Abstract

This report introduces important ideas in the verification, validation and predictive capability assessment, commonly referred to as V&V, with a focus on engineering models. The report discusses the importance of V&V in light of the need for the everincreasing trust in model predictions. Since all models are only approximations to the real physical process they represent, they cannot be made more scientific, by adding complexity or wielding more computational power. This report introduces the tools and techniques required to assess the suitability of models for prognostic pronouncements and lays out how uncertainty quantification can increase the robustness of simulations. The report defines several key and often misunderstood concepts and reviews existing standards for V&V. It then presents the V&V process and delves deeper into verification, validation and predictive capability estimation. The report discusses why V&V must be an integral part of scientific computation and advocates for its continual use in modelling activities. Verification activities are reviewed with discussion on both code and solution verification. Different aspects of validation are presented and some advice is given about each, from the collection of validation data to the construction of validation measures in the presence of epistemic uncertainty. Predictive capability estimation is also discussed with some practical guidelines for different steps. Throughout the discussion about V&V the reader is pointed to key references allowing them to further investigate important concepts. This report is the second part of the DAWS series of reports on uncertainty quantification. The first of the series, which underpins many aspects discussed in the present report can be found here. Subsequent reports will focus on key topics in uncertainty propagation, calibration and sensitivity analysis.

This is v. 1.2.

Changes from the previous version:

• Updated abstract with information and links to reflect new reports and activities.

• Introduced version information.

Changes in v. 1.1.

• fixed a typo in the caption of Figure 3.12

If you need access to previous versions, please email p.hristov@liv.ac.uk