Hydrodynamics of ECC Water Bypass and Refill of Lower Plenum at PWR-LOCA

Transcription

1 Journal of Nuclear Science and Technology SSN: (Print) (Online) Journal homepage: Hydrodynamics of ECC Water Bypass and Refill of Lower Plenum at PWR-LOCA Kazuharu OKABE & Yoshio MURAO To cite this article: Kazuharu OKABE & Yoshio MURAO (1987) Hydrodynamics of ECC Water Bypass and Refill of Lower Plenum at PWR-LOCA, Journal of Nuclear Science and Technology, 24:1, , DO: 1.18/ To link to this article: Published online: 15 Mar 212. Submit your article to this journal Article views: 18 View related articles Full Terms & Conditions of access and use can be found at

2 Journal of NUCLEAR SCENCE and TECHNOLOGY, 24( 1), pp, (October 1987). 785 Hydrodynamics of ECC Water Bypass and Refill of Lower Plenum at PWR-LOCA Kazuharu OKABEt and Yoshio MURAO Japan Atomic Energy Research nstitute* Received August 1, 1986 Revised January 12, 1987 n order to develop the hydraulic model of the ECC water bypass and the refill period at a PWR-LOCA, flashing transient and the downcomer Counter Current Flow Limit (CCFL) experiments were conducted with the large scale Cylindrical Core Test Facility (CCTF). The ECC water was bypassed by the two phase mixture swelled from the lower plenum to the downcomer at the flashing transient experiments. This swelling behavior was predicted well by using the void fraction correlation proposed by Okabe & Murao. The CCFL correlation based on Battelle experiments predicted well the bypass of the ECC water and its penetration into the lower plenum at the CCFL experiment. These swelling and CCFL models were combined to form the best estimate analytical model which was applied to a large break LOCA of a commercial PWR. Calculated results showed that () the proposed model predicted the existence of water at the initiation of the refill, (2) the present licensing calculation was conservative because it assumed no water at this time, and (3) this conservativeness mainly came from the neglect of the water in the downcomer at the ECC water bypass period. KEYWORDS: blowdown, ECCS, hydrodynamics, loss of coolant, lower plenum, PWR type reactors, refill, swelling, two-phase flow, water bypass. NTRODUCTON t is an important problem to predict the fuel clad temperature at a hypothetical large break loss-of-coolant accident (LOCA) for the safety evaluation of a pressurized water reactor (PWR). Peak clad temperature depends significantly upon the estimation of the refill phase (between the time of the end of ECC water bypass and the initiating time of ref!ooding) because the fuel clad temperature is expected to increase almost adiabatically during this period. Previous studies on this ECC water bypass and the refill behavior mostly covered counter current flow (ascending steam flow and falling down water flow) experiments(t)-(3) and the flashing transient experimentsc 4 >. However, these were based on small scale experimental facilities and also they did not discuss in detail the whole hydraulic behavior from the end of ECC water bypass until the refill of the lower plenum. The objective of this study is to clarify the whole hydraulic behavior for this period. Several experiments were conducted by using the large scale facility to simulate the end of blowdown and succeeding refill period. Based on the findings of these experiments, hydraulic models were proposed and applied to the calculation of the end of ECC water bypass and the succeeding refill period at a commercial PWR-LOCA. Chapter ll describes these experiments. Results and discussions are presented in Chaps. m and N. Concept of analytical models proposed here and its application to a commercial PWR-LOCA are summarized in Chap. V. * Tokai-mura, baraki-ken t Present address: Mitsubishi Atomic Power ndustries, nc., Shibakouen, Minato-ku, Tokyo

