DEVELOPMENT AND VALIDATION OF A NEW DRIFT FLUX MODEL IN ROD BUNDLE GEOMETRIES

DEVELOPMENT AND VALIDATION OF A NEW DRIFT FLUX MODEL IN ROD BUNDLE GEOMETRIES Ikuo Kinoshita and Toshihide Torige Institute o Nuclear Saety System, Inc. 64 Sata, Mihama-cho, Mikata-gun, Fukui 99-25, Japan kinoshita@inss.co.jp; torige.toshihide@inss.co.jp Minoru Yamada MHI Nuclear Engineering Co., Ltd. 3-3- Minatomirai, Nishi-ku, Yokohama-city, Kanagawa 22-84, Japan minoru2_yamada@mnec.mhi.co.jp Takashi Hibiki School o Nuclear Engineering, Purdue University 4 Central Drive, West Laayette, IN 4797-27, USA hibiki@purdue.edu ABSTRACT In the authors previous studies, low low data were taken systematically rom the Purdue University rod bundle acility at atmospheric pressure with air and water as working luids. The data set was used to develop a new drit lux correlation or accident analyses to predict void ractions over a wide range o twophase low conditions in rod bundle geometries. In this study, the new correlation was implemented in the RELAP5/MOD3 code, and the void raction prediction perormance was compared with the correlation which was installed in the original RELAP5/MOD3 code. Numerical analyses carried out or the Purdue University rod bundle tests showed that the results obtained using the new correlation were in good agreement with the test data and represented an improvement over those obtained rom the correlation. Numerical analyses carried out or the void proile tests in the THTF at ORNL, however, showed that the new correlation developed under low low atmospheric pressure air/water conditions was less accurate or low low high pressure steam/water conditions. The new correlation needs improvements in order to scale the pressure eects appropriately. KEYWORDS Drit lux, Rod bundle, Void raction, Low liquid low, RELAP5. INTRODUCTION Existing drit lux correlations or rod bundles may not appropriately consider two-phase low mechanisms in low low and low pressure conditions during accidents. For instance, at low liquid low and low pressure conditions, downward low at the outer perimeter has been observed in a large diameter pipe [] and large concentrations o gas have been measured at the central higher velocity regions o large diameter channels [2]. These characteristics are attributed to recirculating low patterns. Furthermore, large concentrations o void have been observed in the central regions o a rod bundle at low pressure and low low conditions [3]. Thereore, similar to large diameter pipes, recirculating low patterns may occur in rod bundles at low low NURETH-6, Chicago, IL, August 3-September 4, 25 3

and low pressure conditions. Low pressure systems introduce greater buoyancy orces because o an increased density dierence between the continuous and dispersed phases. On the other hand, at high pressure and high liquid low conditions, the decreased buoyancy orces and increased turbulence would prohibit any recirculating low patterns rom developing in rod bundles. Earlier drit lux correlations or rod bundles may have placed priority on use in high pressure and high liquid low conditions because these are more representative o a prototypic power plant. This may provide an explanation or why recirculating low patterns have not been previously considered or drit lux models in rod bundles. However, it may be important or a rod bundle drit lux correlation to consider these two-phase low mechanisms. For instance, low low and low pressure conditions are relevant or a loss-o-coolant accident (LOCA) or a loss-o-rhrs (residual heat removal systems) event during mid-loop operation. In the authors previous studies [4, 5], low low data were taken systematically rom the well-scaled Purdue University 8 x 8 rod bundle acility at atmospheric pressure with air and water as working luids. Results indicated a signiicant increase in the distribution parameter when mixture volumetric lux was relatively low. The collected data were used to develop a new drit lux correlation or accident analyses to predict void ractions over a wide range o two-phase low conditions in rod bundle geometries. The Hibiki and Ishii correlation or drit velocity [6] that relected the eects o recirculating low patterns was implemented in the new correlation. The distribution parameter was then back calculated rom the onedimensional drit lux model general expression using the drit velocity and the associated data. It was conirmed that the distribution parameter increased at low volumetric lux conditions beore exponentially decreasing as supericial velocity increased, and approached asymptotically to a constant value which matched the distribution parameter determined or rod bundles at high low and high pressure conditions. These characteristics were associated with recirculation low patterns. A perormance analysis demonstrated that the new correlation improved upon existing correlations and had an average relative error o ±4.5% when predicting the void raction o the database utilized in this work. In this study, the new correlation is implemented in the RELAP5/MOD3 code [7], and numerical analyses are carried out or the Purdue University rod bundle experiments to investigate the applicability o the new correlation to the RELAP5 code. In addition, numerical analyses are carried out or the two-phase mixture level swell tests in the Thermal Hydraulic Test Facility (THTF) at the Oak Ridge National Laboratory (ORNL) [8, 9] to investigate the applicability o the new correlation to low liquid low and high pressure conditions expected in a small break LOCA. 2. THE NEW DRIFT FLUX MODEL IN ROD BUNDLES 2.. Overview o the Rod Bundle Correlation Development The drit lux model [] is given by jg v C j V, () g gj where vg,, jg, C, j, and Vgj are gas velocity, void raction, gas volumetric lux, distribution parameter, mixture volumetric lux, and drit velocity, respectively. The < > and << >> notations designate area-averaged and area-averaged void weighted mean values, respectively. In general, closure relations or the distribution parameter and the drit velocity are developed or application in certain low regimes, geometries, luid parameters or conditions, etc. Much work has been dedicated towards calculating drit lux parameters at various conditions. In order to develop the NURETH-6, Chicago, IL, August 3-September 4, 25 3

