Rapid Prediction of Flow and Concentration Fields in Solid-Liquid Suspension of Slurry Electrolysis Tank

Slurry electrolysis (SE), as a hydrometallurgical process, has the characteristic of multi-tank series connection, which leads to various stirring conditions and complex solid suspension state. For computational fluid dynamics (CFD) require highly computing resources, it was proposed to use a combination of CFD and machine learning to construct a rapid prediction model for the liquid flow and solid concentration fields in the SE tank. By scientifical selecting calculation samples through orthogonal experiments, the comprehensive dataset covering a wide range of conditions was established while effectively reducing simulation number and providing reasonable weights for each factor. Then a prediction model of the SE tank was constructed using the KNN algorithm. The results shown that with an increase of levels in the orthogonal experiments, the prediction accuracy of the model significantly improves. Among them, the model established with 4 factors and 9 levels can accurately predict the flow and concentration fields, for the regression coefficient of average velocity and solid concentration can achieve 0.926 and 0.937, respectively. Compared with the traditional CFD, the response time of fields information predicting was reduced from 75 hours to 20 seconds, which solves the problem of serious lag in CFD applied to actual production, and meets the real-time production control requirements.