Computation-expensive problems have become more common in nuclear industry, originating from expensive analysis with the aim of reaching high accuracy. To overcome this challenge, surrogate modelling techniques are often proposed.
Passive containment cooling system (PCCS) is an essential pattern to ensure the containment integrity in the case of accidental conditions for AP600 reactor, so determination of the most influential parameter as regards the heat transfer rate is of great significance to the optimization design and security analysis of nuclear power plants.
This paper contributes to uncertainty quantification in condensation phenomena for PCCS. The purpose is to create a surrogate model instead of a deterministic model for sensitivity analysis on the account of saving computational time. The case used is an experiment carried out in 1998 for which the data is publicly available and was selected here as a showcase of the meta-model construction. In this paper, the database conducted by Anderson in 1998 is used to validate the computational fluid dynamics (CFD) model. By coupling Open-source CFD code code_saturne and uncertainty analysis code OpenTURNS from the SALOME Platform, a representative sample of input and output parameters is obtained using the design of experiments (DoE) technique. Thus, a surrogate model, including kriging and polynomial chaos metamodel, is constructed. In this way, sensitivity analysis of condensing efficiency is performed, which demonstrates the propagation of modeling uncertainty of condensation phenomena.