3 786 J. Nucl. Sci Techno, H. EXPERMENT 1. Test Facility Test facility used in this study is the Cylindrical Core Test Facility (CCTF)(s) of Japan Atomic Energy Research nstitute (JAER), which is the integral system test facility designed to simulate the flow conditions in the primary system of a PWR during the refill and reflood phase of a LOCA. The vertical dimensions of the system components are nearly the same as those of a 1, MWe-PWR with four primary loops. The flow area of the system components is based on the core flow area scaling ratio of 1:21.4. Figure 1 shows the bird's eye view of CCTF. The pressure vessel consists of the annular downcomer, the upper and lower plenums and the non-nuclear core in which 1,824 electrically heated rods are installed. Figure 2 shows the dimension of the pressure vessel (PV) and the location of the instrumentations. The gap of downcomer is 61.5 mm and the volume of the lower plenum is 1.3 m 3 without internal structures. The facility has three intact loops and a broken loop. Each loop consists of hot leg and cold leg pipings, and active steam generator and an orifice which simulates the resistance of a primary coolant pump. The ECC water is injected into each intact cold leg through angled injection port. Fig. 2 Pressure vessel of CCTF Sturn injection nozzle Top of heated section BottomC'f heated section Lovver plenum Unit:mm Pressure regulation valve Steam generator Exhaust line,. ntact loop 1i Broken loop Fig. 1 Bird's-eye view of CCTF -22-

4 Vol 24, No. 1 (Oct 1987) Test Conditions and Procedures Four experiments were conducted to simulate the hydraulic behavior from the end of blowdown until the refill phase at a PWR-LOCA. Three of them were classified as the flashing transient experiments and the other was the downcomer Counter Current Flow Limit (CCFL) experiment. Flashing transient experiments simulated the depressurization and the fluid behavior at the later portion of the blowdown phase of a PWR LOCA. Test parameters are shown at Table 1. Table 1 Test conditions Flashing transient experiment C2-2 nitial system pressure (MPa). 53 Containment pressure (MPa).2 nitial lower plenum water level 2. Blowdown valve opening time (s) PV side SG side 1 ECC injection Start time (s) 6 Flow rate (m'/s).8 C2-ll not used (after 22 s) Water temperature (K) Loop isolation No Yes Core reflood simulation (core was heated or not) No No C No Yes Down comer CCFL experiment.2.2. not used Yes No Primary system was pressurized to a specified pressure (.53.6 MPa) and the lower plenum of PV was filled to a specified level (.5 or 2. m) with saturated water. The tests were initiated by the opening of the blowdown valves installed in the broken cold leg and after the flashing of the saturated water in the lower plenum started, ECC water was injected into three intact cold legs. Containment pressure was regulated at almost constant pressure (about.2 MPa). nitial water levels of test C2-2 and C2-11 were 2. and.5 m at C2-14 respectively. High initial water level prolonged the flashing transient and the detailed data could be obtained for this period. Low initial water level test (C2-14) was also conducted because the typical licensing calculation for a PWR-LOCA predicted a little of water at the later portion of the blowdown. At test C2-11, all loops were isolated mechanically at pump orifices. This loop isolation caused the steam flow direction from core to the lower plenum, which is the typical direction at the later portion of a PWR-LOCA. After the end of the refill of the lower plenum, the reflooding phase of the heated core was simulated only at the test C2-14 and the core was not heated at the other tests. Downcomer CCFL experiment was conducted to obtain the data of the counter current flow because this situation was shortly observed in the flashing transient experiments. n this experiment, ascending stam flow in the downcomer was established by the saturated steam injection into the upper plenum with the loop isolations. The ECC water was injected into three intact cold legs in the same way as one of the flashing experiment. The ECC water was heated to near the saturation temperature to avoid the rapid condensation. Containment pressure was also regulated at.2 MPa. Test parameters of this experiment are also shown at Table 1. Pressure of the upper plenum, differential -23-

5 788 J. Nucl. Sci Techno/., pressure and the fluid temperature of the lower plenum were measured and reported in this paper although much more detailed instrumentation systems were installed in the CCTF. Figure 3 shows the sketch of fluid behaviors of three flashing transient experiments at about s after the blow down valve open and one of the downcomer CCFL experiment at about s after the start of the test. ECC Water injection <::::J Steam flow -Waterflow ///////// ////////// //_//!////; ///////// ffi. C2-2 C2-11 TEST RESULTS 1. Fluid Behavior in Flashing Transient Experiments Test results of three flashing transient experiments are presented and compared in this chapter. Figure 4 shows the pressure transients of upper plenum of these experiments. When the blowdown valve was opened (time= s), the depressurization of the primary system was initiated. After the system depressurized rapidly, the depressurization rate decreased and then pressure reached a steady state. Pressure of the upper plenum began to recover in 3 s at the test C2-14 because the steam was generated in the heated core after the end of the refill to simulate the reflood phase. Pressure difference between containment and the upper plenum in the later period of the test C2-ll was caused by the mechanical seal at the pump orifice. Start times of ECC injection are also shown in Fig. 4, and it was found that the depressurization was not much influenced by the subcooled water injection. Figure 5 shows the lower plenum differential pressure measured at the locations shown in C2-14 Downcomer CCFL Fig. 3 Sketch of fluid behavior ,;------r ,------,.6.4 C2-2 Pressure of -- Upper plenum ----Containment.2,.,-..._... ' C2-11 Refill end.6 J,., Refill start ECCwater injection stan, C2-14 '" ' "'...., Time(s) 8 Fig. 4 Pressure upper plenum 1-24-

6 Vol 24, No. 1 (Oct. 1987) 789 Fig. 2. Two kinds of differential pressure data were the same as each other. This differential pressure indicates the water inventory in the lower plenum since the frictional and accelerational pressure drop are estimated to be negligible in these tests. Fluid behaviors in Fig. 5 can be classified as follows: (1) Mass depression period: Two phase mixture swelled from the lower plenum into the downcomer due to the flashing by the depressurization. (2) Refill period (a): Although the two phase mixture stopped swelling, steam generation still continued by the depressurization, which caused the counter current flow condition (ascending steam and falling water).1.1 F---- (1) C Downcomer water collapse (2) (3) (1) (2) (3) C2 14 (1) (2) (3) 2 - Downcomer water collapse :J Refill end Refill start ( 1) Swelling of two phase mixture (2) Counter current flow (3) Complete ECC water penetration Time (s) Fig. S Differential pressure of lower plenum at flashing transient experiments 1 in the downcomer. (3) Refill period (b): After the steam generation in the lower plenum terminated, ECC water penetrated into the lower plenum completely. n the period (1 ), injected ECC water was carried to the break point by the swelled two phase mixture through the downcomer, in other words, ECC water was bypassed. This ECC water bypass continued about 2 s in the tests C2-2 and C2-11 as shown in Fig. 5. n the test C2-14, this swelling behavior did not continue for a long time because of the low initial water inventory in the lower plenum. The rapid decrease of the water inventory of the early phase of the test C2-11 was caused by the loop isolations, namely, the expanded steam by the depressurization could not be released through hot legs but suppressed the two phase mixture level to the lowest elevation of the barrel in the lower plenum. When the depressurization rate became lower, swelling of two phase mixture terminated and the recovery of the water inventory in the lower plenum (refill) started (Period (2)). As shown in Fig. 5, the rapid recovery of the water inventory was observed at the initiation of the refill of the test C2-2. This rapid water increase in the lower plenum was caused by the collapse of two phase mixture at the downcomer due to the termination of the ascending steam flow. Actually, the increased water mass was evaluated to be the same as water inventory in the downcomer before the end of ECC water bypass. This collapse of two phase mixture took longer time ( 4-5 s) than the free falling time in the downcomer (estimated as about 1 s) because the ascending steam flow still continued in the downcomer due to the flashing water velocity. Rapid increase of water inventory by the collapse of two phase mixture in the downcomer was also observed at the test C2-11. n the period (3), water increase rate in the lower plenum basically corresponded to the ECC water injection rate into the cold legs. n the test C2-11 loop mechanical seal suppressed the increase rate of water level after the water level increased to the higher elevation than the lowest end of the barrel. -25-

