Scaling and extrapolation of hydrogen distribution experiments

Abstract

Important physical processes within a containment system, which govern the long-term behaviour under severe accident conditions have been analyzed with respect to the scaling of relevant test rigs. This analysis has been performed under contract with the Commission of the European Communities on the basis of the equations processed within a typical long term containment analysis code like the CONTAIN code. An improved set of conservation equations for the involved components (air, hydrogen, vapor etc.), has been subject to a detailed dimensional analysis. This resulted in a set of dimensionless parameter groups which determine the similarity requirements necessary to warrant simple extrapolation of the results to full size reactor containments. The discussion of the physical meaning of the parameter groups showed that steady-state and transient events could possibly require different similarity criteria. For the establishment of a long term natural circulation pattern the strong coupling with a preceding LOCA blowdown process has been demonstrated. Inevitable distortions of mixing effects for hydrogen may be expected for small scaled experiments, if it is not possible to perform experiments with a thoroughly scaled distribution of heat sources and heat sinks. The hydrogen source terms into the containment must be properly related to a volumetric scaling concept. The understanding of the transport process is strongly dependent on a detailed analysis of the Strouhal Number to avoid misinterpretation of the results. Only limited overall similarity seems achievable. The implications of the similarity requirements with the applied empirical correlations or constants in combination with a chosen nodalisation concept have been addressed. Code verification is based on comparison of selected measured with calculated parameters which leads to conclusions concerning an optimal choice of correlations and the proper nodalisation of the test rigs. Heat transfer correlations and local flow resistance determination are important empirical elements for a successful reanalysis of experiments. Facility dependent code verification can only be avoided, if the dimensional dependence of the empiricism in code application is assessed.