(R) Use of Model Verification and Validation in Product Reliability and Confidence Assessments
|Publication Date:||1 February 2020|
This SAE standard outlines the steps and known accepted methodologies and standards for linking Model V&V with model-based product reliability assessments. The standard's main emphasis is that quantified values for model-based product reliability must be accompanied by a quantified confidence value if the users of the model wish to claim use of a "Verified and Validated" model, and if they wish to further link into business and investment decisions that are informed by quantitative second-order risk and benefit cost considerations.
The purpose of this standard is to provide a process for linking demonstrated and accepted standards for model V&V with peer reviewed and published methods and standards for model-based assessments of product reliability. The result will be a standard that enables a design process providing validated model-based assessments of product reliability at a quantified confidence level. This will justify and enhance the success of investments and business decisions that are based on model assessments of product reliability.
Applicability and Intended Use
The intended use of this standard is in design and assessment situations where the reliability of a physical product is being assessed using a combination of a numerical model and relevant physical test data. This is not the same as either (a) reliability assessment of a product via "go versus no-go" testing, or (b) reliability assessment of software through algorithmic computer simulation. In the situation for which this standard was written, there is usually limited test data; therefore, it is a fair consensus that, under such constraints, a numerical value for assessed reliability has little credibility without an accompanying numerical value for assessed confidence. It is the goal of this standard to provide the basic essential steps toward such an assessment: use of verified and validated models to quantitatively assess reliability at a specified quantitative confidence. To enable implementation, the standard references numerous published and peer-reviewed procedures that can meet the specifications of this standard.
This standard does not specify or mandate a particular numerical procedure to meet its requirements. Instead, this document provides a list (which may evolve in the future) of methods for both Model V&V and Product Reliability (at confidence) assessment that can be tailored to the risk and business considerations of a particular design or product.
A complete, but simple, procedure is given in Appendices A and B. Expanded procedures are discussed in 5.1.