7 79 J. Nucl. Sci. Techno/., 2. Fluid Behavior in Downcomer Counter Current Flow Experiment n this experiment, steam was injected continuously into the upper plenum and the ascending steam flow continued in the downcomer because of the mechanical loop isolations at the pump orifice.s. Figure 6 shows the superficial steam velocities in the downcomer region. Because the injected steam partially condensed by ECC water at the downcomer region, two velocities are shown in Fig. 6, namely, one was calculated from the injected steam mass flow rate and the other was from the released steam mass flow rate through the break point, as follows : where j g : Superficial velocity in downcomer mg : Measured steam mass flow rate pg : Saturated steam density Ave : Flow area of downcomer.!.5 > g E 1l ll ( ) 2r , 15 1 r---""' \ \ '-._ l ', --- Calculated from injected steam flow - Calculated from injected steam flow Time (s) Fig. 6 Superficial steam velocity at downcomer \... Figure 7 shows the differential pressures of the lower plenum, which corresponds to the water head and frictional pressure drop. Two kinds of differential pressure data are plotted in Fig. 7, one was measured inside the barrel and the other was measured outside the barrel. Water accumulation rate was the medium value between two values calculated from two data of differential pressures in Fig. 7 by neglecting the frictional pressure drop because the steam flowed reversely in these regions (inside and outside the barrel), and the frictional pressure drop effected reversely on these two kinds of 3.15r , E g -a 1.1 C Steam injection start l!.5 ECC water e a Me-..rement t ouuk:le barrel 11) 12) 13) oosol Time(s) Period (1) Ascending steam flow prevented water down flow (2) Counter current flow (3) Lower plenum was full of water Fig. 7 Differential pressure of lower plenum at counter current flow experiment data. Figure 7 shows that ECC water penetration into the lower plenum was prevented because of high velocity of the ascending steam flow at about 5 s, in other words, ECC water was bypassed. ECC water bypass was also observed in the flashing tests, as described in the previous chapter, although the ascending fluids were different (steam in this test and two-phase flow in the flashing transient tests). ncrease of water inventory in the lower plenum was observed after ECC water bypass period. This period was the counter current flow condition because the steam ascended continuously through the downcomer in this experiment. Counter current condition was also observed in the refill period (a) of the flashing transient experiments discussed in the previous section. n the following chapter, quantitative prediction for the transient of water inventory in these experiments is discussed. N. DSCUSSONS 1. Prediction of Swelling of Two Phase Mixture At first, it is important to predict the swelling behavior due to the flashing by the depressurization in the lower plenum since the swelling of two phase mixture into the downcomer prevents the penetration of ECC water into the lower plenum. -26-

8 Vol 24, No. 1 (Oct. 1987) 791 Okabe et al. (s) proposed the application of his void fraction correlation derived from the air-water experiment of the lower plenum geometry to this swelling behavior. They reported that the void distribution in the lower plenum was unique and could not be predicted by previously proposed correlations derived from the tube geometry experiments. Okabe also compared the predictions of the swelling behavior in the CREARE flashing transient experiments by various models and reported the best agreement of his model with experimen- tal data. The CCTF used in this study are larger than the facility of CREARE experiments and well simulates PWR, therefore, comparison of several analytical model in CCTF experiments was studied again in this chapter. n each calculation pressure transient was given as boundary condition to avoid the effect of break flow model, and the ECC injection was not simulated. Features of applied models are summarized at Table 2 and the following paragraphs. Table 2 Analytical models for two phase mixture swelling in lower plenum Model Nodes in lower plenum Steam-water slip model Note () (2) (3) (4) 8 TRAC-PD2 drift fluid model 8 TRAC-PD2 modified drift flux model Wilson's correlation was used instead of the original model. Wilson Okabe & Murao Wilson's void fraction correlation was converted to the slip formulation. ditto Lower plenum node is defined as the volume below the barrel.! Pressure tronsient is given) BREAK BREAK {Pressure tronsient is given) Core BREAK Pressure transient) ( S gtven Core Volume of primary system is simulated Down comer Lower plenum Downcamer Lower plenum TEE (a) C2-2 (b) C2-ll Fig. 8(a), (b) Nading schematic of TRAC-PD2 for C2-2 and C2-ll TEE ( 1 ) One-dimensional drift flux model equilibrium hydrodynamics. ts one-dimensional The TRAC-PD2 code< 6 > was used for this calculation. This code features a three-dimensional treatment of the pressure vessel with two velocities, non-equilibrium hydrodynamics and of the loops with drift flux model and nonlower drift flux model was applied to the plenum of the CCTF with multi-node simulation. Slip between steam and water is calculated by the incorporated correlation in TRAC-PD2 by considering the flow pattern map. No ding schematics are shown in Fig. 8(a), (b). -27-

