- Email: [email protected]
Downcomer boiling phenomena during the reflood phase of a large-break LOCA for the APR1400
Available online at www.sciencedirect.com
Nuclear Engineering and Design 238 (2008) 2064–2074
Downcomer boiling phenomena during the reflood phase of a large-break LOCA for the APR1400 B.J. Yun ∗ , D.J. Euh, C.-H. Song Korea Atomic Energy Research Institute, 1045 Daedeok-daero, Dukjin-dong 150, Yuseong-gu, Daejeon 305-353, Republic of Korea Received 29 March 2007; received in revised form 11 October 2007; accepted 19 October 2007
Abstract Downcomer boiling phenomena in a conventional pressurized water reactor has an important effect on the transient behavior of a postulated large-break LOCA (LBLOCA), because it can degrade the hydraulic head of the coolant in the downcomer and consequently affect the reflood flow rate for a core cooling. To investigate the thermal hydraulic behavior in the downcomer region, a test program for a downcomer boiling (DOBO) is being progressed for the reflood phase of a postulated LBLOCA. Test facility was designed as a one side heated rectangular channel which adopts a full-pressure, full-height, and full-size downcomer-gap approach, but with the circumferential length reduced 47.08-fold. The test was performed by dividing it into two-phases: (I) visual observation and acquisition of the global two-phase flow parameters and (II(a)) measurement of the local bubble flow parameters on the measuring planes along five elevations. In the present paper, the test results of Phase-I and a part of Phase-II(a) were introduced. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Downcomer boiling phenomena in a conventional pressurized water reactor has an important effect on the transient behavior of a postulated large-break LOCA (LBLOCA), because it can degrade the hydraulic head of the coolant in the downcomer, and consequently affect the reflood flow rate for a core cooling and finally result in a failure of the nuclear fuel rods. Safety analysis using the RELAP5 code showed that the boiling phenomena occurs significantly in the downcomer region of the Korean advanced pressurized water reactor APR (advanced power reactor) 1400 which adopts direct vessel injection (DVI) nozzles instead of the conventional cold leg injection (CLI) nozzles (Cho et al., 2002), and it affects the core reflood flow rate in the reflood phase of a postulated LBLOCA. Fig. 1 shows the major thermal hydraulics phenomena which are observed from the RELAP5 in the LBLOCA reflood phase of the APR1400. In
∗
Corresponding author. Tel.: +82 42 868 8756; fax: +82 42 868 8362. E-mail address: [email protected] (B.J. Yun).
0029-5493/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.nucengdes.2007.10.024
contrast, a downcomer boiling was not observed in the TRACM results. This difference between the simulations was mainly attributable to the adoption of different thermal hydraulics models for the downcomer, especially in the interfacial friction models: RELAP5 uses an interfacial friction model based on the drift flux, whereas TRAC-M uses a modified Blasius friction correlation (Lee et al., 2002). It shows that the thermal hydraulic models of the existing best extimate (BE) codes should be evaluated for the successful application to the downcomer boiling phenomena. The downcomer boiling phenomena was not an important issue in previous cold-leg injection systems, and thus there have been few studies on a downcomer boiling. Sudo and Akimoto (1982) performed a separate effect test on downcomer boiling phenomena. The test was performed using a slab geometry, in which the channel gap and height were 0.2 and 6.5 m, respectively. Also it was performed in a transient condition, and it showed the possibility of a degradation of the hydrostatic head of the coolant due to a downcomer boiling. However, the downcomer gap of the test facility was smaller and its wall was thinner than those of the APR1400, and thus it is difficult to directly extrapolate their test results to the APR1400. In addition, local
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Nomenclature a Dh g G h H l P q T u W x¯
area (m2 ) hydraulic diameter (m) gravity constant (m/s2 ) mass flux collapsed water level measured by DP transmitters (m) distance between the DP transmitter taps (m) length (m) pressure (kPa) heat flux (kW/m2 ) temperature (◦ C) velocity (m/s) mass flow rate (kg/s) void fraction weighted average value of x operator of area averaging
Greek letters α local void fraction μ viscosity (N-s/m2 ) ρ density (kg/m3 ) σ surface tension (N/m) Subscripts b bubble f liquid phase g steam phase R Ratio sys system
2065
a slab channel which preserves the wall thickness and the gap size of a downcomer with those of the APR1400. In the test, the simulator for the reactor vessel wall was heated up to 300 ◦ C before the test was started. From the test, the heat release rate from the downcomer wall to the fluid and the local liquid velocity at the top of a test channel were obtained. However, his test channel was limited to 1.5 m in height and the system pressure was maintained at an atmospheric pressure during the test. And thus, the effects of the height and system pressure were not fully investigated. The difficulty in separately evaluating the thermal hydraulic models related to a downcomer boiling and developing the relevant physical models have resulted in a separate effect-test program for a downcomer boiling by KAERI (Korea Atomic Energy Research Institute). The test program is being performed by using a two-stage approach: (I) visual observation and acquisition of the global two-phase flow parameters and (II) measurement of the local two-phase flow parameters. The Phase-I test has been completed and the Phase-II test is still being progressed. The Phase-II test was initiated to acquire information on the internal flow structure of a two-phase flow which is required for an evaluation and the development of the thermal hydraulics models for the best estimation codes. The Phase-II test is divided into Phase-II(a) and Phase-II(b). The former is for a measurement of the local bubble parameters and the latter is for the local liquid parameters. The flow conditions of the Phase-II follows those of the test matrix of Phase-I. In the present paper, the results of the Phase-I and a part of the Phase-II(a) tests were introduced. 2. Experimental facility
two-phase flow parameters such as the local void fraction and the interfacial friction factor, which provide information on an internal flow structure, were not measured in the test. Recently, Lee (2004) performed a downcomer-boiling test for the APR1400 in the transient flow condition. He also used
The DOBO (downcomer boiling) test facility is designed to simulate the downcomer region below a cold leg in the late reflood phase of a postulated LBLOCA, in which an ECC injection from the SIT is terminated. During this period, most of the thermal hydraulic parameters in the primary system reach quasi-steady-state conditions, and hence the DOBO facility is designed for a steady-state operation. 2.1. Scaling approach
Fig. 1. Major thermal hydraulic phenomena in the primarily system of the APR1400 under the LBLOCA reflood phase.
The design of the DOBO facility was essentially based on a volume scaling law (Nahavandi et al., 1979), which preserves the pressure, temperature, and height. In the standard volume scaling law, the length scale in the lateral direction is determined by the square root of the scaled area ratio. However, the size of the downcomer gap in the DOBO facility is maintained at that of the APR1400, whereas the circumferential length of the downcomer wall is reduced 47.08-fold. It should be noted that information on the multidimensional phenomena in the lateral direction can be lost when using a volume scaling law because it is derived based on the assumption of a one-dimensional vertical flow. However, the present scaling law improves this weak point because one of the two lateral coordinates was considered in the design.
2066
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 2. Geometrical comparison of reactor downcomer of the APR1400 and test section of the DOBO facility.
