Modeling of adsorption isotherm of freeze‐dried mahaleb powder using artificial neural network and calculation of thermodynamic sorption properties

Abstract

AbstractIn this study, adsorption isotherms of freeze‐dried mahaleb powder were obtained using gravimetric method at four different temperatures (5, 25, 35, and 60°C). Experimental adsorption data were modeled using empirical models and artificial neural network. The GAB model gave the best fit among the empirical models with high R2 (0.9908–0.9982) and low P% (4.71–8.40%) values. However, artificial neural network (ANN) model having 10 hidden neurons optimized using Lavenberg‐Marquardth training algorithm was superior to the empirical models with the highest R2 (0.9965–0.9999) and lowest P% (0.74–8.52%) values. The ANN model explained the adsorption behavior of mahaleb powder ideally at all temperatures. In addition to the adsorption kinetics of mahaleb powder, thermodynamic properties such as isosteric heat of sorption, differential entropy and Gibbs free energy were also determined. Isosteric heat of sorption values were high when moisture contents of the powders were low. Also, differential entropy values followed the same trend which indicated that the adsorption process was driven by enthalpy. Gibbs free energy at the isokinetic temperature of 458.48 K was determined as −306.7 J/mol. Gibbs free energy values were below zero at all different temperatures. The results showed that ANN model can be suitable for predictive control systems in the food industry and the data obtained for thermodynamic properties can be useful for the storage of mahaleb powder in industrial scale.Practical ApplicationsAn intelligent model, which is artificial neural network was used for modeling of adsorption behavior of freeze‐dried mahaleb powder and it predicted the adsorption data better than empirical models. Thermodynamic properties which are mostly important for the prediction of energy consumption of a process were also determined. The data of the freeze‐dried mahaleb powder can be significant for the food industry in terms of prediction of shelf‐life and energy requirements.