Severe Slugging Flow Identification from Topological Indicators

Abstract

In this work, a topological data analysis pipeline was used to identify the onset of severe slug flow in offshore petroleum production systems. Severe slugging is a multiphase flow regime known to be very inefficient and potentially harmful to process equipment. Data from a pressure sensor located in wells is utilized to obtain topological indicators capable of revealing the occurrence of severe slugging. Signal data were processed by means of Takens embedding to produce point clouds, analyzed by persistent homology. Topological methods based on persistence diagrams are shown to be useful in identifying severe slugging and in classifying different flow regimes from pressure signals of producing wells with supervised machine learning.