Measurement Uncertainty Evaluation Involving Significant Uncorrected Systematic Errors

摘要

In the field of Metrology, it is often necessary to evaluate measurement uncertainty in the presence of significant uncorrected systematic errors. This typically occurs with multifunction and/or multipoint measurement systems, such as digital multimeters and electronic balances, to name two. It is generally impractical to assign corrections for the many functions and ranges. As a result, these types of systems are usually evaluated with respect to a nominal value and assigned a bilateral confidence interval. In the strictest sense, evaluations of that sort are not in harmony with the Guide to the Expression of Uncertainty in Measurement [1], (hereafter referred to as the "GUM") in the event of remaining uncorrected systematic errors of a significant magnitude relative to the performance specification assigned to the system. According to the GUM, those residual systematic errors must be corrected, in order to legitimately consider the remaining combined random effects as a standard or expanded uncertainty. A deeper look into the philosophy and intent of the GUM, however, reveals a methodology for combining significant uncorrected systematic errors with random effects and still legitimately consider the combination as a standard or expanded uncertainty, provided certain qualifications are made. This paper describes that GUM-consistent methodology, the necessary qualifications, and provides examples of application.