new correlation that accounts or two-phase low mechanisms at low low and low pressure conditions, the data reported rom Clark et al. [4] were utilized which were collected systematically rom the well-scaled 8 x 8 rod bundle acility at atmospheric pressure low low conditions with air and water as working luids. The drit velocity was determined or the data by using the correlation by Hibiki and Ishii [6]. Hibiki and Ishii ormulated the drit velocity correlation or large low channels that spans stagnant to high liquid low conditions. A signiicant attribute o their correlation is that gas velocity, namely void raction, is used to appropriately scale the eect o recirculating low patterns at low low conditions on drit velocity. The distribution parameter was then back calculated rom Eq. () using the drit velocity and the associated data. Figure shows the distribution parameter values obtained rom the low low and low pressure data [5]. Here the superscript + designates a non-dimensional value by considering bubble rise velocity which is given by the denominator in Eq. (2). j j 2 g / 4 2, j j g / 4 (2) As shown in Fig., the distribution parameter irst increases at low volumetric lux conditions beore exponentially decreasing as volumetric lux increased, and approaches asymptotically to a constant value. These regions may be characterized by diering two-phase low structures. The initial increase in the distribution parameter towards a imum value, C, may be attributed to an increasing concentration o gas in the high velocity (central) regions o the low channel and downward liquid low at the outer perimeter. On the other hand, increasing the low rate beyond the conditions or C enhances cocurrent low which would eectively decrease the distribution parameter. This decrease approaches a constant value designated as C j. From the data trend, the value o Cj is approximately., which matches the distribution parameter determined by Ozaki et al. [] or rod bundles at prototypic conditions. The mixture volumetric lux condition at which C occurs is reerred to here as j. As liquid C volumetric lux increases, j increases, and C C decreases. These transitions are expected to be related to changes in two-phase low structure. Distribution Parameter, C [-] 4 3 2 Data Correlation <j > =. m/s <j > =. m/s <j > =. m/s <j > =. m/s <j > =.25 m/s <j > =.25 m/s <j > =.5 m/s <j > =.5 m/s <j > =.75 m/s <j > =.75 m/s <j > =. m/s <j > =. m/s Distribution Parameter, C [-] 3 C 2 C L C j = <j + > <j + > C C H <j > =.5 m/s C j =. 2 3 4 5 6 7 Non-D. Mixure Volumetric Flux, <j + > [-] 5 5 2 25 3 35 Non-D. Mixure Volumetric Flux, <j + > [-] Figure. Distribution parameter data and schematic illustration [5]. NURETH-6, Chicago, IL, August 3-September 4, 25 32