9 792 ( 2 ) Modified one-dimensional drift flux model Because water velocity was not large during this swelling behavior of the two phase mixture, Wilson's void fraction correlation(?) was judged to be applicable. Wilson's correlation can be written as where a: Void fraction V a: Steam velocity. Slip velocity between water and steam V r was assumed to be equal to the steam velocity calculated by the Wilson's correlation as The original slip correlations in TRAC-PD2 code were exchanged for this correlation in this model. Noding schematic is the same as model (1 ). ( 3 ) One node calculation with Wilson's< 7 > void fraction correlation (CREARE model) n this model, lower plenum was described as a single node. n the lower plenum node, thermal equilibrium and homogeneous (uniform distribution of void fraction) conditions were assumed and the exit steam flow rate was calculated from the average void fraction of the lower plenum with Wilson's correlation. Water exit flow rate was calculated by the expansion rate of the two phase mixture in the lower plenum. n other words, slip in the exit flow was taken into account by using Wilson's void fraction correlation in the same way as (2). This model has been developed by CREARE( 4 ) and is termed CREARE model. ( 4) One node calculation with new void fraction (Proposed model) Okabe et a/. ls) conducted the small scale air-water experiment simulating the flashing in the lower plenum and they proposed new void fraction correlation as follows : where a=ao+ (1-ao)jgT, ( 2) c=.85/(a,a,a,) a,= ((Pt-Pa)l Pa)" '" a,= (Dj..;' q /(pl- Pa))" '"' a,= (g f..;' u l(pt-pa))" '". J. Nucl. Sci TechnoL, This correlation includes the effect of the vertical distribution of void fraction in the lower plenum, considering the phenomena that the liquid deficient layer is formed at the upper portion of the lower plenum due to the entrainment of the liquid into the downcomer by the ascending gas flow. And this new correlation was used in the slip calculation instead of Wilson's correlation of model (2). n this model, lower plenum node is defined as the volume lower than the lowest end of the barrel. This model is valid in the safety evaluation because the present licensing calculation shows much lower liquid level in the lower plenum than the lowest end of the barrel. Multi-node simulation (models ( 1) and (2)) was applied to tests of C2-2 and C2-11, because one node simulation is enough to simulate the test C2-l4 of low initial water inventory. Because Okabe's void fraction correlation is applicable to the case when two phase mixture level is below the lowest elevation of the barrel (typical situation of the end of blowdown phase of a PWR-LOCA), model (4) was applied only to tests C2-11 and C2-14. The lower plenum fluid condition of the test C2-2 (liquid level stayed at upper level than the end of the barrel) is not the situation of a commercial PWR-LOCA. Calculation of C2-11 was initiated at 4 s after the water level became lower than the barrel. Measured and calculated differential pressures of the lower plenum are compared in Fig. 9(a) (c). For the tests of C2-11 and C2-14, model (4) (one node simulation with the new void fraction correlation) gave the best prediction of the decrease of water inventory, and also the calculated times of the termination of swelling of two phase mixture (when the decrease of water inventory terminated) are the same as the initiating time of refill in the experiments. The conclusion is the same as one of the previous study by Okabe et a/. (s) for CREARE experi- -28-

10 VoL 24, No. 1 (Oct. 1987) 793 ments. Therefore, model ( 4) was judged to be applicable to the prediction of the swelling behavior of two phase mixture in the lower plenum. Model (2), modified version of TRAC-PD2 ::;: :::1., c. 12 c :!! :::1., e! c..e c e! Q) -.2 C , 121 -,... Experiment... '\ Calculation... ttl TRAC PO 2 ' ll (2) Modified TRAC (3) Wilson Time l sl Experiment Colculation () TRAC PO 2 (2) Modified TRAC (3) Wilson (4) Proposed (4)' t '... -::::::: _ t (11 o--,:'::o: :'2o-=----3, '--'--'4o.2.1. C2-14 Fig. 9(a)- (c) Time l sl Experiment Cotcutotian (3) Wilson 141 (41 Proposed Time l sl Comparison of decrease of differential pressures in lower plenum (tests C2-2, C2-11 and C2-14) code (Wilson's correlation is used instead of the slip correlation of the original version) predicted well the result of C2-2 as shown in Fig. 9 although model (3) (one node simulation with Wilson's correlation) overpredicted the mass inventory in all three cases. This predictions for C2-2 showed the importance of the account of the vertical void distribution in the lower plenum, which was also included model (4). 2. Prediction of Water Accumulation (refill) in Lower Plenum As described in Chap. ill, refill (water accumulation) period was classified into two phases as follows: () All of injected ECC water penetrated into the lower plenum after the depressurization terminated. n this period, water accumulation rate was easily calculated from the ECC water injection rate. (2) Complete penetration of ECC water into the lower plenum was suppressed by the ascending steam flow through the downcomer. n this period, the CCFL correlation is required for the predictions of falling water flow rate into the lower plenum. Experimental data in the downcomer CCFL test of this study were compared with the previous studies to obtain the CCFL correlation. As shown in Fig. 5, two kinds of ascending steam velocities became nearly the same at about 6 s (when the complete bypass was observed) and from 14 to 16 s. Battelle conducted the air-water CCFL experiment{3) at the pressure vessel shown in Table 3, and they correlated their data by using the Kutateladze number which was defined as K= (Px/f:) 112 /[ga(pl-peh'' (x: or g.) Table 3 Comparison of test vessel Battelle CCTF Scale 1/15 2/15 "'1/5 Downcomer Gap size (em) Length (em) Average annulus circumference (em)

