Mass transfer in osmotic dehydration of food products: comparison between mathematical models

Osmotic dehydration (OD) is a method to partially reduce water of fruits, vegetables, meat or fish aiming to increase the shelf life or as a pre-treatment in the processing of dehydrated foods. The aim of this work is to present the models most used in mathematical modelling of experimental data obtained from the OD processes, correlating the water loss and solid gain of the food product with process variables, and to perform a comparative and critical analyses of the different models and ability to fit data. The osmotic solution concentration, the temperature, the level of agitation and the geometry of the product are some of the operating parameters that will be mainly focused. Azuara’s, Peleg’s, Page’s, the Penetration, Magee’s, Weibull’s, Toupin et al.’s, Marcotte et al.’s, the Hydrodynamic mechanism, Spiazzi and Mascheroni’s, Seguı´ et al.’s, Crank’s, Hough et al.’s were the models approached. These were classified into empirical and semi-empirical, phenomenological and mechanistic, and the advantages and disadvantages were presented. An extensive list of applications of the different models to the osmotic dehydration, in variable ranges of operating conditions, of fruits, vegetables, meat and fish is provided in this work. Furthermore, equivalences between parameters of different models were established, based on the affinity of the functions used in the equations of the models, these equivalences allowing a better understanding of the adequacy of the different models to fit the same experimental data. A decision tree is provided in order to allow the selection of the most adequate model(s) to fit and predict experimental data from OD processes. All this information could assist and be helpful to researchers in the choice of the most adequate model(s) to fit experimental data, as well as to predict the water loss and solid gain of food products during OD processes.This work was supported by National Funds from FCT through Project PEst-OE/EQB/LA0016/2013. The first author acknowledges the financial support of CAPES (1528/13-0)