Nuclear Science and Engineering / Volume 182 / Number 1 / January 2016 / Pages 13-53
Technical Paper / Special Issue on the RELAP5-3D Computer Code / dx.doi.org/10.13182/NSE15-79
This paper discusses the role and the depth of the analysis required for merging, on one hand, suitable experimental data and, on the other hand, qualified code calculation results. The availability of an experimental and calculated qualified database is of utmost importance for the validation and qualification of codes. Such a database can be used not only for demonstrating that the code results are reliable and for performing an independent code assessment but also as a basis for developing and validating an uncertainty methodology. As discussed in several other papers and guidelines, an uncertainty methodology must rely on the availability of a qualified code and qualified procedures. The development of a Standardized Consolidated Calculated and Reference Experimental Database (SCCRED) that includes documentation such as the reference data set of the facility and of the tests, the qualification report of the code calculations, and the engineering handbook constitutes an approach envisaged also by the International Atomic Energy Agency to set up a qualified experimental and calculated database for verification and validation (V&V) purposes of computational tools and uncertainty methodologies.
In order to frame and to outline the role of a qualified database for performing a best-estimate and uncertainty analysis (UA), a summary of the approaches for performing the uncertainty evaluation is provided distinguishing among the methods based on propagation of input uncertainties, the methods based on code output accuracy propagation, and the predictive modeling methodology. The main issues from the review of the uncertainty methods are the following: (a) the identification of the uncertainty-method-user effect to be considered in addition to the more well-known code-user effect when the BEPU (Best Estimate Plus Uncertainty) approach is selected to perform accident analysis and (b) (partially connected also with the first issue) the need for validation of the uncertainty methods (in the same way as validation of a computer code is a fundamental prerequisite for application of the code).
To address the two issues above, the Code with the capability of Internal Assessment of Uncertainty (CIAU) is discussed in detail. In particular, it shall be noted that the CIAU method (which belongs to the methods based on code output accuracy propagation) needs a qualified set of experimental and code calculation results as input for performing a qualified intrinsically validated UA, which has a more limited uncertainty-method-user effect in comparison to the methods based on propagation of input uncertainties. Following the above consideration, the creation of SCCRED as a standardized consolidated reference experimental and calculated database constitutes a prerequisite for the development and application of the CIAU method, but at the same time such a qualified database can also be used for the V&V process of methods based on propagation of input uncertainties contributing to limit the uncertainty-method-user effect.