The approach of the present scaling law has the major characteristics of the volume scaling law, such as a preservation of the velocity scale, heat flux, and hydrostatic head of the water column. In addition to these, the present scaling makes it possible to preserve some important local phenomena such as the axial void-fraction distribution, the thickness of the bubbly boundary, and the degree of subcooling of the ECC water. However, the design of the DVI nozzle and its location does not follow the scaling law because it was found that the ECC water injected via DVI nozzles falls down along the core barrel wall in the form of a liquid film (Yun et al., 2002). Therefore, its design focused on a regeneration of the water film along the wall. For this, the diameter of the DVI nozzle of the DOBO facility was enlarged when compared to the 47.08-fold reduction for the APR1400, and it is intruded into the test section.
The major scaling ratios of the DOBO facility are summarized in Table 1, and the detailed geometrical design of the test section is compared with that of the APR1400 in Fig. 2. 2.2. Fluid system The DOBO facility consists of a test section, a steam–water separator, a condenser, a heat exchanger, a drain pump, a storage tank, an air injection and ventilation system, a pre-heater and an injection pump, as shown in Fig. 3. The maximum operational pressure is 500 kPa. The test section is designed with a slab geometry that simulates the high-temperature wall of a reactor vessel from the cold leg to the flow skirt. However, its length was extended by 1.3 m to allow for an installation of a single DVI nozzle at the top of the test section and to drain water
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074 Table 1 Scaling ratio of the DOBO facility Parameter
Elevation Gap size ratio Width ratio Area ratio Volume ratio Velocity ratio Flow rate ratio Gravity ratio Pressure ratio Temperature ratio
Scaling law Scaling ratio
DOBO
1 1 lR aR aR 1 aR 1 1 1
1 1 1/47.08 1/47.08 1/47.08 1 1/47.08 1 1 1
at the bottom—but there is no heat input to this region. The width, depth, and total height of the test section are 0.3, 0.25, and 6.4 m, respectively. Only one of the four side walls (which has a length of 5.1 m) simulates a high-temperature reactor vessel wall, and thus the heat transfer area becomes 1.53 m2 . The heat is generated by 207 cartridge heaters that are inserted in the wall. The maximum available heat flux is 100 kW/m2 , which corresponds to the maximum anticipated value at the termination of an ECC injection flow from the SIT, and this is controlled by
2067
a silicon-controlled rectifier unit. Glass windows are installed on one adjacent and opposite walls of the heated wall to allow for a visual observation. The separator is connected to the outlet pipe of the test section for the phase separation of a two-phase flow. The separated water and steam phases are respectively cooled and condensed by two heat exchangers. The temperature of the water in the storage tank is controlled by the heat removal rates of these heat exchangers. The circulating water is injected into the test section at the top and drained at the bottom of the test section by two centrifugal pumps. The mass flow rate required for the DOBO facility during the reflood phase after emptying the SIT tanks is 1.3–1.4 kg/s. The injected water mass flow rate of the test facility is maximally 3.3 kg/s and it is controlled by a control valve. The water level in the test section can be maintained at a fixed value by controlling the drain flow rate at the bottom of the test section. The water can be heated up to 7 ◦ C by a pre-heater at the designed maximum flow rate. The temperature of the ECC water injected via the DVI nozzle can be maintained at a target value by controlling the temperature of the accumulated water in the storage tank and the applied heat of the pre-heater. The system pressure is controlled by an air injection and venting system.
Fig. 3. Schematics of fluid system of the DOBO facility.
2068
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 4. Instrumentation of test section.
2.3. Instrumentation Several kinds of commercially available instruments are installed for the boundary mass and energy flow rates. The location of each instrumentation is shown in Fig. 4. The mass flow rates of the circulating water are measured by two Coriolis meters that are installed at the inlet and outlet pipes of the test section. The estimated uncertainty for a measured mass flow is 0.3% of its read value. The system pressure is measured at the top and bottom of the test section by two SMART-type PTs (pressure transmitter). Seven SMART-type DP (differential pressure) transmitters (LT-1 to LT-7) are uniformly spaced along the two pressure impulse taps of the pressure transmitters for measuring both the water level and the axial void fraction. An additional wide-range DP (LT-8) is installed to check on the measured differential pressure of the seven DP transmitters and to control the water level at the downcomer. The estimated uncertainty of each DP and PT reading is 0.1% of FS (full scale). The fluid temperatures are measured by K-type thermo-couples, with their uncertainties estimated at 2.2 ◦ C. These are installed at the inlet and outlet pipes, and also at the top and bottom of the test section. The surface temperature of a heated wall (TW) is also measured by thermocouples as shown in Fig. 4. The applied power for the wall heating
is measured by a power meter, with a reading uncertainty of 0.35%. In addition to the above commercially available instruments, specially developed, 5-conductance probes are installed for a measurement of the local bubble parameters such as the void fraction and three-dimensional bubble velocity. The detailed design of a 5-conductance probe is shown in Fig. 5. Total of five sensors are configured in a probe in order to investigate the multidimensional bubble behavior in a highly swirling two-phase flow (Euh et al., 2005a). For the benchmark of a local void fraction and bubble velocity, a separate test was performed in an air–water flow condition. The uncertainty of the void fraction was found to be 6.1% from a benchmark test (Euh et al., 2005b). The bubble velocity is measured by the signals from the two-sensors located at the up- and downstreams in the configuration of the sensors. It is benchmarked with a photographic method by using a highspeed camera whose deviation is 3.01%. The test showed that a distortion of the measured bubble velocity due to the other three sensors is negligible up to 1.8 m/s of the bubble velocity. The traversing unit moves the probe in a two-dimensional direction by controlling the radial and rotational components. The probes are installed axially at five elevations of the test
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
2069
Table 2 Summary of the experimental condition Parameter
TECC (◦ C)
Psys (kPa)
WECC (kg/s)
q (kW/m2 )
R1 R2 R3 R4 R2-1a
110.1 110.2 109.6 109.5 110.7
162.8 161.4 166.5 170.8 158.5
1.22 1.16 1.20 1.20 1.33
50.2 69.7 82.1 91.1 70.5
wall in the test section. Data acquisition began after all of the parameters had reached a steady state. 4. Experiments and results Total of five tests were performed in the reflood flow condition, according to the test conditions summarized in Table 2. However, the actual mass flow rate of the R1-R4 tests was 10% lower than the ideally scaled one. The R2-1a is a test to measure the local bubble flow parameters, of which the general thermal hydraulic conditions are similar to the R2. 4.1. Global two-phase flow parameters Fig. 5. Details of the local 5-conductance probe.
