A process to evaluate commercial software packages that estimate measurement uncertainty through simulation

Coordinate Measuring Machines (CMMs) can be used to measure a great variety of parts. Essentially, any three-dimensional surface can be evaluated. Characterizing the measurement uncertainty in a way that maintains traceability in accordance with international standards can be difficult since there are many variables that influence a measurement, such as geometry type, nominal scale, fixturing, orientation, number of measuring points, and probe configuration. Task-specific uncertainty is measurement uncertainty related to a particular measurement scenario. The idea of using statistically- based simulations, in particular Monte-Carlo type simulations, to help determine these uncertainties, has been previously introduced as a practical method. The work in this dissertation describes the concepts and implementation of two commercially available software simulation packages, VCMM and PUNDIT. A process used for evaluating commercial software packages was implemented by way of comparing software estimates for uncertainty to those calculated from calibrated artifact measurement (substitution method) for a variety of measurands. This process indicates directly how well each software performs for the cases tested, but also provides a framework for future verification processes. The results indicate that realistic uncertainty estimates can be attained by simulation, in the context of the experimental conditions, though a broader scope for the simulations, inclusive of multiple conditions for each measurement task, gives improved results