Relating Margin to Engineering Performance

• Activity Number: 331
• Type: Invited
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Rene Lynn Bierbaum*+
• Companies: Sandia National Laboratories
• Keywords: QMU ; Margin ; One-shot item ; Surveillance

Abstract

The basic ways in which Quantification of Margin and Uncertainty (QMU) is related to engineering performance will be discussed. QMU is similar to other testing and simulation activities where the intent is to use the results to make statements about engineering performance. Experience has shown that there are criteria that should be considered prior to performing a QMU analysis. This checklist includes questions such as: Are the data to be used in a QMU analysis being taken on units that are sufficiently representative of those in the field? Will the testing fidelity be adequate? Given sampling variation, what is the confidence that we have in the inference we make about the proportion of units outside of the performance requirement? How does the parameter that will be analyzed with QMU relate to engineering performance? In addition to this, estimation of performance impacts from QMU poses a surprisingly complex challenge for any parameter because it requires a good understanding of the "performance requirement". This checklist is helpful in ensuring that full benefit is obtained from QMU and that the results can be related to performance. [SAND2012-9391A]