Thresholding Values and Fuzzy Logic Fusion in Visualisation of Gas-Oil-Water Horizontal Flow using Dual-Modality Electrical Tomography

Two different data fusion approaches, namely the thresholding and fuzzy logic-based data fusion, have been proposed to address the challenges in the use of dual-modality electrical tomography (ERT-ECT) for visualising gas-oil-water horizontal flow. This paper introduces both methods and compares their performance on visualisation of six common flow regimes in horizontal pipe flow of industrial-scale multiphase flow testing facility, including stratified flow, wavy stratified flow, slug flow, plug flow, bubbly flow, and annular flow. The paper reports experimental results which reveals specific advantages and limitations of each approaches. Overall, the threshold-based approach, including data fusion and phase decomposition, is simple at the aspects of understanding and implementation. It is fast and requires little computing resources, and therefore suitable for online purpose. The selection of thresholding values, however, have obvious impacts on the results since the determination of the thresholding values requires prior knowledge, which could be affected by site and operation factors. In contrast, fuzzy logic-based method is much more complicated than the thresholding method since it utilises the fuzzy inference system and decision tree to perform the fusion and carry out the decomposition respectively. Accordingly, this advanced method costs more time and computing resources to derive the final results. However, it does not need prior knowledge on threshold values and hence is insensitive to the changes of membership functions, which makes this method more robust than the thresholding value method. In addition, the results from visualisation of industrial flows demonstrated the fuzzy logic fusion method has a better performance on reporting phase boundaries.