Utilization of Big Data Analytics and Advanced Formation Modelling for Detection of Kicks in Drilling Operations

Abstract

A Kick, the influx of formation fluid into the wellbore while drilling, poses a major challenge to drilling operations and can spiral out of control into blowouts with severe fatal, fiscal and environmental consequences. Kicks characteristically have a higher occurrence when drilling in relatively unexplored formations and with the combined factors of a waning era of easy oil and increasing energy demand, the consequent push for petroleum exploration in unconventional formations demands better techniques to detect and control kicks. This work has detection of kicks as its objective. Traditional methods of detecting kicks by monitoring drilling mud levels in the tanks has proven to be cumbersome and error prone and it leaves little time for an effective response. Thus, the use of analytics of real time drilling data and advanced formation modelling is presented as an approach to create a better representation of the drilling environment sub-surface and identify potential threats of a kick along the course of drilling (with respect to the trajectory as well as decisions to be made following that course). The methodology seeks to create a comprehensive model that defines relevant physical parameters whose values can be used as data sets which describe the ongoing drilling process and its relationship with the background formation with the aim of bringing forth information which would give a representation of consequent events. Notable parameters include, porosity, rock density, drill string hook load, weight on bit (WOB), mud density, formation fluid resistivity, rate of penetration, ultra sound speed across media, drilling trajectory amongst others, all relative to time. The background formation is aptly described in discretized grid blocks and is then cross-matched with the real-time data from the drillstring to double-check the actual position of the drillstring at any point in time. The interactions of the formation with the drillstring trajectory are computed as described by the grid blocks in contact with the drill string trajectory as well as adjacent grid blocks. The data describing the formation can be regularly updated to represent whatever sensitive changes that might have occurred in the formation while drilling. This solution, though notably complex is well within the capacity computing power available in the upstream petroleum industry and shows great promise to eliminate all the disastrous consequences that arise from late detection of kicks.