11 794 J. NucL Sci. Techno/., Test data of the air-water experiment of Battelle and the steam-water experiment of this study were plotted by using the Kutateladze number in Fig. 1. t was found from Fig. 1 that CCTF data were consistent with the Battelle data although the fluids and the scale of the vessels used in the test were different each other. H--..._ CCTF 4. x ,.2 C2-2 l Complete ECC water penetration 12) Counter current flow 3.. ATTELL 3 l 2.5 B' o ; l' /15- sccle model 2/15- sccle model 1.5 K,; Fig. 1 Comparison of counter current flow data The previous studies on the counter current flow, proposed the correlation based on the modified Wallis parameter(!) which was defined as j= [pxjj:/ g La (pt- pg)] 112 However, the reasonable definition of the characteristic length La in 1: could not be obtained for the different scaled tests including the Battelle and CFTF. Therefore, the data in Fig. 1 were approximated in this study with the following equation of the Kutateladze number as follows : ( 3) Complete ECC water penetration assumption and CCFL correlation (2) were applied to the CCTF tests, which results are shown in Fig. ll(a) "'(c). Time changed from period (2) to (1) in Fig. ll(a)"'(c) was the end time of the depressurization. Ascending steam flow rate was estimated from the averaged flashing rate due to depressurization in the lower plenum (1 kg(s, kg/s, 1.3 kg/s for C2-2, C2-ll and C2-14, respectively). These values overestimated the steam flow rate (underestimated the water accumulation) for C2-2 and C2-14 because the a.. ::!:!! :::J en e c..1 cu -Cl.1 3 Time 5 7 lsl.2r ; :-;"',-,..., C2-11 ll/ 9 - Experiment --- Calculation ) Complete ECC woter penetration {2) Counter current flow o.o2.':;o---=3'=o---='4o=----=5'="o..,..so.,---_j7o C ) Time (s) 1 - Experiment Calculation ( l Complete ECC water paetrotion (2) Counler currenl flow O.O : ", '::..,2,_3,4 Fig. ll(a)"' (c) Time lsl Comparison of water accumulation in lower plenum (tests C2-2, C2-11 and C2-l 4) steam flowed out not only to the downcomer but also partially to the core. As shown in Fig. ll(a)"'(c), water accmulation in the lower plenum was generally predicted well by these calculations. Difference between the experiment -3-

