Electromagnetic Tomography for Multiphase Flow in the Downhole Annulus

Downhole annulus multiphase fluid monitoring during drilling is technically tricky, and the existing downhole overflow monitoring means the specific overflow type and state cannot be determined. For the monitoring of downhole multiphase fluids, this article proposes the downhole annulus electromagnetic tomography (AEMT) detection technology and focuses on the extreme learning machine (ELM) algorithm to realize downhole annulus image reconstruction. For the special characteristics of electromagnetic tomography (EMT) in downhole annulus, this article first establishes a theoretical model of AEMT mathematics and proposes the ELM image reconstruction algorithm for downhole annulus fluids. Second, the 3-D simulation model of the AEMT is established. The 3-D simulation model of the AEMT random phase distribution multiphase flow measurement is constructed by co-simulation of MATLAB with COMSOL, and 12000 groups of simulation data samples are established. Finally, this article builds an annulus fluid multiphase flow test platform for static experimental comparison and analysis. The results show that the average correlation coefficients of the test set of the downhole ELM image reconstruction algorithm are increased by 139.27%, 138.65%, and 725.15%, and the average relative errors are reduced by 66.5%. The average relative errors are reduced by 66.5%, and the average relative errors are decreased by 66%. The average relative error decreased by 66.34%, 68.41%, and 81.18% compared with the three classical algorithms, local binary pattern (LBP), Landweber, and Tikhonov. This article realizes the EMT based on the ELM algorithm for the first time in downhole annulus flow channel, which provides the theoretical basis for the future real-time image monitoring of downhole annulus multiphase fluids.