Early Kick Detection Using Adaptive Analytics with Downhole Accelerometer Data

Abstract

High frequency downhole vibration data includes more hidden information than low frequency surface data. This paper discusses monitoring high frequency acceleration data for early kick detection because the accelerator sensor values’ trend is monitored rather than processed. When gas, fluid, or oil kick occurs, fluid influx reduces fluid viscosity in the annulus, causing degradation of the damping factor. The sensor installed on the drillpipe detects the velocity/acceleration modification, resulting in the damping factor modification and includes an analytical model to calculate the effect of the damping ratio on the acceleration calculations. Fluid influx and migration in the wellbore strongly affects the damping factor. This paper discusses a method of deconvoluting sensor values that use a combination of minimum entropy deconvolution and the Teager-Kaiser energy operator to remove the noise, unwanted sensor values, and the likelihood of false predictions. The trend of the final intrinsic mode functions (IMFs) at each depth is continuously monitored to predict formation influx, if any. A novel concept of monitoring incremental IMF and IMF energy at each depth is introduced, revealing a wealth of information and simplifying the process of monitoring and analyzing the vast amount of available data. The methodology developed extracts essential information from high frequency vibration data to make real-time data monitoring straightforward, reliable, and fast.