section to investigate a propagation of the local bubble flow parameters. 3. Experimental conditions and procedure The experimental condition was obtained by applying the scaling ratios in Table 1 to the RELAP5 results (Lee and Oh, 2005). A single failure of a safety injection system was assumed in the RELAP5 calculation, and thus two of the four HPSIs (high pressure safety injections) were working. The test was performed from the condition at 250 s after the initiation of a LBLOCA, because the analysis results from RELAP5 showed that a downcomer boiling occurs during this time period. The reflood flow rate of the APR1400 was estimated at about 62.5 kg/s from RELAP5, and thus 1.33 kg/s of ECC water should be delivered to the test section. During this period, the system pressure at the top of the test section changed from 170 to 160 kPa. In the test, the ECC water injected via a DVI nozzle into the test section simulates the penetrating water from the cold-leg level to the lower downcomer region in the APR1400. Therefore, the temperature of the injected ECC water was maintained the same as that of the ECC water where the mixture level exists in the downcomer of the APR1400. RELAP5 estimated that the degree of a subcooling of the penetrated ECC water at this point was 15 ◦ C at 250 s, and it decreased by up to 3 ◦ C at the end of the event. At this time, the heat flux of the downcomer wall of the APR1400 changed from 75 to 50 kW/m2 . In the test, the heat flux was changed to cover this range. The test was started by injecting heated ECC water via a single DVI nozzle located at the top of the test section. The water mixture level was maintained above the top of the heated
In the test, ECC water was delivered from the top of the test section and drained at the bottom. It should be noted that bubbles generated by the heated wall flow upward, and thus countercurrent flows of the steam and water phases are expected. Fig. 6 shows an illustration of the characteristics of the downcomer boiling phenomena observed through the transparent windows of the test section (Yun et al., 2006). As seen in the figure, a distinct bubble boundary layer, which is a typical characteristic of the subcooled boiling flow, occurs in the lower and middle region and a bulk boiling is observed in the upper region. The detailed characteristics of the downcomer boiling can be explained region-by-region as follows. In the lower region of the test section, the void formation was negligible. The onset of a nucleate boiling (ONB point) occurred in this region, and the generated bubbles remained on the heated wall until their diameter increased to a few millimeters, at which point they were detached from the wall. After this point, the detached bubbles slid along the wall, and thus the thickness of the bubbly boundary layer gradually increased. Downstream of this region, the void fraction and the thickness of the bubbly boundary layer started to increase rapidly, which is the point of an onset of a void where a significant void fraction is developed. In the middle region of the test section, cap shaped bubbles which are generated by a coalescence of the small bubbles were accompanied with small bubbles even though the void fraction was less than a few percent, and these slid up the wall. The bubble boundary layer thickened rapidly as it traveled upward, with the thickness being changed due to a periodic swirling motion of the bubble clusters comprising of groups of small bubbles. The swirling motion from the heated wall to the cold-water core region was due to the unbalanced distributions of the phases and velocities of the two phases between the inside and outside of a bubble boundary layer. In the adjacent cold wall, small amounts
2070
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 7. Distribition of an axial average void fraction.
Fig. 6. Illustration of downcomer boiling phenomena observed in the DOBO facility.
of stagnant or flowing-downward bubbles were also observed periodically. The swirling motion of the high-temperature bubble clusters can enhance the interfacial heat transfer between the two phases. In the upper region, a well-mixed bulk boiling occurred that expanded toward the middle and bottom regions as the heat flux increased. The observed flow regime was a churn-turbulent flow. Bubble breakup – rather than a bubble coalescence – was dominant in this highly turbulent region. For a qualitative investigation, the distribution of an average axial void-fraction was obtained using DPs. Here, the average void fraction, α, is obtained from the measured collapsed level according to α = 1 −
h H
to LT-7. As shown in the figure, the void fraction was negligible below an elevation of 2.8 m, but it was significant above it for all the cases. This indicates that an increased heat flux resulted in an increase in the void fraction above 2.8 m. The overall average void fraction of the heated region was less than 9% for a given maximum heat flux condition (R4 case). This low void fraction is somewhat different from the result of the best-estimate code (e.g., the void fraction from the RELAP5 calculation was more than 40%, even in the middle region, Lee and Oh, 2005). The 9% void fraction resulted in a 9% reduction of the collapsed water level and then a 3% decrease in the reflood flow rate. This fact implies that a degradation of the hydrostatic head for a core reflooding is not so much significant as predicted by RELAP5, and that the thermal hydraulic models of the RELAP5 code related to a downcomer boiling should be improved. Fig. 8 shows the degree of a subcooling of the fluid at the bottom of the test section. In the test, the subcooling of the temperature of the injected ECC water was 1.5 to 5.0 ◦ C. The
(1)
Fig. 7 shows the axial distribution of the average void fraction according to the applied heat flux. The mixture level was maintained between the two pressure taps of LT-7, and thus the void fraction exceeded 0.4 at LT-7 for all the cases. The decrease in the void fraction of LT-7 with an increasing applied heat flux was due to the collapsed water level being controlled by a widerange level transmitter LT-8, which covers the range from LT-1
Fig. 8. Degree of subcooling of at the top and bottom of the test section.
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 9. Measurement points for the local bubble parameters.
figure also clearly shows that the ECC water did not reach a saturated temperature and that the degree of the liquid subcooling at the bottom of the test section was maintained in the range of 4.3–5.5 ◦ C. It should be noted that the temperatures of the injected ECC water for the R3 and R4 tests were higher than those for R1 and R2, which must have affected the degree of a subcooling at the bottom. 4.2. Local bubble flow parameters The local bubble parameters were measured by 5conductance probes at the 70.5 kW/m2 of a heat flux condition which corresponds to the R2-1a of Table 2. The experimental condition was similar to that of the R2 case of Phase-I, however, it should be noted that the injection flow rate via the DVI nozzle was increased by about 10% when compared to the R2 case, which is a closer initial condition to the safety analysis code results than the R2 case. Total of five planes are selected for the local measurements along the heated test section and each measuring plane was located at the midpoint between each DP tab. On each measuring plane, six measuring paths perpendicular to the heated wall were chosen (see Fig. 9). Among them, five paths (P1–P5) are located on a half of the measuring plane and the remainder (P6) is located on the other half of the plane for a check on the symmetry of the obtained data. The data was acquired at 25 points along each measuring path and thus measurements were performed at 150 points on each measuring plane. The data acquiring time of the 5-conductance probe was 30 s with a 50 kHz sampling rate at each point. Fig. 10 shows the propagation of the local void fraction along the test section. At the lower elevation where a boiling
2071
Fig. 10. Propagations of local void fraction in the lower downcomer.
is initiated, the steam is concentrated near the heated wall. As the generated steam flows upward, the void profile becomes wide and a distinct bubbly boundary layer was found along test channel. The distributions of the local void fraction at the three lowest elevations show that the local void fraction near the sidewall is higher than that of the center. It is mainly due to the unavoidable sidewall effect of the present test. In the corner between the heated surface and sidewall, the liquid velocity must be lower than the center of the channel due to the increased wall shear stress. The contour plot of the surface temperature of the heated wall confirms this fact as well as shown in Fig. 11. As in the figure, the surface temperature of the sidewall is higher than that of the center at lower elevations. In contrast to these cases, at the two highest high void fraction regions, a peaking of the void profile moves to the central
Fig. 11. Temperature distribution of the heated wall.
