Evaluation of Two Noniterative Electrical Resistance Tomography (ERT) Reconstruction Algorithms for Air-Core Measurements in Hydrocyclone

Air-core dynamics during the hydrocyclone operation play an essential role in determining the separation efficiency. Precise monitoring of air-core dynamics during operation can suggest possible strategies to prevent underflow discharge problems. Electrical resistance tomography (ERT) facilitates online analysis of the internal behavior of air-core, because of its high temporal resolution. Existing results for ERT-based air-core reconstructions in hydrocyclones employ general algorithms that do not utilize process knowledge. In this paper, two noniterative methods, viz., monotonicity and factorization algorithms, which are process-aware, are evaluated to estimate air-core diameters in hydrocyclone under feed pressure variations. To further investigate the performance of these algorithms, comparison studies are made with Gauss-Newton (GN) and Total Variation (TV) algorithms. Initially, the algorithms are compared on simulated phantoms for qualitative analysis and to select hyper-parameters. A threshold method is developed for quantitative analysis of experimental data to obtain a crisp radius value. The ground truth is calculated from image processing on images obtained from a video camera. It is observed that GN, monotonicity, and factorization methods result in a similar performance on the experimental data. However, for very fine meshes, monotonicity and factorization algorithms are much faster than GN and TV methods.

Automated summary

This study evaluates the performance of two noniterative methods, monotonicity and factorization algorithms, in estimating air-core diameters in hydrocyclones. The results show that these two methods perform similarly on the experimental data and are faster than GN and TV methods.