|
|
4.3 Easy and Not-So-Easy Methods of Uncertainty Analysis
Carroll Croarkin
Mark Levenson
Jack Wang Statistical Engineering Division, ITL
Theodore Doiron Precision Engineering Division, MEL
John Wehrmeyer Eastman Kodak Concepts of uncertainty analysis for metrology were outlined in a two-day workshop presented in conjunction with the Measurement Science Conference in Pasadena, CA on Feb. 2-3, 1998. Croarkin began the workshop with a discussion of the ISO policy on uncertainty and a case study that illustrated the the estimation of type A components of uncertainty from multiple sources, expanded uncertainties and effective degrees of freedom. Doiron discussed his experience and reliance on check standards for computing uncertainties for dimensional measurements. Wehrmeyer provided an industrial persepective with suggestions for how 'easier' assessments of uncertainty can sometimes be inferred from guard-banding techniques and other sources. Wang illustrated how an uncertainty assessment can be made from an inter-laboratory study, in this case, for measurements on 10 volt zener standards. In his first talk, Levenson outlined a method for computing uncertainties of results from a linear calibration line, and in a second talk, he showed a case study of how formal propagation or error was used to compute uncertainties for radiometry measurements. Croarkin concluded with a discussion of pros and cons of propagation of error and an uncertainty budget that combined results from type A evaluations with type B evaluations with propagation of error.
Date created: 7/20/2001 |