An efficient 3D SCG-based image reconstruction for pipeline multiphase flows

Electrical Impedance Tomography (EIT) has been used as a visualization and measurement tool in many fields such as medical imaging, geophysical prospecting and industrial processing applications. Many research groups are now turning their attention to the full three dimensional image reconstruction algorithms. Considerable challenges are associated with the software aspects of 3D EIT systems due to the requirement for accurate 3D forward problem modelling and the derivation of 3D image reconstruction algorithms. This work aims to provide an efficient 3D EIT reconstruction and imaging for pipeline multiphase flows based on the sensitivity theorem using conjugate gradients (SCG) methods. The paper presents a 3D SCG based inverse solution in conjunction with a novel 3D EIT dual-plane zigzag sensor. The sensing system was developed based on the main concepts of maximally including the longitudinal information to eliminate 3D effects and equally distributing the electric field around a relevant narrow sensing section of pipeline flow. The proposed 3D SCG solution reconstructs the conductivity distribution in a cylindrical 3D domain where a narrow section of the pipeline is focused. The algorithm assumes two sensing planes, with 16 electrodes per plane with a zigzag arrangement, mounted in a pipeline section whose length justifies the assumption that it is infinite in the forward model. Based on the specific features of pipeline flows and the geometrical structure of the new sensor, the modified algorithm utilise non- uniform 3D mesh elements along the longitudinal axis of tubing domain and the infinitive far definition for two sides of the domain. This method is convenient for dynamic 3D flow visualisation and velocimetry with cross- correlation method, which provides a balance between the requested imaging precision and computation speed. A comparison of the performance between the traditional 2D SCG algorithm and the modified 3D SCG algorithm to 3D effect are reported in this paper.