Sandia’s Uncertainty Calculator: A Case Study in the Graded Approach to Software Quality Assurance

Where practicable, the total end-to-end test-and-calibration program cost would serve as the ultimate measurement quality metric (MQM). Total cost includes both the capitalization and ongoing costs that support product quality (sometimes called cost of quality) and the consequence costs (sometimes called cost of poor quality) that result from imperfect measurement and products. End-to-end means capturing costs from the entire traceability chain: from measurement standards to end products. Minimizing this MQM, total end-toend cost (TETEC), equates to optimizing measurement quality assurance (MQA). Lacking easily available measurement and performance data automatically fed to modeling software, organizations have found cost metrics unimaginable or impracticable, so their measurement programs instead target more easily computed MQMs, such as false-accept risk or simpler proxies thereof, setting minimum, but sub-optimal, quality levels. However, modern computing systems and software, such as laboratory management systems with testpoint- level traceability, rapidly approach the point at which the TETEC MQM will become practicable. Preparing for this eventuality, the NCSLI 173 Metrology Practices Committee has developed models that relate costs to measurement program information such as product specifications, test and measurement uncertainties, calibration intervals and reliability targets. Applications include optimizing overall program MQA, but also estimating the value of metrology and return on equipment investments, selecting instruments, designing test and calibration processes, designing products. This paper applies the cost models to case studies and examples to illustrate some applications.