1. Reflux Condensation Heat Transfer of Steam-Air Mixture under Gas-Liquid Countercurrent Flow in a Vertical Tube Oct 7, 2004 Institute of Nuclear Safety System, Inc. Institute of Nuclear Technology Technical Support Project Takashi Nagae

2. 1 Purpose Shutdown PSA evaluation in Japan found that the mid-loop operation showed a high core damage probability. Mid-loop Operation ・・・ The RCS inventory is so low that it may decrease to the center line of the reactor coolant piping. Core decay heat is cooled by RHR Loss of RHR function Boiling away of the core at an early stage One of possible alternative cooling method is Reflux condensation by SG Fig 1 Mid-loop operation Reactor Vessel RHR Pump Pressurizer

3. 2 Reflux condensation ・・・ Core heat is removed by boiling Steam flows to the SG and condenses inside tubes Condensate on the up-flow side flows back to the core Fig 2 Reflux cooing Fig 3 Condensation with noncondensable gases in U-tube To investigate the reflux condensation heat transfer, we had experiments of reflux condensation and developed the new heat transfer models To estimate realistic availability of reflux condensation heat transfer we must consider following realistic conditions (1)Existence of noncondensable gases ← degrade heat transfer (2)Gas-liquid countercurrent flow ← flow regime effect to heat transfer  Default model in RELAP5 → is not confirmed whether they are applicable for condition (1)(2). （ It is reported that the model underestimates heat transfer in NUREG report. ）  Suggested models in other researchers → are applicable only for narrow condition

4. 3 Experiment Condenser tube inner diameter 19.3mm Condenser tube wall thickness 3.04mm Pressure ※ 0.1, 0.2, 0.4[MPa] Inlet steam flow rate ※ 0.45 ～ 1.9[g/s] Inlet air mass flow rate 0.03 ～ 0.18[g/s] ※ Condition during reflux condensation Fig 4 Test section (Double-pipe, concentric-tube heat exchanger) Table 2 Test conditions Measured temperatures are (1)Mixture of steam and air : T g (2)Condenser tube outer wall : T w,o (3)Coolant water: T c at 9 distances from the condenser tube by thermocouples New heat transfer models Local heat flux q’’ and interface condensation heat transfer coefficient h i are calculated and evaluated Nusselt numbers Nu i.

5. 4 Calculations － Calculation of local heat flux q’’ － (1)Inlet side of test section (2) Outlet side of test section － Calculation of interface condensation heat transfer coefficient － 1/K (z) = r w,i ln(r w,o /r w,i )/λ w (z) + 1/h f (z) + 1/h i (z) Overall heat resistance tube wall liquid film interface q’’ (z) = K(z) (T g (z) - T w,o (z)) Nu i (z) = h i (z)d w,i /λ s (z) K : overall heat transfer coefficient r w,i : tube inner radius r w,o : tube outer radius λ w : thermal conductivity of tube h f : heat transfer coefficient for liquid film Nu : Nusselt number d w,i : tube inner diameter λ s : thermal conductivity of film Nusselt number for the condensate film is obtained by applying the modified McAdams correlation to the Nusselt analysis for falling laminar film on a cold plate 外管内管 Steam & air Liquid film Coolant Outer tube Inner tube

6. 5 Experimental result RELAP5 default heat transfer model underestimate the heat transfer coefficients Moon’s empirical correlation F = h tot / h f = 2.58x10 -4 Re g 0.200 Re f 0.502 Ja -0.642 W air - 0.244 （ 6119< Re g <66586, 0.140< W air <0.972, 0.03<Ja<0.125 ） No measurement in low temperature region Exploration of the correlation overestimate the heat transfer coefficient Ｆ ：非凝縮性ガスによる熱伝達の劣化係数 h fpt ：非凝縮性ガスを含む場合の凝縮熱伝達率 h f ：純粋蒸気の Nusselt による凝縮熱伝達率の理論値 Re g ：蒸気・空気の混合ガスレイノルズ数、 Re g ：液膜のレイ ノルズ数 Ja : ヤコブ数、 W rir ：局所の空気質量流量比 Pressure = 0.1MPa Inlet steam flow rate = 1.23g/s, Inlet air flow rate = 0.06g/s T g ： Mixture of Steam and air temperature T w,o ： Condenser tube outer wall temperature T c ： Coolant temperature Fig 5 Temperature profile (at steady state) Test condition

