Transient modeling of extraction columns: Parameter estimation, uncertainty analysis, and operation optimization

Abstract Despite extensive modeling efforts in extraction research, transient column models are rarely applied in industry due to concerns regarding parameter identifiability and model reliability. To address this, we analyzed uncertainty propagation from estimated parameters in a previously introduced column model and assessed identifiability via ill‐conditioning analysis. Extrapolation was evaluated by estimating parameters from fluid dynamic experiments and validating with mass transfer data. Using a DN50 pulsed sieve tray column, parameters , and the swarm exponent were estimated through maximum likelihood, as they directly affect online‐measurable drop diameter and hold‐up . Incorporating these estimates, the model reproduced , , and stationary and transient concentration profiles with relative errors below 10%. For optimization, we applied the model to minimize solvent demand for 99% product recovery. Uncertainty analysis showed coefficients of variation below 4%, confirming model reliability.