Table I. The new drit-lux correlation or rod bundles [5].39.39 jg jg Vgj Vgj, B e V gj, P e Hibiki and Ishii [6] C L or j j g C C C C Ishii [2] C CH or j j C. j C H CH j..84e j C C j C L jg C j j C j m j b C min.284.25,.52 j j Vgj m ; b C C For the new correlation, the distribution parameter was approximated by a linear increasing and exponential decreasing region as show schematically in Fig.. A boundary condition was used to distinguish between each region as C C L or j j C or j j C H C. (3) As gas volumetric lux degreased to zero, the initial distribution parameter was set to C j.. This represented an even distribution o void raction over the rod bundle cross section. On the other hand, the exponential decrease was described using a least squares curve itting approach. C. j H j..84e (4) The transition condition in Fig. was associated with a ical void raction,, corresponding to a transition in the two-phase structure. The relationship between ical void raction and dimensionless liquid volumetric lux was approximated by min.284.25,.52. j (5) The relationship between j and C j was determined by the ical void raction and the dritlux general expression () as ollows. j m j b (6) C NURETH-6, Chicago, IL, August 3-September 4, 25 33

Vgj m ; b C C (7) Finally, it was necessary to scale the distribution parameter based on luid properties or system pressure. This was done by incorporating the correlation by Ishii [2]. The new correlation is detailed in Table I. 2.2. Perormance Analysis The drit-lux correlations are compared against the low liquid velocity and low pressure data [4]. A comparison between the void ractions predicted by the new correlation and the measured values is shown in Fig. 2. Here, there are no distinct biases present in void raction prediction by the new correlation and the random error is mostly within the ±% error band. A comparison between the void ractions predicted by the correlation [3] and the measured values is shown Fig. 2. The correlation, also known as the Chexal-Lellouche correlation, was developed to cover the ull range o conditions in rod bundles and implemented in the RELAP5/MOD3 code. Disagreement is observed between the correlation and data trends at lower void ractions. This error may be attributed to the empirical relations o the correlation not appropriately considering secondary low patterns. Calculated Void Fraction, < > [-].8.6.4.2 Data <j > =. m/s <j > =. m/s <j > =.25 m/s <j > =.5 m/s <j > =.75 m/s <j > =. m/s % Error Calculated Void Fraction, < > [-].8.6.4.2 Data <j > =. m/s <j > =. m/s <j > =.25 m/s <j > =.5 m/s <j > =.75 m/s <j > =. m/s % Error.2.4.6.8 Measured Void Fraction, < > [-].2.4.6.8 Measured Void Fraction, <> [-] Figure 2. Void raction prediction error [5]; the new correlation and the correlation. 3. PURDUE UNIVERSITY ROD BUNDLE TEST ANALYSES The new correlation is implemented in the RELAP5/MOD3 code, and numerical analyses were carried out or the Purdue University rod bundle tests [4] to investigate the applicability o the new correlation to the RELAP5 code. The analysis results are compared with those by the correlation implemented in the original RELAP5 code. NURETH-6, Chicago, IL, August 3-September 4, 25 34

#7 units in mm Acrylic Casing 4 mm Impedance Meter Electrode Pressure Port and Impedance Meter Elevations Spacer Grid Elevations #6 6.7 mm #5 #4 6.7 mm 2.7 mm Acrylic Rods 4 mm 2958 234 836 79 394 266 #3 #2 95 788 346 2669 2229 3 mm # 464 Figure 3. Rod bundle test section [4]; cross sectional view and axial view. 3.. Outline o Purdue University Rod Bundle Tests Two-phase low low data were taken systematically rom a well-scaled 8 x 8 acrylic rod bundle test acility at atmospheric pressure with air and water as working luids [4]. Figure 3 shows the rod bundle test section as cross-sectional view and axial view. The casing o the rod bundle was made o transparent acrylic plates in a 4 x 4 mm square duct. The acrylic rods had a diameter o 2.7 mm and a pitch o 6.7 mm. There were 7 spacer grids located along the length o the test section to limit vibration and simulate spacers in the prototypic design. The impedance meters 7 were located at z/d H, Case =.7, 9.,., 2.2, 3., 4.5, and 2., respectively. The seven pressure ports were at the same locations. The height o the test acility was 3 m. Area average void raction measurements were perormed using impedance meters 4 and 6. These locations were ar enough downstream rom the spacer grids so that their eect o interacial structure was negligible [4]. The six dierent liquid low conditions utilized were j =.,.,.25,.5,.75,. m/s. 3.2. Analysis Methods Nodalization or the analyses is shown in Fig. 4. The bundle domain was represented by a RELAP5 pipe component divided into 26 volumes. The part o the test section was represented by ine node rom impedance meters 4 to 6. To inject water and air into the bottom o the bundle domain, time-dependent volumes and time-dependent junctions were connected to the bottom o the bundle domain. The injection low rates were controlled so that the liquid and gas volumetric luxes at the measuring position matched those Time TMDPVOL Dependent 5Vol.9 (Pressure boundary) Single SNGLJUN Jun.8 8 SNGLVOL Single Vol.7 7 Single SNGLJUN Jun.6 6 (26) (25) (24) (23) (22) (2) (2) (9) #6 2 (8) (7) (6) (5) (4) #4 (3) (2) Pipe PIPE 5 5 () () 4 (9) 2 (8) 6 (7) (6) (5) (4) (3) (2) () Time Dependent TMDPJUN Jun.2 2 Time TMDPJUN Dependent 44Jun.4 Time Dependent Vol. Time Dependent Vol.3 TMDPVOL (Water) TMDPVOL 3 (Air) Outlet plenum Bundle domain Figure 4. Nodalization or the analyses. NURETH-6, Chicago, IL, August 3-September 4, 25 35

