Abstract

Numerical simulations play a pivotal role in the development and testing of novel technologies, enabling the improvement of product quality whilst reducing costs and shortening time to market. Increasing reliance on numerical simulations in product development and verification demands for robust simulation governance, i.e. procedures to ensure the reliability of the predictions based on numerical simulations. As part of the research project SPRUCE (Standardization Practices for a Responsible Use of Computational models in Engineering), a survey was recently conducted among simulation experts in Sweden. The outcomes of the survey suggest that the current practice for quality assurance of numerical simulations in industrial applications is largely dependent on the individual responsibility of the analysts, possibly assisted by supervising experts within the same organization. Furthermore, the analysis results are rarely questioned by designers and managers who seek guidance in numerical simulations to drive their own decisions. The prevalence of subjective judgement without a structured, possibly standardized, evaluation method introduces an element of uncertainty which is hard to estimate and hinders the process of reaching consensus among experts (who, for example, might attribute different weights to the approximations made in the models about physics, solution method, loads, etc). Overconfidence in numerical models enhances the risks associated with choosing numerical simulations over physical testing, which might have dramatic consequences, for instance in the verification of the conformity of products to safety requirements (see the point about “Simulation supporting certification”, in the NAFEMS Simulation Agenda). In this study, we present a review of quality assurance concepts, methods, and standards in the context of engineering simulations. Particular emphasis is given to the process of building confidence in the results of numerical simulations to support certification in highly regulated industries such as road transport safety, construction, and offshore. The role of verification, validation, and simulation process and data management in the overall quality assessment will be examined. A comparison between virtual and physical testing will be outlined, with special regard to the quality requirements on the data generated to support decisions in product development and certification processes. Preliminary results from the project STEERING (STandardization Experiments for Enhanced Reliability and confidence of engineerING simulations) on the potential of standardization as a key driver for quality improvement in Computer-aided Engineering (CAE) will be illustrated.