Title:
Statistical methods for process parameter development of niobium alloy C-103

Thumbnail Image
Author(s)
Palacios, Daniel
Authors
Advisor(s)
Stebner, Aaron
Advisor(s)
Editor(s)
Associated Organization(s)
Series
Supplementary to
Abstract
The growth of the Additive manufacturing (AM) as a process has resulted in its appli- cation to various new materials. However, for each new material that is developed with AM, process parameter development must be conducted in order to be able to fabricate end-use parts. Given either extreme properties of certain materials, or a lack of existing literature, the development of process parameters can often be a slow, expensive process. This work demonstrates a method that can be used to determine optimal process parameters for a material by beginning with single laser tracks (with no deposition) and then moving up to a thin wall specimen. This method is developed in the Julia programming language and then applied in a case-study for the Niobium alloy C-103. Single laser tracks are used to determine bounds in the process space and then a sequential learning approach is used to sample across this space in order to determine optimized parameters for the density of the final part. The application of this method resulted in just over 50 specimens being created, resulting in the determination of multiple parameter sets that yielded parts with greater than 99% relative density, as well a process map that can be used to identify other regions of the process space that may be of interest. The resulting model was then analyzed using Shapley values to determine relationships between process parameters and density. Shapley value analysis found that the z-step and dwell time played no role in the model’s prediction for density. Build speed and the mass flow rate had a mean absolute contribution that was nearly double that of the laser power.
Sponsor
Date Issued
2023-08-28
Extent
Resource Type
Text
Resource Subtype
Thesis
Rights Statement
Rights URI