o the tests. The injected water and air temperature were the same as the tests. The outlet plenum was simulated by a single volume. To get a ree outlow boundary condition, the time-dependent volume and a normal junction were connected to the outlet plenum. The pressure was the same as that o the test. Analysis cases in this study are listed in Table II. Table II. Analysis Cases Series No. Measured j Measured j g Measured void [m/s] [m/s] raction [-] Series..6 2.2.4.6 Series 2..2.25 6.33.5.7 3.3. Analysis Results 3.3.. Series ( j =. m/s, j g =.629 m/s) Figure 5 compares the predicted void raction proiles by the new and the correlations or the test carried out at j =. m/s, j g =.629 m/s. When the axial positions become higher, the predicted void ractions slightly increase. This is because the gas pressures are lower at the higher axial position. In this sense, the RELAP5 code implementing the new correlation makes a reasonable prediction as well as the correlation. In Fig. 5, the measured void raction at the position o the impedance meters #4 is also plotted. The void raction predicted by the new correlation is within the ±% error band o the measured value. On the other hand, the void raction predicted by the correlation is below the ±% error band o the measured value. 3.3.2. Series 2 (j =.2 m/s, j g = 6.33 m/s) Figure 5 compares the predicted void raction proiles by the new and the correlations or the test carried out at j =.2 m/s, j g = 6.33 m/s. As is Fig. 5, when the axial positions become higher, the predicted void ractions slightly increase. This case also shows that the RELAP5 code implementing the new correlation makes a reasonable prediction as well as the correlation. In Fig. 5, the measured void raction at the position o the impedance meters #6 is also plotted. Although both the void ractions predicted by the new correlation and the correlation are within the ±% error band o the measured value, the new correlation makes a better prediction than the correlation. 3.3.3. Void raction prediction Figure 6 compares the predicted void ractions by the new and the correlations, arranged along with the non-dimensional gas volumetric luxes. In the case o Series (j =. m/s), the new correlation leads to better agreement with the data than the correlation. This is because the new correlation appropriately considers two-phase low mechanisms in stagnant liquid low conditions. On the other hand, in the case o Series 2 (j ~. m/s), there is no signiicant dierence in the void raction prediction accuracy between the new and the correlations. This may be due to the recirculating low pattern being diminished at high low conditions. In act, as shown in Fig., the increase o the distribution parameter at j =. m/s is not so signiicant compared to that at j =. m/s. NURETH-6, Chicago, IL, August 3-September 4, 25 36

