2001). Field testing during drilling in the Burgos field in Northern Mexico in 2002 revealed several practical issues, including inconsistent data processing, which resulted in excessive false alarms and difficulties in managing the complex code required to calculate probabilities and normalize and sum the results. As a result, field personnel did not accept the technology.