Numerical calculations for air-water tests on CCFL in different-scale models of a PWR hot leg

Air-water CCFL (countercurrent flow limitation) tests were previously carried out at Kobe University using the 1/5th scale rectangular channel and 1/15th scale circular tube simulating a PWR hot leg. Then numerical calculations for these tests and full-scale PWR conditions were made using the CFD code, FLUENT6.3.26. At Forschungszentrum Dresden-Rossendorf (FZD), similar tests were previously carried out for both air-water and steam-water flows using the 1/3rd scale rectangular channel simulating a PWR hot leg installed in the pressure chamber of the TOPFLOW facility. In this paper, numerical simulations for the air-water CCFL tests of FZD using FLUENT6.3.26 are presented and compared with the experimental data obtained at Kobe University and FZD. In the calculations, the VOF (volume of fluid) model or two-fluid (2F) model was used. In the 2F model, we used the combination of three correlations on the interfacial friction coefficients as a function of void fractions, which had been validated for the 1/15th and 1/5th scale tests at Kobe University. Calculation parameters were the air flow rates and air inlet locations, which were at the top of the reactor vessel simulator simulating the FZD test facility (inlet 1) and the opposite side of the hot leg junction simulating the test loops at Kobe University (inlet 2). Conclusions were as follows : (1) the calculated CCFL characteristics using the 2F model for the FZD tests agreed well with the 1/15th scale circular tube data obtained at Kobe University and the calculated results for full-scale PWR conditions, which supported the validity of the 1/3rd scale rectangular channel to simulate CCFL in circular tubes; (2) there were no significant differences between the calculated CCFL characteristics with the air inlet 1 and inlet 2, which indicated that the air inlet location did not influence CCFL behavior in a hot leg; and (3) comparison with the FZD data showed that the calculations using the 2F and VOF models overestimate