Evaluating Inter-laboratory Comparison Data

• Abstract:

  The primary purpose of inter-laboratory comparisons is to demonstrate that the uncertainty specifications of the calibration measurement capabilities of the participating laboratories are correct. The most common criterion for assessing a participating laboratory’s results is whether the normalized error |Eni| is ≤ 1. Most comparison reports we reviewed properly include uncertainty components related to the transfer standard (uTs ) and the repeatability of the calibrations (urepeati ) in the uncertainty of the value reported by a participant. Unfortunately, high values for either uTs and urepeati decrease |Eni| , making it easier to achieve passing results in a comparison that uses a poor transfer standard or for a participant that delivers unstable measurements. A review of past comparison reports shows that this problem occurs for many measurands, including flow, temperature, and pressure. Improved comparison criteria were proposed by [1] to counteract the flaws of the |Eni| ≤ 1 criterion by introducing the possibility of inconclusive results and a probability-based approach. In this paper, we define comparison uncertainty ucomp as the root-sum-of-squares of uTs and urepeati and find it a better tool for assessing the power of the comparison than uTs alone. We applied the comparison evaluation criteria to recent comparison results to illustrate their benefits over the |Eni|≤ 1 criterion. In general, the newer criteria confirm prior determinations, but in some cases passing results for the |Eni|≤ 1 criterion would be found inconclusive.
• Keywords:

  Key comparison; Evaluation criteria; Inconclusive results
• Download:

  IMEKO-TC9-2019-022.pdf
• DOI:

  tc9-2022.022

Event details: