INTRODUCTION OF A ROBUSTNESS COEFFICIENT IN OPTIMIZATION PROCEDURES - IMPLEMENTATION IN MIXTURE DESIGN-PROBLEMS .3. VALIDATION AND COMPARISON WITH COMPETING CRITERIA

Abstract De Boer, J.H., Smilde, A.K. and Doornbos, D.A., 1992. Introduction of a robustness coefficient in optimization procedures: implementation in mixture design problems. Part III: Validation and comparison with competing criteria. Chemometrics and Intelligent Laboratory Systems , 15: 13–28. The robustness coefficient (RC) introduced in Part I is an optimization criterion measuring the robustness or ruggedness of a product or process. The RC has been developed for mixture problems, especially for situations where the independent (mixture) variables have a varying variance/ covariance structure. In this article the RC is validated and compared with two other robustness criteria using simulation experiments. Validation is performed by comparing the three criteria with simulated values while varying a number of potentially influential factors. The sensitivity of the three robustness criteria to repeated sampling is also investigated.