On-line flow regime identification of severe slugging in multiphase liquid and gas pipelines

An improved technique is presented for the flow regime identification in pipelines-rise system simulating offshore oil pipelines. The technique is based on signal analysis to classify flow pattern and to extract key feature parameters from pressure fluctuation at riser base in different flow regime. Through the analysis of these characteristic parameters, the flow regulation is obtained and the flow regime is successfully distinguished and identified using least square support vector machine artificial intelligence technology. On-line experimental results demonstrate that the proposed technique can recognize severe slugging in pipelines and has reached better recognition accuracy which is about 96.4%.