Machine Learning approaches for Anomaly Detection in Multiphase Flow Meters

Abstract

Abstract Multiphase Flow Meters (MPFM) are important metering tools in the oil and gas industry. A MPFM provides real-time measurements of gas, oil and water flows of a well without the need to separate the phases, a time-consuming procedure that has been classically adopted in the industry. Evaluating the composition of the flow is fundamental for the well management and productivity prediction; therefore, procedures for measuring quality assessment are of crucial importance. In this work we propose an Anomaly Detection approach to MPFM that is effectively able to hand the complexity and variability associated with MPFM data. The proposed approach is designed for embedded implementation and it exploits unsupervised Anomaly Detection approaches like Cluster Based Local Outlier Factor and Isolation Forest.