Automated Horizontal Slurry Flow Regime Recognition Using Statistical Analysis of the ERT Signal

Abstract

Flow regime recognition is not only useful for characterisation of the flow, but also for the purpose of modeling, system controls and optimization and correction of flow regime dependent flow meters. This paper proposes a new indirect method for on-line recognition of the active horizontal slurry flow regime using statistical signal analysis of measurements obtained with a high performance Electrical Resistance Tomography system (ERT). Significant features of the ERT signal are extracted from both time domain and frequency domain. A set of experiments were carried out using a pilot-scale slurry flow loop, through which a mixture of sand and tap water was pumped into a 50mm inner-diameter test section. All common slurry flow regimes are considered in the recognition scheme, including the transitional regime boundaries covering the transport velocity range of 1.5-5 m/s. Two types of sand are used in the experiments, medium (75-900μm) and coarse (150-2200μm), each with different throughput volumetric concentration, 2% and 10%. 1.2 m transparent pipe section was included into the test section, so as to visually inspect the prevailing flow regime and capture photographic images of the flow. A code was developed not only to render the recognition of the active flow regime, but also to visualise the distribution of the solid particles across the pipe cross-section and display the mean solids volume fraction. The evaluation of the proposed recognition method suggests 90.32% successful rate.