ADS-79D-HDBK Downloaded from https://www.everyspec.com The V&V activities involve an iterative process. To effectively validate a model of interest, verification needs to be performed first. The proof-of-concept and verified computational model is employed to conduct preliminary analysis to facilitate proper setup for validation experiments. The experimental outcomes are used to benchmark simulation outcomes obtained from the computational model. Based on the outcomes of comparisons between simulation and experiments, various model calibrations or enhancements will be made to improve the model‘s predictive capability, if necessary.
The primary objective of validation is to establish a desirable level of confidence that a model or algorithm under consideration reasonably represents intended physical reality. Therefore, validation needs to be carefully planned and performed to cover the region of primary interest. To establish a database for further validation, the scope of validation, in terms of validation domain, needs to be determined first. Ideally, the validation domain should cover entirely the domain of intended application. Practically, however, there is often a budgetary or schedule limit to prevent excessive scope of validation.
This is particularly important for high fidelity physics based computational models for a complex system. Therefore, understanding the nature of the physics of the underlying computational model plays an essential role in establishing a feasible and tangible validation plan to ensure the accuracy and applicability of the model. Validation strategies can be further categorized based on amount of overlap between the validation domain and application domain.
As shown in Figure K-4, the ―Application Domain‖ designates the region where predictive capability from a computational model is needed for potential applications, while the ―Validation Domain‖ represents a region where the proper understanding of relevant physics is well established and confidence of the associated computational model has been quantitatively demonstrated by satisfactory agreement between predictive model and experiments. FIGURE K-4. Illustration of Various Validation Sceneries 64 64 SAND 2003 – 3769, Verification, Validation, and Predictive Capability in Computational Engineering and Physics, Sandia National Laboratories, 2003. 196