Mixture Experiment Approaches: Examples, Discussion, and Recommendations

A mixture experiment involves varying the proportions of two or more ingredients, called components of the mixture, and studying the changes that occur in the measured properties (responses) of the resulting end products. Other factors that affect the response(s) such as process variables or the total amount of the mixture may also be studied in the experiment. Five approaches for designing mixture experiments and analyzing the data are considered: component proportion, mixture-amount, mixture-process variable, mathematically independent variables, and slack variable. The approaches are illustrated using examples from the literature. The various approaches are discussed and recommendations are made for choosing among the approaches when faced with a mixture experiment problem. A five-component waste grout example for which several mixture experiment approaches are possible is discussed.