2072
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 13. Evaluation of average interfacial friction models of existing BE codes.
velocity is very small or negative near the opposite wall of the heated surface in the two highest region, since the upward bubble motion is suppressed by the downward liquid flow. The ECC water that is falling down along the side and opposite walls of the heated surface is one of the reasons. 4.3. Evaluation of the interfacial friction models
Fig. 12. Propagations of local bubble velocity in the lower downcomer.
region of a measuring plane. This trend reflects the visual observation well, in that the bubbly boundary layer thickness increases rapidly as the elevation becomes higher. The surface temperature of the heated wall increases as it goes downward as seen in Fig. 11. It is mainly resulted by an increased saturation temperature of the fluid as it goes downward. The average void fraction can be obtained from the distribution of the local void fraction. Fig. 7 shows a comparison of the average void fractions obtained from the local void fraction (R2-1a) and DPs (R2). It shows that the average void fractions coincide well with each other, and confirms that the R2-1a and R2 tests are repeatable and the present measurement method and the acquired data of the local void fraction are reliable. Fig. 12 shows the contour plot of a local three-dimensional bubble velocity obtained from a 5-conductance probe. Its comparison with the local void fraction shows that the magnitude of the local bubble velocity follows that of the local void fraction. Moreover, the main direction of the bubble motion is an axial one, and the lateral motion of the bubbles from the heated wall to the opposite cold wall is not so significant. It means that the periodic swirling motion of the bubble cluster, that was observed visually, has a large time period. In addition to this, the bubble
The average interfacial friction coefficient was obtained from R2-1a for an evaluation of the interfacial friction models of the existing safety analysis codes. In the test, the interfacial friction factor is determined by simplified momentum equations of each phase under the assumptions that the flow is one-dimensional and the wall friction is neglected as follows: Ci,avg =
α (1 − α)(ρf − ρg )g 2
(Vg − Vf )
(2)
Here, Vg is obtained by an area averaging of the local axial bubble velocity with a local void fraction. And, Vf is calculated from the injection mass flow rate at a DVI nozzle. The observed flow regimes of the downcomer boiling are a bubbly flow with accompanying cap bubbles and a churnturbulent flow, and thus the interfacial friction models for these flow regimes of the existing BE codes were evaluated against the present data. Table 3 shows a summary of the interfacial frictional models of BE codes. Fig. 13 shows a comparison of the interfacial friction models adopted in the systems analysis codes with the present data. As seen in the figure, most of the models are not satisfactory. TRAC-original (Spore et al., 1993) model was developed based on the bubble drag of each bubble and the profile of a slip velocity. However, it overestimates the interfacial friction coefficients by five times when compared with the present data. It means that the original TRAC model will overestimate the void fraction. In the case of TRAC, the modified Blasius model was recommended firstly for a better prediction of the UPTF test, and then it was also used for the downcomer boiling model of the APR1400 (Lee et al., 2002). However, the present comparison shows that the Blasius model underestimates it too much. It
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
2073
Table 3 Summary of interfacial friction coefficients of existing BE codes
explains the reason for no downcomer boiling in the APR1400 case as stated in the introduction. RELAP5 (2001) adopts a drift-flux model for the interfacial friction coefficient. Drift-flux model considers the distributions of a void fraction and a slip velocity between two phases. In the present work, Kataoka and Ishii (1987) model was used as a drift-flux model because both the downcomers of the APR1400 and the DOBO facility are treated as a large pipe. It is known from the RELAP5 that the criterion for a large pipe is 0.08 m in diameter. As seen in Fig. 13, the RELAP5 model underestimates it by two times when compared with the present data. However, it is fairly good when compared with the TRAC cases. This comparison suggests that the higher void fraction prediction of the APR1400 is caused by other models related to a subcooled boiling. However, to check on the detailed subcooled boiling model in RELAP5, more sophisticated information is required such as the distribution of the water temperature, and so on. This is one of the motivations for the Phase-II(b) test which will be implemented as a series of the present work. All of these comparisons show that the current interfacial friction model should be improved for a better prediction capability of the existing BE codes.
5. Conclusions For an investigation of the downcomer boiling phenomena experimentally, a separate effect test was performed in the DOBO facility. Observations of the test revealed the occurrence of a countercurrent subcooled boiling flow, and the creation of a distinct bubble boundary layer whose thickness varied dramatically with the applied heat flux. The data showed that the channel average void fraction was small at the anticipated heat flux condition of the reflood phase of a LBLOCA, and thus the reduction of the hydraulic head for the core reflood phase was not too severe in the present test condition, with a subcooling margin of 4.3–5.5 ◦ C at the bottom of the test section. The local two-phase parameters such as the void fraction and bubble velocity were also measured by using a newly designed local 5-conductance probe. The local distributions, of various bubble parameters, reflect the visualization observation well. Using the data, an evaluation of the interfacial friction models of the existing system analysis codes was performed. It showed that most of the models for the interfacial friction coefficients should be improved for an enhancement of the prediction capability of the downcomer boiling phenomena.
2074
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
In further studies, the data base for the Phase-II(a) will be extended. In addition, a Phase-II(b) test for the local liquid velocity and the local fluid temperature will be undertaken by using specially designed instrumentation consisting of a local bi-directional flow tube and a thermocouple. And finally, thermal hydraulics models for the downcomer boiling phenomena will be developed by using the measured data. Acknowledgements The authors wish to express their deep appreciation to Mr. W.M. Park and Dr. H.K. Cho at KAERI and Mr. B.U. Bae at Seoul National University for their assistance of the experimental work. This study has been carried out under the nuclear R&D program by the Korean Ministry of Science and Technology. References Cho, S.J., Kim, B.S., Kang, M.G., Kim, H.G., 2002. The development of passive design features for the Korean next generation reactor. Nucl. Eng. Design 201, 259–271. Euh, D.J., Yun, B.J., Song, C.-H., 2005a. Measurement method of interfacial area concentration for highly fluctuating bubbly flow. KNS Autumn Meeting. Euh, D.J., Yun, B.J., Park, W.M., Song, C.-H., 2005b. Investigation of the Transport of the Bubble Parameters in Air/Water Flow Conditions, ICAPP05, Seoul, Korea.
Kataoka, I., Ishii, M., 1987. Drift-flux model for large diameter pipe and new correlation for pool void fraction. Int. J. Heat Mass Transfer 30 (9), 1927– 1939. Lee, D.W., 2004. An Experimental Study of Thermal-hydraulic Phenomena in the Downcomer with Direct Vessel Injection of APR1400 during LBLOCA Reflood Phase. Ph.D. Thesis. KAIST, Korea. Lee, S.W., Kim, H.G., Oh, S.J., 2002. The post-test analyses of CCTF C2-4 using TRAC-M and RELAP codes. In: Proceedings of the 3rd Korea–Japan Symposium on Nuclear Thermal Hydraulics and Safety (NTHAS3), Kyeongju, Korea. Lee, S.W., Oh, S.J., 2005. Experimental benchmarking and safety analyses of APR1400 large break LOCA scenario. In: Proceedings of the 5th International Congress on Advances in Nuclear Power Plants (ICAPP05), Seoul, Korea. Nahavandi, A.N., Castellana, F.S., Moradkhanian, E.N., 1979. Scaling laws for modeling nuclear reactor systems. Nucl. Sci. Eng. 72, 75– 83. RELAP5/Mod3.3 Code Manual, 2001. Vol. IV: Models and Correlations. Information Systems Laboratories Inc. Spore, J.W., et al., 1993. TRAC-PF1/Mod2 Volume I Theory Manual, NUREG/CR-5673. Los Alamos National Lab. Sudo, Y., Akimoto, H., 1982. Downcomer effective water head during reflood in postulated PWR LOCA. J. Nucl. Sci. Technol. 9 (1), 31–45. Yun, B.J., Euh, D.J., Song, C.-H., 2006. Investigation of the downcomer boiling phenomena during the reflood phase of a postulated large-break LOCA in the APR1400. Nucl. Technol. 156 (1), 56–68. Yun, B.J., Kwon, T.S., Park, W.M., Park, J.K., Song, C.-H., Euh, D.J., Chu, I.C., 2002. Direct ECC bypass phenomena in the MIDAS test facility during LBLOCA reflood phase. J. KNS 34 (5), 421–432.