7. 6 Development of Heat transfer models Correlation for local Nusselt number is obtained as a function of the steam-to-air partial pressure ratio and plotted in Fig 6. Nu i = 120 (P s /P a ) 0.75 ， (Nu <500) (1) Eq (1) is not valid for turbulent flow region and we can ’ t neglect the influence of gas flow Fig 7 Comparison between measurements and calculation (Nusselt numbers)  To develop the correlation in turbulent flow region, steam Reynolds number is adopted to Eq (1). Nu i = 120 (P s /P a ) 0.75 max(1.0 ， aRe, s b ) (Re, s ≦ 5000 ， a=0.0012 ， b=1.0) (2)  Comparing between the measurement and calculation from Eq (2) shows good agreement not only in laminar flow region but also in turbulent flow region. Eq （１） Laminar flow Turbulent flow Fig 6 Nusselt numbers Correlation for the local heat transfer coefficients Turbulent flow Laminar flow

8. 7 Improvement of Heat transfer models Additional experiment (increasing air mass flow to 0.2-1.0g/s) to improve the correlation in turbulent flow region Fig 8 Comparison between measurements and Eq(2) (air mass flow: 0.2-1.0g/s ） □ 0.2MPa △ 0.4MPa In low heat transfer area, Eq(2) underestimate the Nusselt numbers Estimation only by the steam Reynolds number is not enough when air mass flow rate increases In low heat transfer area, Eq(2) over estimate the Nusselt numbers Effect of Re,s ｂ (b=1) is too big +50% -50%

9. 8 Improvement of Heat transfer models The steam Reynolds number Re,s in Eq(2) was changed to steam-air mixture Reynolds number and Eq(3) was derived (a = 0.0035 ， b = 0.8)(3) Comparing between the measurement and calculation from Eq (3) shows good agreement not only in laminar flow region but also in turbulent flow region including the air mass flow increasing condition Fig 9 Comparison between measurements and Eq(3) (air mass flow: 0.03 ～ 1.0g/s ） ◇ 0.1MPa □ 0.2MPa △ 0.4MPa +50% -50%

10. 9 Temperature measurements by thermocouples may contain errors, so calculated local heat transfer coefficients may have errors To evaluate the accuracy of calculation, we calculated the mixture of steam and air temperature profile and compared with measurements It was verified that Eq (3) effectively simulate the temperature profile. We confirmed the validity of Eq (3) as heat transfer model Fig 10 Comparison between measurements and calculation (temperature profile) Evaluation of Heat transfer models Eq (3)

11. 10 Evaluation of Heat transfer models Comparing local heat transfer coefficient between measurements and Eq (3) （ 1/h c = 1/h f + 1/h i ） Condenser tube inner diameter 16.56mm Condenser tube wall thickness 1.25mm Pressure ※ 0.1, 0.15, 0.25[MPa] Inlet steam flow rate ※ 0.37 ～ 0.91[g/s] Inlet air mass flow rate 0.15 ～ 0.68[g/s] Moon’s experiment (test condition) Fig 11 Comparison between measurements and calculation (h c ) ◇ 0.1MPa □ 0.15MPa △ 0.25MPa +50% -50% We confirmed the validity of Eq (3) as heat transfer model in with Moon ’ s experiment Measurement limitation Comparing local heat transfer coefficient between measurements and Eq (3)

12. 11 To estimate realistic availability of reflux condensation heat transfer we must consider following realistic conditions 1. Existence of noncondensable gases 2. Gas-liquid countercurrent flow ７． Summary  An experimental facility was constructed to study reflux condensation heat transfer in the riser section of PWR U-tubes  New heat transfer models were developed ① Correlation for local Nusselt number was obtained as a function of the steam-to-air partial pressure in laminar flow region ② In turbulent flow region steam-air mixture Reynolds number was adopted  It was verified that New heat transfer models effectively simulate the temperature profile

13. 12 ８． Future plan Incorporation of new models into RELAP5 Validation of new models in RELAP5

14. Reference

15. 14 Comparison with RELAP5 heat transfer models Comparing local heat transfer coefficient with measurements, Eq (2) and RELAP5 models 層流域 乱流域 層流域 乱流域 Fig 9 Interface condensate heat transfer coefficient （ Comparison with RELAP5) Laminar flowTurbulent flow 0.1MPaUnderestimate (20 ～ 50%) Good agreement 0.4MPaUnderestimate （ 10 ～ 25% ） Underestimate （ 45 ～ 50% ） RELAP5 condensation heat transfer models tend to underestimate the heat transfer coefficient in all region. We will incorporate the new models into RELAP5 as a option and we will be able to calculate SG reflux condensation more accurate than default models. ※ Now we are under additional experiment because data in turbulent flow region is not sufficient. Table 3 Calculation with RELAP5 heat transfer models Turbulent flow Laminar flow Eq (2) measurement P steam /P air hi (W/m 2 K)