Modeling Thin-Layer Drying of Turnip Slices Under Semi-Industrial Continuous Band Dryer

Abstract

This study aims at investigating the thin-layer drying behavior of turnip slices in a multistage semi-industrial continuous band dryer. Turnip slices with the thickness of 4 mm were used for the drying experiments. The experiments were conducted at three air temperatures of 45, 60, and 75C, three air velocities of 1, 1.5, and 2 m/s, and three belt linear speeds of 2.5, 6.5, and 10.5 mm/s with three replications for each treatment. To estimate the drying kinetic of turnip slices, six mathematical models were used to fit the experimental data of thin layer drying. Consequently, the Midilli et al. model was selected as the best mathematical model to describe the drying kinetics of the turnip slices. The effective moisture diffusivity varied from 8.37 × 10−10 to 4.82 × 10−9 m2/s. The energy of activation varied from 12.80 to 26.31 kJ/mol using Arrhenius type equation. After well training of the ANN models, proved that the ANN model was relatively better than the empirical models. The best neural network for the prediction of moisture ratio (MR) and drying rate was feed forward back-propagation with 4-10-10-2 structure, training algorithm of Bayesian regulation and threshold functions of tansig-purelin-logsig. The best R2 value for predication of MR and drying rate were 0.9990 and 0.9619, respectively. Practical Applications Turnip is one of the oldest cultivated vegetables that have been used for human expenditures since prehistoric times. Drying process has been used for decades in food processing industries for efficient long-term preservation of final products. It has the potential to deliver safe food products through enzyme inactivation and microbe destruction. Advantages of semi-industrial continuous band drying are including short processing time, high mass transfer coefficients, low energy consumption, and high quality. To model experimental drying data, many correlations that are available in literature may be used. However, artificial neural network (ANN) methodology has become increasingly popular recently because of its capability of giving more general and precise results as also presented in this study.