Thin-layer drying of tea leaves: Mass transfer modeling using semi-empirical and intelligent models

Moisture content is a critical factor in quality and shelf-life of foods and agricultural products. This research dealt with prediction of moisture ratio of tea leaves using intelligent genetic algorithm-artificial neural networks (multilayer perceptron, MLP; and radial basis function, RBF) and semi-empirical models during different thin-layer drying processes (i.e. sun, air, hot air, and microwave drying). Effective diffusivities were found in the range of 7.5x10(-7) to 9457.2x10(-7) m(2)/h, which the highest D-eff value was achieved for microwave drying. Moisture ratio data were modeled using fourteen semi-empirical equations among which Henderson and Pabis, Henderson and Pabis-modified, two-term-modified and Wang and Singh models received highest correlation coefficients. However, the prediction efficiencies of MLP (MSE, NMSE and MAE of 0.0084, 0.0597 and 0.0722, respectively) and RBF (MSE, NMSE and MAE of 0.0043, 0.0973 and 0.0564, respectively) networks were found to be more competent than semi-empirical models and therefore could be applied successfully for predicting moisture ratio of tea leaves during different drying processes. (c) All Rights Reserved