12 Vol 24, No. 1 (Oct. 1987) and the calculation in the counter current flow period of C2-2 and C2-14 mainly caused by the overestimation of steam flow rate into the downcomer as described before. However, in the complete penetration period of the test C2-2, the additional steam flow was generated by the heat transfer from the structural metal of the lower plenum to the water and the steam flow caused slightly slower water accumulation than the prediction. At the test C2-11, loop mechanical isolation at pump orifices caused the increase of pressure of upper plenum and it suppressed the water accumulation above the lowest end of the barrel. V. APPLCATON TO COMMERCAL PWR-LOCA CALCULATON n order to apply the findings in CCTF tests to the prediction of the hydrodynamics in the lower plenum from the end of ECC water bypass until the refill period of a commercial PWR LOCA, several calculations were performed based on the following model : () Swelling of two phase mixture period: The ECC water does not penetrate into the lower plenum during the period when the two phase mixture continues to swell from the lower plenum into the downcomer. This swelling behavior is calculated by the model ( 4) of Chap. ill, that is, the lower plenum below the barrel is simulated as a single node and the steam exit velocity from the two phase mixture of this lower plenum node is calculated with the void fraction correlation of Eq. ( 2 ). (2) Counter current flow period: After the termination of the two-phase mixture swelling into the downcomer, penetration of ECC water into the lower plenum is initiated. The penetrating water flow rate through the downcomer is controlled by the counter current flow correlation of Eq. ( 3 ). (3) Complete ECC water penetration period: After the ascending steam flow due to depressurization terminates, water flow rate into the lower plenum is equal to the ECC water injection rate into the cold legs. This model was applied to a large break LOCA of a commercial PWR and the result was compared with the typical licensing calculation( 9 ). 795 The referred pressure transient as the boundary condition was taken from the calculation result with RELAP4-MOD6( 9 ), which is shown in Fig. 12. CREARE( 4 ) showed that the swelling model did not influenced so much the pressure transient of the lower plenum. The ECC water injection rate and the downward core steam flow rate calculated with RELAP4-MOD6 were also used as the boundary conditions. TRAC-PDi 6 ) best estimate oriented calculation for a large break LOCA of the typical 1, MWe PWR was also referred Time after break ( s l Fig. 12 Pressure transient at large break LOCA of commercial PWR by RELAP4 code Sensitivity studies were performed on the start time of the calculation, the initial water inventory in the lower plenum and the effect of supported water at the upper portion of the downcomer. t was found that the start time and the initial water inventory were not so sensitive to the remained water at the initiating time of the refill period. Start time of the calculation of Cases 1 and 3 was 16 s after the break and the start time of Case 2 was set to 8 s. nitial lower plenum water inventory of Cases 1 and 3 was taken from TRAC-PD2 calculation and the result of RELAP4-MOD6 was used in Case 2. And Case 3 accounted the effect of the falling of supported water at the upper portion of the downcomer. Ascending steam flow rate used for the counter current flow calculation at the downcomer was evaluated at the sum of the downward core steam flow rate calculated by RELAP

13 796 J. Nucl. Sci. Techno!., MOD6 and the steam generation rate due to the flashing of the water in the lower plenum of this calculation. Figure 13 shows the comparison of the calculated water inventory of the lower plenum. Homogeneous model of RELAP4 predicted the rapid decrease of the water inventory, and refill was initiated from no water condition of the lower plenum. And there was also no water in the downcomer at the initiation of the refill, because the artificial mass subtraction from the reactor vessel was required in the safety evaluation calculation. On the other hand, proposed model estimated the remained water at the initiation of the refill and the earlier end of the refill. These calculations showed the conservativeness of the licensing (RELAP4) calculation because the early end of the refill reduced the adiabatic increase of the clad temperature. However, the time difference was not so large bcause the steam flow rate calculated by RELAP4 was large enough to prevent the ECC water penetration until the end of ECC bypass. Case 3 showed the earlier initiation and end time of the refill than the other cases. n Case 3, supported water mass in the downcomer was calculated from the water inventory at the upper node of the downcomer of RELAP4 calculation. This result showed that the account of the supported water in the downcomer before the end of ECC water bypass was important for the prediction of the reflooding initiating time. E :J c r ' 2 \ -- Present calculation ( ) nitial value from TRAC (2) nitial value from RELAP (3) ncludes D.C water RELAP4 TRAC.!: ll 1 \ ' ',, \ \ \ 1 2 Time after break ( s ) Fig. 13 Comparison of calculated transient water inventory in lower plenum This study showed the conservativeness of Although the verification of Eq. ( 1 ) the present licensing calculation, however, was shown by the large scale experiments future effort is required for resolving the follow- (CREARE & CCTF), additional experiment ing problems : with larger size is desirable. () Effect of subcool temperature on counter (3) Best estimate calculation for downcomerflow at large scale downcomer region : steam flow rate at end of blowdown Equation ( 2 ) was derived from the experi- phase : ments of air-water and high temperature The steam flow rate calculated by RELAP4 water-steam. Subcooled ECC water of a was given as the boundary condition in this commercial PWR may condense much study. ts flow rate might be overestimated amount of the ascending steam and may because the thermal-equilibrium model of penetrate into the lower plenum earlier RELAP4 might overestimate the condensathan the estimated value by Eq. ( 2 ). tion rate by ECC cold water at the cold legs (2) Applicability of Eq. ( 1 ) to large scale and the depressurization rate. vessel : -32-4

