Water sorption isotherms of freeze-dried milk products: applicability of linear and non-linear regression analysis in modelling

Water sorption of various freeze-dried milk products was modelled using several water sorption isotherm models, most of which proved to be applicable. Sorption models were fitted to experimental data using linear and non-linear regression analysis. Both methods gave almost the same prediction of water sorption when the model had a good fit. The GAB model was considered to be the most applicable in predicting water sorption in practical applications, as the use of one universal model is desirable. Time-dependent changes, e.g. lactose crystallization above glass transition, were taken into account in the water sorption modelling. Water contents and relative humidities which allowed changes in physico-chemical properties were not included in the modelling, because of unsteady amounts of sorbed water.