Evaluating a High-Dimensional Constrained Mixture Experiment Design with Replacement Points

• Activity Number: 321
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Scott K. Cooley* and Greg F. Piepel and John D. Vienna and Michael J. Schweiger
• Companies: Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory
• Keywords: mixture experiment ; optimal experimental design ; nuclear waste glass

Abstract

Methods used to evaluate two related mixture experiment designs for a 14-component nuclear waste glass study are presented and illustrated. An initial design was generated to provide a statistically optimal set of boundary points for a design space specified by single-component and multiple-component constraints. The specified design space represented an expansion of glass composition regions explored in previous studies by allowing higher proportions of certain components. Several glass compositions from the initial design did not result in acceptable glasses during testing. Each unacceptable composition was modified by glass scientists based on subject matter knowledge, to obtain acceptable, replacement glass compositions. This modification process typically required several iterations of adjusting the composition and attempting to form glass from the adjusted composition. The compositional adjustments generally focused on a few key components. The final design, containing the replacement compositions, and the initial optimal design were then compared graphically and using several statistical measures of design performance.