Nuclear Engineering and Design 238 (2008) 2064–2074
Downcomer boiling phenomena during the reflood phase of a large-break LOCA for the APR1400 B.J. Yun ∗ , D.J. Euh, C.-H. Song Korea Atomic Energy Research Institute, 1045 Daedeok-daero, Dukjin-dong 150, Yuseong-gu, Daejeon 305-353, Republic of Korea Received 29 March 2007; received in revised form 11 October 2007; accepted 19 October 2007
Abstract Downcomer boiling phenomena in a conventional pressurized water reactor has an important effect on the transient behavior of a postulated large-break LOCA (LBLOCA), because it can degrade the hydraulic head of the coolant in the downcomer and consequently affect the reflood flow rate for a core cooling. To investigate the thermal hydraulic behavior in the downcomer region, a test program for a downcomer boiling (DOBO) is being progressed for the reflood phase of a postulated LBLOCA. Test facility was designed as a one side heated rectangular channel which adopts a full-pressure, full-height, and full-size downcomer-gap approach, but with the circumferential length reduced 47.08-fold. The test was performed by dividing it into two-phases: (I) visual observation and acquisition of the global two-phase flow parameters and (II(a)) measurement of the local bubble flow parameters on the measuring planes along five elevations. In the present paper, the test results of Phase-I and a part of Phase-II(a) were introduced. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Downcomer boiling phenomena in a conventional pressurized water reactor has an important effect on the transient behavior of a postulated large-break LOCA (LBLOCA), because it can degrade the hydraulic head of the coolant in the downcomer, and consequently affect the reflood flow rate for a core cooling and finally result in a failure of the nuclear fuel rods. Safety analysis using the RELAP5 code showed that the boiling phenomena occurs significantly in the downcomer region of the Korean advanced pressurized water reactor APR (advanced power reactor) 1400 which adopts direct vessel injection (DVI) nozzles instead of the conventional cold leg injection (CLI) nozzles (Cho et al., 2002), and it affects the core reflood flow rate in the reflood phase of a postulated LBLOCA. Fig. 1 shows the major thermal hydraulics phenomena which are observed from the RELAP5 in the LBLOCA reflood phase of the APR1400. In
∗
Corresponding author. Tel.: +82 42 868 8756; fax: +82 42 868 8362. E-mail address: [email protected] (B.J. Yun).
0029-5493/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.nucengdes.2007.10.024
contrast, a downcomer boiling was not observed in the TRACM results. This difference between the simulations was mainly attributable to the adoption of different thermal hydraulics models for the downcomer, especially in the interfacial friction models: RELAP5 uses an interfacial friction model based on the drift flux, whereas TRAC-M uses a modified Blasius friction correlation (Lee et al., 2002). It shows that the thermal hydraulic models of the existing best extimate (BE) codes should be evaluated for the successful application to the downcomer boiling phenomena. The downcomer boiling phenomena was not an important issue in previous cold-leg injection systems, and thus there have been few studies on a downcomer boiling. Sudo and Akimoto (1982) performed a separate effect test on downcomer boiling phenomena. The test was performed using a slab geometry, in which the channel gap and height were 0.2 and 6.5 m, respectively. Also it was performed in a transient condition, and it showed the possibility of a degradation of the hydrostatic head of the coolant due to a downcomer boiling. However, the downcomer gap of the test facility was smaller and its wall was thinner than those of the APR1400, and thus it is difficult to directly extrapolate their test results to the APR1400. In addition, local
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Nomenclature a Dh g G h H l P q T u W x¯
area (m2 ) hydraulic diameter (m) gravity constant (m/s2 ) mass flux collapsed water level measured by DP transmitters (m) distance between the DP transmitter taps (m) length (m) pressure (kPa) heat flux (kW/m2 ) temperature (◦ C) velocity (m/s) mass flow rate (kg/s) void fraction weighted average value of x operator of area averaging
Greek letters α local void fraction μ viscosity (N-s/m2 ) ρ density (kg/m3 ) σ surface tension (N/m) Subscripts b bubble f liquid phase g steam phase R Ratio sys system
2065
a slab channel which preserves the wall thickness and the gap size of a downcomer with those of the APR1400. In the test, the simulator for the reactor vessel wall was heated up to 300 ◦ C before the test was started. From the test, the heat release rate from the downcomer wall to the fluid and the local liquid velocity at the top of a test channel were obtained. However, his test channel was limited to 1.5 m in height and the system pressure was maintained at an atmospheric pressure during the test. And thus, the effects of the height and system pressure were not fully investigated. The difficulty in separately evaluating the thermal hydraulic models related to a downcomer boiling and developing the relevant physical models have resulted in a separate effect-test program for a downcomer boiling by KAERI (Korea Atomic Energy Research Institute). The test program is being performed by using a two-stage approach: (I) visual observation and acquisition of the global two-phase flow parameters and (II) measurement of the local two-phase flow parameters. The Phase-I test has been completed and the Phase-II test is still being progressed. The Phase-II test was initiated to acquire information on the internal flow structure of a two-phase flow which is required for an evaluation and the development of the thermal hydraulics models for the best estimation codes. The Phase-II test is divided into Phase-II(a) and Phase-II(b). The former is for a measurement of the local bubble parameters and the latter is for the local liquid parameters. The flow conditions of the Phase-II follows those of the test matrix of Phase-I. In the present paper, the results of the Phase-I and a part of the Phase-II(a) tests were introduced. 2. Experimental facility
two-phase flow parameters such as the local void fraction and the interfacial friction factor, which provide information on an internal flow structure, were not measured in the test. Recently, Lee (2004) performed a downcomer-boiling test for the APR1400 in the transient flow condition. He also used
The DOBO (downcomer boiling) test facility is designed to simulate the downcomer region below a cold leg in the late reflood phase of a postulated LBLOCA, in which an ECC injection from the SIT is terminated. During this period, most of the thermal hydraulic parameters in the primary system reach quasi-steady-state conditions, and hence the DOBO facility is designed for a steady-state operation. 2.1. Scaling approach
Fig. 1. Major thermal hydraulic phenomena in the primarily system of the APR1400 under the LBLOCA reflood phase.
The design of the DOBO facility was essentially based on a volume scaling law (Nahavandi et al., 1979), which preserves the pressure, temperature, and height. In the standard volume scaling law, the length scale in the lateral direction is determined by the square root of the scaled area ratio. However, the size of the downcomer gap in the DOBO facility is maintained at that of the APR1400, whereas the circumferential length of the downcomer wall is reduced 47.08-fold. It should be noted that information on the multidimensional phenomena in the lateral direction can be lost when using a volume scaling law because it is derived based on the assumption of a one-dimensional vertical flow. However, the present scaling law improves this weak point because one of the two lateral coordinates was considered in the design.
2066
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 2. Geometrical comparison of reactor downcomer of the APR1400 and test section of the DOBO facility.
The approach of the present scaling law has the major characteristics of the volume scaling law, such as a preservation of the velocity scale, heat flux, and hydrostatic head of the water column. In addition to these, the present scaling makes it possible to preserve some important local phenomena such as the axial void-fraction distribution, the thickness of the bubbly boundary, and the degree of subcooling of the ECC water. However, the design of the DVI nozzle and its location does not follow the scaling law because it was found that the ECC water injected via DVI nozzles falls down along the core barrel wall in the form of a liquid film (Yun et al., 2002). Therefore, its design focused on a regeneration of the water film along the wall. For this, the diameter of the DVI nozzle of the DOBO facility was enlarged when compared to the 47.08-fold reduction for the APR1400, and it is intruded into the test section.
