Examining the efficiency of the stirred structure packed extraction column with hands-on investigation using simulation incorporating artificial neural networks

Liquid-liquid extraction (LLX) is a fundamental application of separation processes utilized in various industries. Extraction columns are essential equipment in LLX operations. Designing and scaling up extraction columns necessitates a comprehensive understanding of hydrodynamic parameters and mass transfer performance. Therefore, this study investigates the mass transfer characteristics of Scheibel column. A total of 42 experimental data points were collected to explore the impact of operating conditions on the overall mass transfer coefficient. The results indicated that rotor speed (N) significantly affects K_od, while the flowrates of the continuous and dispersed phases have a minimal impact. Subsequently, K_od was determined using previous models, but none of them accurately predicted K_od. As a result, a new correlation was proposed to estimate K_od as function of the Reynolds (Re) and Weber (We) numbers, and the holdup of dispersed phase (x_d). This new correlation demonstrated excellent agreement with experimental data, making it a valuable tool for designing Scheibel extraction columns. Lastly, four different ANN models were applied to forecast K_od using rotor speed (N), continuous phase velocity (V_c), dispersed phase velocity (V_d), and diffusion coefficient (D_d) as input parameters. MLPNN, RBFNN, and CFFNN models proved to be accurate and reliable in predicting K_od.