.8 j =. m/s j g =.629 m/s Data [4].8 Measuring Position Void raction [-].6.4 Measuring Position Void raction [-].6.4.2.2 j =.2 m/s j g = 6.33 m/s Data [4].5.5 2 2.5 3 Axial Position [m].5.5 2 2.5 3 Axial Position [m] Figure 5. Predicted void raction proiles Series ( j =. m/s, j g =.629 m/s) and Series 2 ( j =.2 m/s, j g = 6.33 m/s). Measured void ractions are also plotted with the ±% error band..8 j =. m/s.8 j ~. m/s Void raction [-].6.4 Void raction [-].6.4.2 Data [4].2 Data [4] 2 3 4 5 Non-D. Gas Volumetric Flux, j + g [-] 2 3 4 5 Non-D. Gas Volumetric lux, j + g [-] Figure 6. Void raction prediction Series and Series 2. 4. ORNL/THTF ROD BUNDLE TEST ANALYSES To investigate the applicability o the new correlation to low liquid low and high pressure conditions expected in a small break LOCA, the numerical analyses by the RELAP5 code implementing the new correlation were conducted or the ORNL/THTF void proile tests [8, 9]. The analysis results are compared with those by the original RELAP5 code with the correlation. 4.. Outline o ORNL/THTF Void Proile Tests The THTF is a high pressure bundle thermal-hydraulics loop, and designed to produce a thermal-hydraulic environment similar to that expected in a small break LOCA. Figure 7 shows a sectional view o the THTF test section. The test section consisted o the bundle domain, shroud box, shroud-plenum annulus, and test section barrel. The bundle domain consisted o 6 heated rods and 4 unheated rods, their pitches were.27cm. The outside diameter o all heated rods was.95cm. These are the same values as typically used or a 7 x 7 uel assembly. The heated rods had a heater surrounded by boron nitride, and a stainless steel sheath outside. The axial void raction proiles were obtained rom dierential pressure measurements. NURETH-6, Chicago, IL, August 3-September 4, 25 37

Figure 7. Rod bundle test section o the THTF [8]; cross-sectional view and pressure instrumentation. Figure 7 indicates locations o those measurements. The bundle domain was divided into 9 dierential pressure cells along the axis length. The volume average void ractions were obtained at the mid-heights o the dierential pressure cell measurements. As or dierential pressure cells including the beginning o the boiling length, the void ractions were obtained at the mid-heights o the two-phase regions o the cells. The void proile tests were started by boiling o water rom the bundle, which was originally illed with water. On reaching the quasi-steady state, subcooled water was supplied rom the bundle domain bottom in order to compensate or the amount o transpiration rom the bundle domain top, and the bundle was uncovered partially. 4.2. Analysis Methods Nodalization or the analyses is shown in Fig. 8. The bundle domain (simulated heated region only) was divided into volumes corresponding to the dierential pressure cells, so that the volume center positions became the void raction evaluation positions. The dierential pressure cells which included the beginning o the boiling length were also divided by the beginning o the boiling length. The heated rods and the shroud box surrounding the bundle domain were simulated as the heat structure. The ractional heat loss rom the shroud box was evaluated in the tests. In the analyses the heat loss was simulated assuming that it was axially uniorm. Figure 8. Nodalization or the analyses. NURETH-6, Chicago, IL, August 3-September 4, 25 38

To inject subcooled water into the bottom o the bundle domain, a time-dependent volume and a timedependent junction were connected to the bottom o the bundle domain. The injection low rate was controlled so that the collapsed water level o the bundle domain matched that o the test, and the injected water temperature was the same as the test. To get a ree outlow boundary condition, the time-dependent volume and a normal junction were connected to the top o the bundle domain. The pressure o the top o the bundle domain was the same as that o the test. Analysis cases in this study are listed in Table III. Table III. Analysis Cases Test No. Fluids Pressure Power (gas/liquid) [MPa] [kw/m] Test 3.9.J steam/water 4.2.7 Test 3.9.K steam/water 4..32 Test 3.9.M steam/water 6.96.2 Test 3.9.N steam/water 7.8.47 Test 3.9.AA steam/water 4.4.27 Test 3.9.BB steam/water 3.86.64 Test 3.9.CC steam/water 3.59.33 Test 3.9.DD steam/water 8.9.29 Test 3.9.EE steam/water 7.7.64 Test 3.9.FF steam/water 7.53.32 4.3. Analysis Results 4.3.. Test 3.9.J The analysis results o void raction proiles or Test 3.9.J are shown in Fig. 9. The dierence o void raction predictions between the new correlation and the correlation becomes larger when the axial positions become higher. This dierence can be understood by the relation between void ractions and steam volumetric luxes. Figure presents the void ractions obtained by applying the new correlation and the correlation directly, arranged according to non-dimensional gas volumetric luxes. Under the test conditions, non-dimensional steam volumetric luxes are in the range 5. For this range, the dierence o void raction predictions becomes larger when the gas volumetric luxes become larger. The void ractions calculated by the new correlation tend to be under-predicted compared with the measured data. Similar results are obtained or simulations o test 3.9.M, AA, BB, CC, DD and EE. 4.3.2. Test 3.9.FF The analysis results o void raction proiles or Test 3.9.FF are shown in Fig. 9. In this case, there is no signiicant dierence o the void raction prediction between the new correlation and the correlation, and both correlations give good agreement with the data. Figure presents the void ractions obtained by applying the new correlation and the correlation directly, arranged according to non-dimensional gas volumetric luxes. Using the test conditions, non-dimensional gas volumetric luxes are in the range. For this range, there is no signiicant dierence between the correlations, and both the correlations lead to similar results and give good agreement with the data. Similar results are obtained or simulations o tests 3.9.K and N. NURETH-6, Chicago, IL, August 3-September 4, 25 39

