Standardization for Numerical Simulations: a Mapping of Current Practice and Challenges in Swedish Industry (pro-ject SPRUCE)

Abstract

Project SPRUCE ("Investigation of Standardization Practices for a Responsible Use of Computational models in Engineering") investigated the role of standardization in quality assurance of numerical simulations for industrial applications. The technical literature on existing standards, analysis guidelines and validation methods has been reviewed and summarized for the purpose of identifying the areas where standardization has the major impact on engineering simulations. A survey among practitioners in Swedish industry was conducted and presented during an open workshop, resulting in a comprehensive mapping of current practices to assess the credibility of numerical simulations as an effective tool to support decision-making. The main outcomes of the analysis of the survey data can be summarized as:
1) the notion of quality for numerical simulations is not standardized.
2) Although the adoption of published standards for numerical simulations is limited, part of the simulation process is often standardized to streamline recurring tasks.
3) Credibility assessment of simulations is mostly relying on subjective judgement of experts and the individual responsibility of the analyst.
Most respondents expressed a positive opinion of standardization and indicated a number of potential benefits expected primarily in the following areas:
1) analysis and data management (e.g. automation, traceability).
2) Simulation governance (i.e. ensuring reliability of results). Commitment of the management into the strategic development of numerical simulation activities is seen as a key requirement to meet challenges related to simulation governance.
3) Certification by simulation (i.e. partially superseding physical testing to assess the conformity of products to legal requirements). Although the replacement of physical testing with numerical simulations is already allowed by some standards, the risks associated with that are often poorly understood and quantified.