Expressing uncertainty using regression and GUM for derived property measurements based on uncertainty in input quantities

In order to validate Integrated Computational Materials Engineering (ICME) thermodynamic property models for high-performance alloy design and advanced manufacturing techniques such as additive manufacturing optimization, it is necessary to quantify the uncertainty during measurement of thermophysical properties of metals and alloys. However, many values reported for these properties in the literature rarely include rigorous uncertainty quantification. The current standard is to report the experimental mean and standard deviation from tabulated experimental data or to plot test results with limited discussion on how error bars are obtained. But this approach does not reveal anything about the underlying systematic or random error propagation. This paper discusses the utility of using a systematic precision-based facility performance evaluation procedure to quantify variability during experimental measurement based on principles developed as part of the Joint Committee for Guides on Metrology (JCGM) Guide to the Expression of Uncertainty Measurement (GUM) recommended protocols. Graphical presentation of results is accomplished using a Measurement Accuracy and Precision (MAP) plot where measurement trueness can be evaluated relative to recognized reference value while precision can be assessed by quantification of GUM uncertainties. As an example, the performance of multiple levitation facilities has been successfully compared using these methods for a variety of metallic elements and industrial alloy systems.