Prediction of liquid - solid wetting efficiency in trickle flow reactors

A comprehensive database of catalyst wetting efficiency measurements in trickle flow regime was gathered from 14 independent studies to conduct a thorough evaluation of the performances of current correlations for the prediction of wetting efficiency during vertical gas-liquid co-current down-flow in randomly packed fixed bed reactors. Cross-examined with the database, several shortcomings arising from a low level of accuracy or a lack in generalization revealed the weakness of existing estimation methods. An approach relying on the combination of artificial neural network computing and dimensional analysis (ANNDA approach) helped to derive a highly accurate correlation for wetting efficiency in trickle flow regime. This correlation yielded an absolute average relative error of 8% and a standard deviation of 10%. The five dimensionless groups intervening in the proposed correlation were a composite two-phase low Reynolds, Reg, liquid Stokes (St), Froude (Fr) and Galileo (Ga) groups and a bed correction factor (Sb).