Void raction [-].8.6.4 Test 3.9.J Data [9] Mixture level [9] Void raction [-].8.6.4 Test 3.9.FF Data [9] Mixture level [9].2.2.5.5 2 2.5 3 3.5 4 Axial Position [m].5.5 2 2.5 3 3.5 4 Axial Position [m] Figure 9. Predicted void raction proiles Test 3.9.J and Test 3.9.FF. Below Mixture Level Below Mixture Level.8.8 Void raction [-].6.4 Void raction [-].6.4.2 Test 3.9.J.2 Test 3.9.FF 5 5 2 Non-D. Gas Volumetirc Flux, j + g [-] 5 5 2 Non-D. Gas Volumetirc Flux, j + g [-] Figure. Void ractions obtained by applying the new correlation and the correlation directly Test 3.9.J and Test 3.9.FF. 4.3.3. Void raction prediction Figure compares the predicted void ractions by the new correlation and the correlation versus the void ractions measured in the ORNL/THTF tests. The void ractions obtained using the new correlation are under-estimated or most o the cases. Figure shows the drit lux plane. The gas velocities predicted by the new correlation tend to reach those by the correlation when system pressures increase. This tendency indicates that the new correlation developed under low low atmospheric pressure air/water conditions is less accurate or low low high pressure steam/water conditions. 5. DISCUSSION Figure 2 shows the behavior o the new correlation as it is scaled to the typical BWR pressure, P = 7 MPa, and PWR pressure, P = 5.5 MPa, reproduced rom igures in [5]. NURETH-6, Chicago, IL, August 3-September 4, 25 4

Void raction [-].8.6.4.2 J K M N AA BB CC DD EE FF J K M N AA BB CC DD EE FF Data J Data K Data M Data N Data AA Data BB Data CC Data DD Data EE Data FF Non-D. Gas Velocity, v g + [-] 2 5 5 J K M N AA BB CC DD EE FF J K M N AA BB CC DD EE FF 2 3 4 5 6 7 8 9 Non-D. Gas Volumetirc Flux, j + g [-] 2 3 4 5 6 7 8 9 Non-D. Mixture Volumetirc Flux, j + [-] Figure. All cases void raction prediction gas velocity prediction. Open symbols represent the data obtained by the tests conducted under pressures o about 4. MPa, while illed symbols the data by the tests under pressures o about 7.5 MPa. 2 j =. m/s 2 j =. m/s Non-D. Gas Velocity, v g + [-] 8 6 4 2 Data [4] P =. MPa (Air-Water) Correlation P =. MPa (Air-Water) P =. MPa (Steam-Water) P = 7. MPa (Steam-Water) P = 5.5 MPa (Steam-Water) Ozaki et al. [] P =. MPa (Steam-Water) P = 7. MPa (Steam-Water) P = 5.5 MPa (Steam-Water) 2 4 6 8 2 Non-D. Mixure Volumetric Flux, j + [-] Non-D. Gas Velocity, v g + [-] 8 6 4 2 Data [4] P =. MPa (Air-Water) Correlation P =. MPa (Air-Water) P =. MPa (Steam-Water) P = 7. MPa (Steam-Water) P = 5.5 MPa (Steam-Water) Ozaki et al. [] P =. MPa (Steam-Water) P = 7. MPa (Steam-Water) P = 5.5 MPa (Steam-Water) 2 4 6 8 2 Non-D. Mixure Volumetric Flux, j + [-] Figure 2. Behavior o the new correlation at plant-scale conditions [5]; j =. m/s and j =. m/s. Values obtained with the correlation o Ozaki et al. [] are also plotted. As pressure increases, the distribution parameter becomes relatively constant and approaches the value o the Ozaki et al. [] correlation, which was developed with regard to primarily high pressure data. However, there remains a dierence between them, which may be because the distribution parameter C o the new correlation does not properly scale the physical eects under high pressure conditions. The new correlation needs improvements in order to scale the pressure eects appropriately. 6. CONCLUSIONS Existing drit lux correlations or rod bundles may have placed priority on use in high pressure and high liquid low conditions because these are more representative o a prototypic power plant. Thereore, recirculating low patterns which may occur at low pressure and low low conditions have not been previously considered or drit lux model in rod bundles. However, it may be important or a rod bundle drit lux correlation to consider these two-phase low mechanisms because, or instance, low low and low pressure conditions are relevant or a LOCA or a loss-o-rhrs event during mid-loop operation. NURETH-6, Chicago, IL, August 3-September 4, 25 4

