Moisture Sorption Isotherms and Isotherm Model Performance Evaluation for Food and Agricultural Products

Abstract

Moisture sorption characteristics of agricultural and food products play important roles in such technological processes as drying, handling, packaging, storage, mixing, freeze-drying and other processes that require the prediction of food stability, shelf life, glass transition and estimation of drying time and texture and prevention of deteriorative reactions. They are useful in the computation of thermodynamic energies of moisture in the products. An understanding of moisture sorption phenomena in products, moisture sorption isotherm (MSI) determination techniques and moisture sorption isotherm model evaluation procedures would be useful in the development or selection, modeling and controlling as well as optimization of appropriate processes to make for enhanced efficiency. The phenomena addressed in this chapter are equilibrium moisture content (EMC)-water activity (aw) relationships and MSI types, temperature influence on isotherms and occurrence of moisture sorption hysteresis. MSI measurement techniques highlighted are the gravimetric, vapor pressure manometric (VPM), hygrometric and inverse gas chromatographic and the use of AquaLab equipment. Commonly used moisture sorption isotherm models (BET, GAB, modified GAB, Hailwood-Horrobin, modified Hailwood-Horrobin, modified Halsey, modified Henderson, modified Chung-Pfost and modified Oswin) were selected, and their evaluation procedures using moisture sorption data were outlined. Static gravimetric technique involving the use of saturated salt solution appears to be the most widely used and recommended method of determining the EMC of agricultural and food products. Most of the MSI models can be fitted to moisture sorption data thorough linearization by logarithmic transformation, while others can be solved using such expression as second-order polynomial. Model goodness of fit can be determined using standard (SE) error of estimate, coefficient of determination (R2), mean relative percentage deviation (P) and fraction explained variation (FEV). The acceptance of a model depends on the nature of its residual plots. A model is considered acceptable if the residual plots show uniform scatter around the horizontal value of zero showing no systemic tendency towards a clear pattern. A model is better than another model if it has lower SE, lower P, higher R2 and higher FEV. Although it appears as if a generalized MSI model is yet to exist, it is recommended that the Ngoddy-Bakker-Arkema (NBA) model should be given thorough going and extensive testing on the MSI of different categories of food as it could prove true to its generalized model posture due to the fundamental nature of its derivation.