The major scaling ratios of the DOBO facility are summarized in Table 1, and the detailed geometrical design of the test section is compared with that of the APR1400 in Fig. 2. 2.2. Fluid system The DOBO facility consists of a test section, a steam–water separator, a condenser, a heat exchanger, a drain pump, a storage tank, an air injection and ventilation system, a pre-heater and an injection pump, as shown in Fig. 3. The maximum operational pressure is 500 kPa. The test section is designed with a slab geometry that simulates the high-temperature wall of a reactor vessel from the cold leg to the flow skirt. However, its length was extended by 1.3 m to allow for an installation of a single DVI nozzle at the top of the test section and to drain water
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074 Table 1 Scaling ratio of the DOBO facility Parameter
Elevation Gap size ratio Width ratio Area ratio Volume ratio Velocity ratio Flow rate ratio Gravity ratio Pressure ratio Temperature ratio
Scaling law Scaling ratio
DOBO
1 1 lR aR aR 1 aR 1 1 1
1 1 1/47.08 1/47.08 1/47.08 1 1/47.08 1 1 1
at the bottom—but there is no heat input to this region. The width, depth, and total height of the test section are 0.3, 0.25, and 6.4 m, respectively. Only one of the four side walls (which has a length of 5.1 m) simulates a high-temperature reactor vessel wall, and thus the heat transfer area becomes 1.53 m2 . The heat is generated by 207 cartridge heaters that are inserted in the wall. The maximum available heat flux is 100 kW/m2 , which corresponds to the maximum anticipated value at the termination of an ECC injection flow from the SIT, and this is controlled by
2067
a silicon-controlled rectifier unit. Glass windows are installed on one adjacent and opposite walls of the heated wall to allow for a visual observation. The separator is connected to the outlet pipe of the test section for the phase separation of a two-phase flow. The separated water and steam phases are respectively cooled and condensed by two heat exchangers. The temperature of the water in the storage tank is controlled by the heat removal rates of these heat exchangers. The circulating water is injected into the test section at the top and drained at the bottom of the test section by two centrifugal pumps. The mass flow rate required for the DOBO facility during the reflood phase after emptying the SIT tanks is 1.3–1.4 kg/s. The injected water mass flow rate of the test facility is maximally 3.3 kg/s and it is controlled by a control valve. The water level in the test section can be maintained at a fixed value by controlling the drain flow rate at the bottom of the test section. The water can be heated up to 7 ◦ C by a pre-heater at the designed maximum flow rate. The temperature of the ECC water injected via the DVI nozzle can be maintained at a target value by controlling the temperature of the accumulated water in the storage tank and the applied heat of the pre-heater. The system pressure is controlled by an air injection and venting system.
Fig. 3. Schematics of fluid system of the DOBO facility.
2068
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 4. Instrumentation of test section.
2.3. Instrumentation Several kinds of commercially available instruments are installed for the boundary mass and energy flow rates. The location of each instrumentation is shown in Fig. 4. The mass flow rates of the circulating water are measured by two Coriolis meters that are installed at the inlet and outlet pipes of the test section. The estimated uncertainty for a measured mass flow is 0.3% of its read value. The system pressure is measured at the top and bottom of the test section by two SMART-type PTs (pressure transmitter). Seven SMART-type DP (differential pressure) transmitters (LT-1 to LT-7) are uniformly spaced along the two pressure impulse taps of the pressure transmitters for measuring both the water level and the axial void fraction. An additional wide-range DP (LT-8) is installed to check on the measured differential pressure of the seven DP transmitters and to control the water level at the downcomer. The estimated uncertainty of each DP and PT reading is 0.1% of FS (full scale). The fluid temperatures are measured by K-type thermo-couples, with their uncertainties estimated at 2.2 ◦ C. These are installed at the inlet and outlet pipes, and also at the top and bottom of the test section. The surface temperature of a heated wall (TW) is also measured by thermocouples as shown in Fig. 4. The applied power for the wall heating
is measured by a power meter, with a reading uncertainty of 0.35%. In addition to the above commercially available instruments, specially developed, 5-conductance probes are installed for a measurement of the local bubble parameters such as the void fraction and three-dimensional bubble velocity. The detailed design of a 5-conductance probe is shown in Fig. 5. Total of five sensors are configured in a probe in order to investigate the multidimensional bubble behavior in a highly swirling two-phase flow (Euh et al., 2005a). For the benchmark of a local void fraction and bubble velocity, a separate test was performed in an air–water flow condition. The uncertainty of the void fraction was found to be 6.1% from a benchmark test (Euh et al., 2005b). The bubble velocity is measured by the signals from the two-sensors located at the up- and downstreams in the configuration of the sensors. It is benchmarked with a photographic method by using a highspeed camera whose deviation is 3.01%. The test showed that a distortion of the measured bubble velocity due to the other three sensors is negligible up to 1.8 m/s of the bubble velocity. The traversing unit moves the probe in a two-dimensional direction by controlling the radial and rotational components. The probes are installed axially at five elevations of the test
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
2069
Table 2 Summary of the experimental condition Parameter
TECC (◦ C)
Psys (kPa)
WECC (kg/s)
q (kW/m2 )
R1 R2 R3 R4 R2-1a
110.1 110.2 109.6 109.5 110.7
162.8 161.4 166.5 170.8 158.5
1.22 1.16 1.20 1.20 1.33
50.2 69.7 82.1 91.1 70.5
wall in the test section. Data acquisition began after all of the parameters had reached a steady state. 4. Experiments and results Total of five tests were performed in the reflood flow condition, according to the test conditions summarized in Table 2. However, the actual mass flow rate of the R1-R4 tests was 10% lower than the ideally scaled one. The R2-1a is a test to measure the local bubble flow parameters, of which the general thermal hydraulic conditions are similar to the R2. 4.1. Global two-phase flow parameters Fig. 5. Details of the local 5-conductance probe.
