Mixed-level supersaturated design application to a robustness study on an organic synthesis

作者：Croguennoc Alice; Frederic Marc; Claeys Bruno Magalie^*; Roger Phan Tan Luu; Sergent Michelle

来源：Chemometrics and Intelligent Laboratory Systems, 2012, 113: 19-25.

DOI：10.1016/j.chemolab.2012.01.005

• Mixed-level supersaturated designs
• Screening
• Process robustness
• D-optimal design

Mixed-level supersaturated designs are designs in which the number of coefficients to estimate is greater than the number of experiments. This type of design is useful in the rapid preliminary investigation of a process with a large number of potentially relevant factors but with only a few of them having important effects. The purpose of this project was to determine the active parameters on a chemical process. A mixed-level supersaturated design of 12 experiments was carried out, allowing us to screen 17 two-level factors and 5 three-level factors. chi(2)-optimality of the design was confirmed. In order to verify the results, a 40-run matrix was built. One parameter was identified as very active in both analyses, and several others were suspected to have an effect on the process. Some differences were observed in the results, regarding the detection of the least influent factors. These are likely due to the loss of information generated by the reduction of runs in the supersaturated design, as this kind of design allows an important reduction of the number of experiments. Further studies should be considered to confirm the activity of some factors.

• 出版日期2012-4-15