Sandia's Uncertainty Calculator: A Case Study in the Graded Approach

Abstract

Software is used everywhere in a calibration lab, from controlling measurement equipment to performing statistical calculations to storing calibration records in a database. Yet the importance and details of software validation and quality assurance are not often discussed. A graded approach to software quality assurance (SQA) is typically used, where software with a higher risk of adverse consequences upon failure requires a higher level of rigor. The Sandia National Laboratories Uncertainty Calculator (Suncal) is open-source, free software for performing various statistical calculations related to metrology, including Guide to the Expression of Uncertainty in Measurement (GUM) and Monte Carlo uncertainty propagation, decision risk analysis, uncertainty in curve fitting, and analysis of variance. The SQA activities for the Suncal evolved over time. In its infancy, Suncal was simply a few code functions shared among a handful of coworkers, and SQA activities were almost nonexistent. As Suncal grew into a more formal library of algorithms with a full-featured user interface, and especially after it was released to the public, the SQA rigor level increased to match the tool’s increased risk and visibility. Suncal, and the SQA activities to ensure it operates correctly are discussed below.