section to investigate a propagation of the local bubble flow parameters. 3. Experimental conditions and procedure The experimental condition was obtained by applying the scaling ratios in Table 1 to the RELAP5 results (Lee and Oh, 2005). A single failure of a safety injection system was assumed in the RELAP5 calculation, and thus two of the four HPSIs (high pressure safety injections) were working. The test was performed from the condition at 250 s after the initiation of a LBLOCA, because the analysis results from RELAP5 showed that a downcomer boiling occurs during this time period. The reflood flow rate of the APR1400 was estimated at about 62.5 kg/s from RELAP5, and thus 1.33 kg/s of ECC water should be delivered to the test section. During this period, the system pressure at the top of the test section changed from 170 to 160 kPa. In the test, the ECC water injected via a DVI nozzle into the test section simulates the penetrating water from the cold-leg level to the lower downcomer region in the APR1400. Therefore, the temperature of the injected ECC water was maintained the same as that of the ECC water where the mixture level exists in the downcomer of the APR1400. RELAP5 estimated that the degree of a subcooling of the penetrated ECC water at this point was 15 ◦ C at 250 s, and it decreased by up to 3 ◦ C at the end of the event. At this time, the heat flux of the downcomer wall of the APR1400 changed from 75 to 50 kW/m2 . In the test, the heat flux was changed to cover this range. The test was started by injecting heated ECC water via a single DVI nozzle located at the top of the test section. The water mixture level was maintained above the top of the heated
In the test, ECC water was delivered from the top of the test section and drained at the bottom. It should be noted that bubbles generated by the heated wall flow upward, and thus countercurrent flows of the steam and water phases are expected. Fig. 6 shows an illustration of the characteristics of the downcomer boiling phenomena observed through the transparent windows of the test section (Yun et al., 2006). As seen in the figure, a distinct bubble boundary layer, which is a typical characteristic of the subcooled boiling flow, occurs in the lower and middle region and a bulk boiling is observed in the upper region. The detailed characteristics of the downcomer boiling can be explained region-by-region as follows. In the lower region of the test section, the void formation was negligible. The onset of a nucleate boiling (ONB point) occurred in this region, and the generated bubbles remained on the heated wall until their diameter increased to a few millimeters, at which point they were detached from the wall. After this point, the detached bubbles slid along the wall, and thus the thickness of the bubbly boundary layer gradually increased. Downstream of this region, the void fraction and the thickness of the bubbly boundary layer started to increase rapidly, which is the point of an onset of a void where a significant void fraction is developed. In the middle region of the test section, cap shaped bubbles which are generated by a coalescence of the small bubbles were accompanied with small bubbles even though the void fraction was less than a few percent, and these slid up the wall. The bubble boundary layer thickened rapidly as it traveled upward, with the thickness being changed due to a periodic swirling motion of the bubble clusters comprising of groups of small bubbles. The swirling motion from the heated wall to the cold-water core region was due to the unbalanced distributions of the phases and velocities of the two phases between the inside and outside of a bubble boundary layer. In the adjacent cold wall, small amounts
2070
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 7. Distribition of an axial average void fraction.
Fig. 6. Illustration of downcomer boiling phenomena observed in the DOBO facility.
of stagnant or flowing-downward bubbles were also observed periodically. The swirling motion of the high-temperature bubble clusters can enhance the interfacial heat transfer between the two phases. In the upper region, a well-mixed bulk boiling occurred that expanded toward the middle and bottom regions as the heat flux increased. The observed flow regime was a churn-turbulent flow. Bubble breakup – rather than a bubble coalescence – was dominant in this highly turbulent region. For a qualitative investigation, the distribution of an average axial void-fraction was obtained using DPs. Here, the average void fraction, α, is obtained from the measured collapsed level according to α = 1 −
h H
to LT-7. As shown in the figure, the void fraction was negligible below an elevation of 2.8 m, but it was significant above it for all the cases. This indicates that an increased heat flux resulted in an increase in the void fraction above 2.8 m. The overall average void fraction of the heated region was less than 9% for a given maximum heat flux condition (R4 case). This low void fraction is somewhat different from the result of the best-estimate code (e.g., the void fraction from the RELAP5 calculation was more than 40%, even in the middle region, Lee and Oh, 2005). The 9% void fraction resulted in a 9% reduction of the collapsed water level and then a 3% decrease in the reflood flow rate. This fact implies that a degradation of the hydrostatic head for a core reflooding is not so much significant as predicted by RELAP5, and that the thermal hydraulic models of the RELAP5 code related to a downcomer boiling should be improved. Fig. 8 shows the degree of a subcooling of the fluid at the bottom of the test section. In the test, the subcooling of the temperature of the injected ECC water was 1.5 to 5.0 ◦ C. The
(1)
Fig. 7 shows the axial distribution of the average void fraction according to the applied heat flux. The mixture level was maintained between the two pressure taps of LT-7, and thus the void fraction exceeded 0.4 at LT-7 for all the cases. The decrease in the void fraction of LT-7 with an increasing applied heat flux was due to the collapsed water level being controlled by a widerange level transmitter LT-8, which covers the range from LT-1
Fig. 8. Degree of subcooling of at the top and bottom of the test section.
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 9. Measurement points for the local bubble parameters.
figure also clearly shows that the ECC water did not reach a saturated temperature and that the degree of the liquid subcooling at the bottom of the test section was maintained in the range of 4.3–5.5 ◦ C. It should be noted that the temperatures of the injected ECC water for the R3 and R4 tests were higher than those for R1 and R2, which must have affected the degree of a subcooling at the bottom. 4.2. Local bubble flow parameters The local bubble parameters were measured by 5conductance probes at the 70.5 kW/m2 of a heat flux condition which corresponds to the R2-1a of Table 2. The experimental condition was similar to that of the R2 case of Phase-I, however, it should be noted that the injection flow rate via the DVI nozzle was increased by about 10% when compared to the R2 case, which is a closer initial condition to the safety analysis code results than the R2 case. Total of five planes are selected for the local measurements along the heated test section and each measuring plane was located at the midpoint between each DP tab. On each measuring plane, six measuring paths perpendicular to the heated wall were chosen (see Fig. 9). Among them, five paths (P1–P5) are located on a half of the measuring plane and the remainder (P6) is located on the other half of the plane for a check on the symmetry of the obtained data. The data was acquired at 25 points along each measuring path and thus measurements were performed at 150 points on each measuring plane. The data acquiring time of the 5-conductance probe was 30 s with a 50 kHz sampling rate at each point. Fig. 10 shows the propagation of the local void fraction along the test section. At the lower elevation where a boiling
2071
Fig. 10. Propagations of local void fraction in the lower downcomer.
is initiated, the steam is concentrated near the heated wall. As the generated steam flows upward, the void profile becomes wide and a distinct bubbly boundary layer was found along test channel. The distributions of the local void fraction at the three lowest elevations show that the local void fraction near the sidewall is higher than that of the center. It is mainly due to the unavoidable sidewall effect of the present test. In the corner between the heated surface and sidewall, the liquid velocity must be lower than the center of the channel due to the increased wall shear stress. The contour plot of the surface temperature of the heated wall confirms this fact as well as shown in Fig. 11. As in the figure, the surface temperature of the sidewall is higher than that of the center at lower elevations. In contrast to these cases, at the two highest high void fraction regions, a peaking of the void profile moves to the central
Fig. 11. Temperature distribution of the heated wall.
