Evaluation of the Siddiqi-Lucas Correlation and Its Modification to Estimate of the Alcoholic Solutions Diffusivity

Because of the importance of diffusion coefficient and its high application in mass transfer calculations, many correlations have been proposed in order to predict it in various systems. In this paper, one of the most simple, general and applicable correlation to estimate diffusivity in liquid systems, Siddiqi-Lucas correlation, selected and its accuracy and capability in calculation of diffusivity of various liquid systems have been investigated. Results indicate that average absolute errors of this correlation in diffusivity prediction are 25.1% and 20.8% in hydrocarbon mixtures and aqueous solutions, respectively. Also, the calculations show that average absolute error for alcoholic solutions is 44.8% and more than 30% of calculated diffusion coefficients in these systems have an error of more than 60% while the in non-alcohols solutions average error is 18.9% and less than 2% of results have an error of more than 60%. Because of the weakness of this correlation in the prediction of diffusion coefficients in systems including alcoholic solvents, and the results of previous studies, literatures which depict higher prediction errors using most relations for alcoholic solutions with respect to non-alcoholic solutions, and since, there is not a particular relation for these systems, a new correlation in the form of Siddiqi-Lucas relation proposed to calculate diffusivity in alcoholic systems. Using this relation, the average error decreased from 44.8% to 28.5% and the results percentage with error more than 70% decreased from 24 to zero. For more verification, the results of proposed relation compared with Wilk-Chang results, Tyn-Calus and Hayduk-Minhas correlations, and the results indicated a good improvement in calculated coefficients and the higher accuracy of the proposed relation which makes its usage reasonable and necessary.