14 VoL 24, No. 1 (Oct. 1987) 797 V. CONCLUSON (1) The ECC water was bypassed by the ascending two phase mixture swelled from the lower plenum into the downcomer at the flashing transient experiments in the large scale Cylindrical Core Test Facility (CCTF). (2) The swelling behavior in the lower plenum was predicted well with the void fraction correlation Eq. ( 2 ). (3) Counter Current Flow Limit (CCFL) cor relation based on Battelle experiments (Eq. ( 3 )) predicted well the ECC water bypass and partial penetration observed at the downcomer CCFL test in the CCTF. (4) Findings in the CCTF experiments were combined to form the best estimate analytical model for the ECC water bypass and the refill of the lower plenum at a PWR LOCA. (5) This model was applied to a commercial PWR-LOCA and the calculated result was compared with the typical licensing calculation. Findings were as follows : (a) the proposed model predicted the existence of water at the initiation of the refill. (b) the licensing calculation was conservative because it predicted no water at this time. (c) this conservativeness mainly came from the neglect of the water in the downcomer at the ECC water bypass period. c: D: g: H: J'. a. J'.. ja : j, : K: K'.. K. l. [NOMENCLATURE] Parameter of Wilson's correlation Equivalent hydraulic diameter (m) Acceleration of gravity (m/s 2 ) Height of lower plenum (m) Wallis parameter for j a Wallis parameter for j 1 Superficial"gas velocity (m/s) Superficial liquid velocity Constant in Eq. ( 2 ) Kutateladze number for j a Kutateladze number for j, (m/s) La : Characteristic length in CCFL Eq. ( 3 ) ma : Steam mass flow rate (kgfs) n : Parameter of Wilson's correlation V a : Vapor velocity (m/s) a: Void fraction ao: Average coid fraction of froth layer a: Overall average void fraction of lower plenum Density of saturated gas Density of saturated liquid Pa : P : a : Surface tension (Superscript) U: Gas, : Liquid, : Average T: Total. ACKNOWLEDGMENT (kg/m 3 ) (kgfm 3 ) (kg/m) The authors are deeply indebted to Messrs. T. guchi, J. Sugimoto, Dr. H. Akimoto and Mr. T. Okubo for their analytical and experimental support. They a(e much grateful to Dr. K. Sato and Dr. T. Shimooke for hearty suggestions. -REFERENCES- ( ) WALLS, G. B.: "One-dimensional Two Phase Flow", (1969), McGraw Hill. ( 2) CROWLEY, C. 1., et al: Down comer effects in a 1/ 5 scale PWR geometry experimental data report, NUREG-281, (1977). ( 3) SEGEV, A., et al.: Development of a mechanistic model for ECC penetration in a PWR downcomer, NUREG/CR-1426, (198). ( 4) CROWLEY, C. J., eta/.: Analysis of flashing transient effects during refill, NUREGjCR-1765, (198).,( 5) HRANO, K., MURAO, Y.: Large scale reflood test, J. At. Energy Soc. Jpn., (in Japanese), 22[ 1], (198). ( 6) L,!LES, D. R., et a/.: TRAC-PD2; An advanced best-estimate computer program for PWR loss of coolant accident analysis, NUREGjCR-254, (1981 ). (7) WLSON, J. F., et al.: Steam volume fraction in a bubbling two phase mixture, Trans. Am. NucL Soc., 5, 15 (1962). ( 8) OKABE, K., MURAO, Y.: Swelling model of twophase mixture in lower plenum at end of blowdown phase of PWR-LOCA, J. Nucl. Sci. Tee/moL, 21[12], (1984). ( 9) FSCHER, S. R., et a/.: RELAP4/MOD6; A computer program for transient thermal-hydraulic analysis of nuclear reactors and related systems, CDAP-TR-3 (1978)