Carl Ehrett, School of Mathematical and Statistical Sciences, Clemson University; D. Andrew Brown, School of Mathematical and Statistical Sciences, Clemson University; Christopher Kitchens, Department of Chemical and Biomolecular Engineering, Clemson University; Evan Chodora, Department of Civil Engineering, Clemson University; Roland Platz, Department of Architectural Engineering, Pennsylvania State University; Leslie Xu, Department of Architectural Engineering, Pennsylvania State University; and Sez Atamturktur, Department of Architectural Engineering, Pennsylvania State University;
In traditional engineering design, material selection involves choosing a material with appropriate properties from a database of known materials, often as a matter of ad-hoc satisficing. Material design usually occurs separately, and without an eye to specific end-uses. We wed these design processes, selecting a material design by modeling its performance outcomes in a particular engineering application. We show that existing techniques for model calibration can be profitably reconceptualized as a method for optimization and applied to solve this material design problem. We demonstrate by calibrating material design parameters to desired performance targets for a wind turbine blade.
Keywords: Uncertainty quantification, Gaussian processes, optimization, Pareto optimality, wind turbines, material design