Explain Modelling for process and recipe

Generally, all problems that appear in food product design can be divided into mixture or process problems, with the latter having the dominant share. Sometimes a problem that seems to be a mixture problem is really a process problem and can only be solved with a corresponding factorial experimental method. As explained above, the difference between a process and a mixture study is quite distinct, and these studies need different statistical experimental techniques to deal with. In practice, it is not easy to distinguish a process problem with a mixture problem, when the food product design is only concerned with recipe or formulation development. To get a better understanding of the difference between them, a short description of performing a factorial experiment for solving a process problem and of running a mixture experiment is given:

1.  A factorial experiment: It studies the effect of some independent variables on food quality indices (response) through varying two or more of these independent variables, such as temperature, time, pressure and pH value. A series of values or test levels of each factor is selected, and certain combinations of their levels are tested.

2.  A mixture experiment: An experiment in which the food quality indices (response) are assumed to depend only on the relative proportions of the ingredient components present in the mixture and not on the amount of the mixture. In such an experiment, if the total amount of the mixture is held constant, the value of the response changes when changes are made in the relative  proportions of the ingredients.

The development of bakery powder is described as a practical example that will help you in understanding the difference between a factorial and a mixture experiment.
 
A premixed bakery powder for biscuit making consists of wheat flour  F and three different chemical compounds A, B and C, which would be tested  in the biscuit making according to a standard bakery experiment. The flour is used as a diluting medium, whereas A, B and C will be effective at different baking temperatures or baking phases. To develop an optimal baking powder formulation from F, A, B and C, the effect of various formulations are tested. Three different statistical experimental approaches are applied.