2072
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
Fig. 13. Evaluation of average interfacial friction models of existing BE codes.
velocity is very small or negative near the opposite wall of the heated surface in the two highest region, since the upward bubble motion is suppressed by the downward liquid flow. The ECC water that is falling down along the side and opposite walls of the heated surface is one of the reasons. 4.3. Evaluation of the interfacial friction models
Fig. 12. Propagations of local bubble velocity in the lower downcomer.
region of a measuring plane. This trend reflects the visual observation well, in that the bubbly boundary layer thickness increases rapidly as the elevation becomes higher. The surface temperature of the heated wall increases as it goes downward as seen in Fig. 11. It is mainly resulted by an increased saturation temperature of the fluid as it goes downward. The average void fraction can be obtained from the distribution of the local void fraction. Fig. 7 shows a comparison of the average void fractions obtained from the local void fraction (R2-1a) and DPs (R2). It shows that the average void fractions coincide well with each other, and confirms that the R2-1a and R2 tests are repeatable and the present measurement method and the acquired data of the local void fraction are reliable. Fig. 12 shows the contour plot of a local three-dimensional bubble velocity obtained from a 5-conductance probe. Its comparison with the local void fraction shows that the magnitude of the local bubble velocity follows that of the local void fraction. Moreover, the main direction of the bubble motion is an axial one, and the lateral motion of the bubbles from the heated wall to the opposite cold wall is not so significant. It means that the periodic swirling motion of the bubble cluster, that was observed visually, has a large time period. In addition to this, the bubble
The average interfacial friction coefficient was obtained from R2-1a for an evaluation of the interfacial friction models of the existing safety analysis codes. In the test, the interfacial friction factor is determined by simplified momentum equations of each phase under the assumptions that the flow is one-dimensional and the wall friction is neglected as follows: Ci,avg =
α (1 − α)(ρf − ρg )g 2
(Vg − Vf )
(2)
Here, Vg is obtained by an area averaging of the local axial bubble velocity with a local void fraction. And, Vf is calculated from the injection mass flow rate at a DVI nozzle. The observed flow regimes of the downcomer boiling are a bubbly flow with accompanying cap bubbles and a churnturbulent flow, and thus the interfacial friction models for these flow regimes of the existing BE codes were evaluated against the present data. Table 3 shows a summary of the interfacial frictional models of BE codes. Fig. 13 shows a comparison of the interfacial friction models adopted in the systems analysis codes with the present data. As seen in the figure, most of the models are not satisfactory. TRAC-original (Spore et al., 1993) model was developed based on the bubble drag of each bubble and the profile of a slip velocity. However, it overestimates the interfacial friction coefficients by five times when compared with the present data. It means that the original TRAC model will overestimate the void fraction. In the case of TRAC, the modified Blasius model was recommended firstly for a better prediction of the UPTF test, and then it was also used for the downcomer boiling model of the APR1400 (Lee et al., 2002). However, the present comparison shows that the Blasius model underestimates it too much. It
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
2073
Table 3 Summary of interfacial friction coefficients of existing BE codes
explains the reason for no downcomer boiling in the APR1400 case as stated in the introduction. RELAP5 (2001) adopts a drift-flux model for the interfacial friction coefficient. Drift-flux model considers the distributions of a void fraction and a slip velocity between two phases. In the present work, Kataoka and Ishii (1987) model was used as a drift-flux model because both the downcomers of the APR1400 and the DOBO facility are treated as a large pipe. It is known from the RELAP5 that the criterion for a large pipe is 0.08 m in diameter. As seen in Fig. 13, the RELAP5 model underestimates it by two times when compared with the present data. However, it is fairly good when compared with the TRAC cases. This comparison suggests that the higher void fraction prediction of the APR1400 is caused by other models related to a subcooled boiling. However, to check on the detailed subcooled boiling model in RELAP5, more sophisticated information is required such as the distribution of the water temperature, and so on. This is one of the motivations for the Phase-II(b) test which will be implemented as a series of the present work. All of these comparisons show that the current interfacial friction model should be improved for a better prediction capability of the existing BE codes.
5. Conclusions For an investigation of the downcomer boiling phenomena experimentally, a separate effect test was performed in the DOBO facility. Observations of the test revealed the occurrence of a countercurrent subcooled boiling flow, and the creation of a distinct bubble boundary layer whose thickness varied dramatically with the applied heat flux. The data showed that the channel average void fraction was small at the anticipated heat flux condition of the reflood phase of a LBLOCA, and thus the reduction of the hydraulic head for the core reflood phase was not too severe in the present test condition, with a subcooling margin of 4.3–5.5 ◦ C at the bottom of the test section. The local two-phase parameters such as the void fraction and bubble velocity were also measured by using a newly designed local 5-conductance probe. The local distributions, of various bubble parameters, reflect the visualization observation well. Using the data, an evaluation of the interfacial friction models of the existing system analysis codes was performed. It showed that most of the models for the interfacial friction coefficients should be improved for an enhancement of the prediction capability of the downcomer boiling phenomena.
2074
B.J. Yun et al. / Nuclear Engineering and Design 238 (2008) 2064–2074
In further studies, the data base for the Phase-II(a) will be extended. In addition, a Phase-II(b) test for the local liquid velocity and the local fluid temperature will be undertaken by using specially designed instrumentation consisting of a local bi-directional flow tube and a thermocouple. And finally, thermal hydraulics models for the downcomer boiling phenomena will be developed by using the measured data. Acknowledgements The authors wish to express their deep appreciation to Mr. W.M. Park and Dr. H.K. Cho at KAERI and Mr. B.U. Bae at Seoul National University for their assistance of the experimental work. This study has been carried out under the nuclear R&D program by the Korean Ministry of Science and Technology. References Cho, S.J., Kim, B.S., Kang, M.G., Kim, H.G., 2002. The development of passive design features for the Korean next generation reactor. Nucl. Eng. Design 201, 259–271. Euh, D.J., Yun, B.J., Song, C.-H., 2005a. Measurement method of interfacial area concentration for highly fluctuating bubbly flow. KNS Autumn Meeting. Euh, D.J., Yun, B.J., Park, W.M., Song, C.-H., 2005b. Investigation of the Transport of the Bubble Parameters in Air/Water Flow Conditions, ICAPP05, Seoul, Korea.
Kataoka, I., Ishii, M., 1987. Drift-flux model for large diameter pipe and new correlation for pool void fraction. Int. J. Heat Mass Transfer 30 (9), 1927– 1939. Lee, D.W., 2004. An Experimental Study of Thermal-hydraulic Phenomena in the Downcomer with Direct Vessel Injection of APR1400 during LBLOCA Reflood Phase. Ph.D. Thesis. KAIST, Korea. Lee, S.W., Kim, H.G., Oh, S.J., 2002. The post-test analyses of CCTF C2-4 using TRAC-M and RELAP codes. In: Proceedings of the 3rd Korea–Japan Symposium on Nuclear Thermal Hydraulics and Safety (NTHAS3), Kyeongju, Korea. Lee, S.W., Oh, S.J., 2005. Experimental benchmarking and safety analyses of APR1400 large break LOCA scenario. In: Proceedings of the 5th International Congress on Advances in Nuclear Power Plants (ICAPP05), Seoul, Korea. Nahavandi, A.N., Castellana, F.S., Moradkhanian, E.N., 1979. Scaling laws for modeling nuclear reactor systems. Nucl. Sci. Eng. 72, 75– 83. RELAP5/Mod3.3 Code Manual, 2001. Vol. IV: Models and Correlations. Information Systems Laboratories Inc. Spore, J.W., et al., 1993. TRAC-PF1/Mod2 Volume I Theory Manual, NUREG/CR-5673. Los Alamos National Lab. Sudo, Y., Akimoto, H., 1982. Downcomer effective water head during reflood in postulated PWR LOCA. J. Nucl. Sci. Technol. 9 (1), 31–45. Yun, B.J., Euh, D.J., Song, C.-H., 2006. Investigation of the downcomer boiling phenomena during the reflood phase of a postulated large-break LOCA in the APR1400. Nucl. Technol. 156 (1), 56–68. Yun, B.J., Kwon, T.S., Park, W.M., Park, J.K., Song, C.-H., Euh, D.J., Chu, I.C., 2002. Direct ECC bypass phenomena in the MIDAS test facility during LBLOCA reflood phase. J. KNS 34 (5), 421–432.