Kick Detection Using Downhole Accelerometer Data

Abstract

The high-frequency downhole vibration data include a greater amount of hidden information than the low-frequency surface data. This paper proposes the monitoring of high-frequency acceleration data for early kick detection. The trend of accelerator sensor values is monitored, rather than processed. When the gas, fluid, or oil kick occurs, the fluid influx reduces the viscosity of the fluid in annulus, which causes the degradation of the damping factor. The sensor installed on the drillpipe detects the velocity/acceleration change that results in the damping factor change. This approach includes an analytical model to calculate the effect of the damping ratio on the acceleration calculations. The fluid influx and migration in the wellbore strongly affect the damping factor. The paper presents a method of deconvoluting the sensor values that uses a combination of minimum entropy deconvolution and Teager-Kaiser energy operator to remove the noise, unwanted sensor values, and likelihood of false prediction. It is then proposed to calculate instantaneous jerk and jerk intensity at each depth. The trend of the final intrinsic mode functions (IMF) at each depth is continuously monitored to predict the formation influx, if any. A novel concept of monitoring the incremental IMF and IMF energy at each depth is introduced. This technique is shown to reveal a wealth of information and simplifies the process of monitoring and analyzing the vast amount of available data. The methodology developed is applied to extract the essential information from high-frequency vibration data to make real-time data monitoring straightforward, reliable, and fast.