Detection of Bubble Flow by Cluster Analysis of Ultrasound Waves’ Spectral Properties

Abstract

As a non-invasive tool, ultrasound waves can be applied to probe gaseous content of the drilling fluid in offshore oil-drilling operations. The approach is believed to improve sensitivity and accuracy of a gas-kick detection system. In this research, four types of bubble flow are designed to simulate undeveloped gas kicks, and their effects on changes of ultrasound waves are investigated. The bubbles are found to have changed power distribution of the sound waves that have been reflected by the bubbles and received by side sensors. The pattern of power spectrum changes around the master frequency is found to be closely related to the type of bubble flow. Such changes are grouped on the basis of cluster analysis, and it is found that bubble strings and bubble groups would produce substantially different effects and that bubble mergences would largely alter spectral property of the sound waves. By establishing relationship between power-change pattern of sound waves and the behavior of a bubble flow, the research is intended to seek a more predictive way of recognizing early-stage gas kicks for offshore oil-drilling practices.