In this study, a new correlation was implemented in the RELAP5/MOD3 code, and the void raction prediction perormance was compared with the correlation which was installed in the original RELAP5/MOD3 code. Numerical analyses carried out or the Purdue University rod bundle tests showed that the results obtained using the new correlation were in good agreement with the test data and represented an improvement over those obtained rom the correlation. Numerical analyses carried out or the void proile tests in the THTF at ORNL showed that the new correlation developed under low low atmospheric pressure air/water conditions was less accurate or low low high pressure steam/water conditions. The distribution parameter o the new correlation needs improvements in order to scale the pressure eects appropriately. REFERENCES. Ohnuki, A., Akimoto, H. and Sudo, Y., Flow pattern and its transition in gas-liquid two-phase low along large vertical pipe, Proc. 2nd Int. Con. Multiphase Flow, Vol. 3, pp.7-23 (995). 2. Ohnuki, A. and Akimoto, H., An experimental study on developing air-water two-phase low along a large vertical pipe: eect o air injection method, International Journal o Multiphase Flow, 22, pp.43-54 (996). 3. Paranjape, S., Ishii, M., and Hibiki, T., Modeling and measurement o interacial area concentration in two-phase low, Nuclear Engineering and Design, 24, pp.2329-2337 (2). 4. Clark, C., Griiths, M., Chen, S.W., Hibiki, T., Ishii, M., Kinoshita, I., and Yoshida, Y., Experimental study o void raction in a 8 x 8 rod bundle at low pressure and low liquid low conditions, International Journal o Multiphase Flow, 62, pp.87- (24). 5. Clark, C., Griiths, M., Chen, S.W., Hibiki, T., Ishii, M., Ozaki, T., Kinoshita, I., and Yoshida, Y., Drit-lux correlation or rod bundle geometries, International Journal o Heat and Fluid Flow, 48, pp.-4 (24). 6. Hibiki, T. and Ishii, M., One-dimensional drit-lux model or two-phase low in a large diameter pipe, International Journal o Heat and Mass Transer, 46, pp.773-79, (23). 7. RELAP5/MOD3.3, Code Manual Volume IV: Models and Correlations, NUREG/CR- 5535/Rev, Prepared or the Division o Systems Research, Oice o Nuclear Regulatory Research. USNRC by Inormation Systems Laboratories, Inc., Rockville, Maryland (2). 8. Felde, D. K. et al., Facility Description THTF MOD3 ORNL PWR BDHT Separate-Eects Program, NUREG/CR-264 (982). 9. Anklam, T. M. et al., Experimental Investigations o Uncovered-Bundle Heat Transer and Two-Phase Mixture-Level Swell Under High-Pressure Low-Heat-Flux Conditions, NUREG/CR-2456 (982).. Zuber, N. and Findlay, J. A., Average Volumetric Concentration in Two-Phase Flow Systems, Journal o Heat Transer, 87, pp.453-468 (965).. Ozaki, T., Suzuki, R., Mashiko, H. and Hibiki, T., Development o drit-lux model based on 8 x 8 BWR rod bundle geometry experiments under prototypic temperature and pressure conditions, Journal o Nuclear Science and Technology, 5, pp.563-58 (23). 2. Ishii, M., One-Dimensional Drit-lux Model and Constitutive Equations or Relative Motion between Phases in Various Two-phase Flow Regimes, ANL-77-47, USA (977). 3. Chexal, B. and Lellouche, G., A Full-Range Drit-Flux Correlation or Vertical Flows (Revision ), Electric Power Research Institute. -NP-3989-SR (986). NURETH-6, Chicago, IL, August 3-September 4, 25 42