PREDICTION OF WATER ACTIVITY IN FOOD SYSTEMS - A COMPUTER-PROGRAM FORPREDICTING WATER ACTIVITY IN MULTICOMPONENT FOODS

The prediction of water activity (a(w)) by mathematical models, in gen eral, has been limited to aqueous solutions of electrolyte and non-ele ctrolyte solutes. The effect of food components, such as fiber, protei ns and fats, on a(w) has not been taken into account when those models are applied to real foods. The estimated a(w) on food systems is grea ter than the actual value reported because of water absorbed by protei ns or bound into fat mesomorphic structures. An interactive computer p rogram developed to accurately predict the a(w) of multicomponent syst ems is able to perform the following tasks: (1) Predict the a(w) of fo od model systems by several well accepted equations, such as Bromley, Pitzer, Norrish, Ferro-Chirife-Boquet, Ross, and Ferro-Benmergui-Chiri fe. (2) Establish similarity levels berween foods of known a(w) and co mposition to new food products where composition is available. (3) Cal culate deviation factors based on previously established similarity le vels, where the presence of fibers, fats, and proteins is taken into a ccount. The analysis of sausage-like products, in the first iteration, showed a deviation factor of 1.07, while sugar confections with gums and pectins had a deviation factor of 1.09. Once a supporting data ban k is created, the deviation factor rapidly converges to one.