A new approach to the optimization of composition and processing parameters for alloy development

The primary motivation of this research is to find ways to reduce time and cost associated with the development and selection of new metallic alloys having highly variable mechanical properties. Particularly, the fracture toughness response as measured by Charpy V-Notch (CVN) values exhibits substantial variability and cannot be modeled via standard regression with its focus on the mean. We use the latest achievements in generalized regression techniques for building linear regression models for right and left tails of the ln(CVN) distribution. We demonstrate how these regression models can be used to search for combinations of chemical compositions and processing parameters that result in steels with optimal tensile yield strength and CVN characteristics. According to our approach, information from prior experimental data is incorporated into mathematical models, the models are used for optimization, and results of optimization are subsequently used in refining the fabrication process. The simulations can be used successively at various stages of the experimental program.