Experimental evaluation of software estimates of task specific measurement uncertainty for CMMs

Estimation of the uncertainty of Coordinate Measuring Machine (CMM) measurements for real, imperfect parts is a very complex undertaking. Not only are there many contributors to the uncertainty that may interact in a non-linear fashion, making it difficult to mathematically determine an uncertainty estimate, but it is also difficult to predict the long-term variation of these parameters. Our work seeks to provide experimental validation of the uncertainties predicted by two different commercial software packages that purport to predict the task-specific measuring uncertainty of CMM measurement results. The validation procedure uses repeated measurements of calibrated artifacts to experimentally determine measurement uncertainties. These measurements can then be simulated in the commercial software packages. The comparison will allow the software to be tested to see if it appropriately accounts for the influences of the operator, environment, and part placement. This paper reports the results of actual part measurements and the predicted uncertainty provided by commercial simulation packages. Differences between experimental and simulated uncertainties are highlighted, and their causes examined.