Comparison and Analysis of the Condensation Benchmark Results

Summary In the frame of the SARnet Network of Excellence, the need was fe lt for assessing the status of condensation models adopted in CFD codes relevant for nuclear reactor containment applications. The motivation for this work was provided by the increasi ngly widespread use of CFD in the analysis of containment behavior consequent to postulated s evere accidents, in which wall condensation can promote containment atmosphere mixing. Since standard models are seldom available in many CFD codes for dealing with c ondensation and in consideration of the different strategies envisaged for analyzing downscaled facilities or full scale containments, this aspect was considered worth of a specific attention. In this aim, after performing a review of the models available to the Participants in the network, appropriate Benchmark Problems were proposed to assess and compare their behavior. The University of Pisa took the charge of coordinating these efforts, proposing an initial step of the Benchmark (identified as the 0 th Step) aimed at comparing code responses among each other and with applicable correlations in the application to a classical problem of condensation on a flat plate; the reference geometrical and operati ng conditions for this step were selected as an idealization of those typical in the CONAN experimental facility, operated at the University of Pisa. Then, the 1 st Step of the activity involved addressing experimental data from the CONAN facility at different stea m mass fractions and velocities and the comparison of the measured condensation rates and of local heat f luxes with code predictions. Both the steps in this activity were fruitful, since they consti tuted a gradual and relatively systematic approach to the actual experimental condit ions, allowing for revising model details and discussing numerical and physical options. Though the compa rison with experiments involved up to now only a limited number of data points, the ac tivity is not considered to be completed and additional experimental data will be offe red in the future to obtain a broader assessment of codes in conditions of interest for sev ere accidents in light water reactors. A. INTRODUCTION In order to determine the risk associated with the presence of hydrog en in nuclear power plant containments during a hypothetical severe accident, predict ive codes are necessary. Computational Fluid Dynamics codes are very promising for this purpose, since