Electrical resistance tomography for process intensification of gas–liquid stirred tanks: A comparison of linear back projection and sensitivity conjugate gradient algorithms

Abstract Electrical resistance tomography (ERT) is a non‐invasive technique widely applied to the study of multiphase flows, where phases exhibit different electrical properties. By injecting currents and measuring boundary voltages, ERT reconstructs the local conductivity distribution, from which dispersed phase hold‐up can be estimated. The inverse problem underlying ERT is ill‐posed, requiring the use of reconstruction algorithms. The most common approach, linear back projection (LBP), provides rapid, real‐time reconstruction but is limited to qualitative imaging due to its low resolution. Today, there is a growing demand for quantitative data to provide validation metrics, which are essential for tasks such as validating computational models hence to improve tools for process intensification. Iterative algorithms, such as the sensitivity conjugate gradient (SCG) method, offer more physically representative reconstruction and better capture of nonlinear behaviour, but require careful tuning of parameters and are computationally intensive. Surprisingly, however, there is a lack in the open literature of a systematic comparison of LBP versus SCG in stirred tank reactors under different hydrodynamic regimes. This work therefore presents a systematic comparison between LBP and SCG for the analysis of gas–liquid flows in a pilot‐scale baffled stirred tank equipped with a Rushton turbine and ring sparger. A gas–liquid stirred tank operating in different fluid‐dynamic regimes has been analyzed to test the reconstruction algorithms under both large and small gradients of dispersed‐phase hold‐up. Therefore, the selected equipment should be considered a benchmark that allows the reconstruction algorithms to be evaluated under different conditions.