Impact on DNB predictions of mixing models implemented into the three-dimensional thermal-hydraulic code Thyc

The objective of this paper is to point out how departure from nucleate boiling (DNB) predictions can be improved by the THYC software. The EPRI/Columbia University E161 data base has been used for this study. In a first step, three thermal-hydraulic mixing models have been implemented into the code in order to obtain more accurate calculations of local void fractions at the DNB location. The three investigated models (A, B and C) are presented by growing complexity. Model A assumes a constant turbulent viscosity throughout the flow. In model B, a k-L turbulence transport equation has been implemented to model generation and decay of turbulence in the DNB test section. Model C is obtained by representing oriented transverse flows due to mixing vanes in addition to the k-L equation. A parametric study carried out with the three mixing models exhibits the most significant parameters. The occurrence of departure from nucleate boiling is then predicted by using a DNB correlation. Similar results are obtained as long as the DNB correlation is kept unchanged. In a second step, an attempt to substitute correlations by another statistical approach (pseudo-cubic thin-plate type Spline method) has been done. It is then shown that standard deviations of P/M (predicted to measured) ratios can be greatly improved by advanced statistics. (author). 7 figs., 2 tabs., 9 refs