Automatic detection of reservoir influx in conventional drilling, managed pressure drilling and dual gradient drilling

Abstract

Reservoir influxes, or kicks, are well control incidents with the potential of severe consequences to health, safety and the environment, as well as economics. Although the main focus will always be to prevent such incidents from happening, drilling crew will also need to be able to spot reservoir influx as quickly as possible.

This thesis presents a method for automated detection of reservoir influx or losses based on simulations of the surface circulation system. Theoretical background for the causes of reservoir influx is presented. The rig circulation system and traditional mudlogging approaches are discussed, as well as a literature study of proposed new methods for the detection of reservoir influx. Focus has been on conventional drilling, but literature and applications for managed pressure drilling and dual gradient drilling have also been included.

MatLab simulation scripts have been generated to investigate the relationship between changes in pump rate and measured volumes at surface. The script is compared to data from the literature, as well as real drilling data, and tuned by the use of adaptive observer technology. Low-pass and high-pass filters are also employed.

Simulations show that the real volume behaviour of the circulation system on a drilling rig can be relatively accurately described through simple programming logics. The scripts demonstrate the possibility to remove these dynamics from the volumes being monitored, so that any observed changes will in fact be real indications of volume change.

While this thesis only presents the basics for such a method, a further development has the capabilities of being incorporated into an automated system. Fully functioning, this system would allow for setting closer alarms on the monitored volume, resulting in reduced amounts of false alarms as well as earlier kick detection.