RELAP5/MOD3.2 ASSESSMENT STUDIES BASED ON THE PACTEL AND PMK-2 LOSS OF FEED WATER TESTS

Transcription

1 Proceedings of the 3rd ASME/JSME Joint Fluids Engineering Conference July 18-23, 1999, San Francisco, California FEDSM RELAP5/MOD3.2 ASSESSMENT STUDIES BASED ON THE PACTEL AND PMK-2 LOSS OF FEED WATER TESTS József Bánáti Lappeenranta University of Technology Department of Energy Technology P.O. Box 20 FIN Lappeenranta, Finland FAX: , banati@lut.fi György Ézsöl Laboratory of Thermal Hydraulics KFKI Atomic Energy Research Institute P.O. Box 49, H-1525 Budapest, Hungary FAX: ezsol@sunserv.kfki.hu Jyrki Kouhia VTT Energy / Nuclear Energy P.O.Box 20 FIN Lappeenranta, Finland FAX: Jyrki.Kouhia@vtt.fi Keywords: RELAP5, PACTEL, PMK, VVER-440, loss of feed water, thermal hydraulics ABSTRACT Geometry of the VVER-440 type reactors are different from pressurized water reactors widely used in the world. Due to the specific features of these nuclear power plants, the computer codes originally developed for PWRs have to be validated for VVER experiments. In order to set up experimental data bases, the PMK-2 and the PACTEL facilities were constructed and upgraded during the past years. A number of computer codes were used for the analysis of the experiments, such as RELAP, CATHARE, APROS and ATHLET. This paper gives an assessment for the results of RELAP5 MOD 3.2 simulation against Loss-Of-Feed-Water type experiments performed in both facilities. The PACTEL test LOF-01 involved an evaporation of the secondary side water using constant core power, while the primary and secondary pressures were controlled within a narrow range. The PMK-2 TLFW experiment simulated a plant transient with total loss of feed. For prevention of core damage, primary side Bleed-and-Feed with secondary side bleed accident management measures were applied. After a brief description of the facilities, the profile of the experiments is presented. Modeling aspects are discussed in the RELAP analyses of the tests. Emphasis is placed on the ability of the code to predict the main trends observed in the tests and thus, an assessment is given for the code capabilities to represent the system transient. The overall agreement was fairly good in both cases, however, the calculated parameters showed strong sensitivity for the requested time step in the LOF-01 analysis. INTRODUCTION Performance of the Loviisa NPP in Finland and the Hungarian Paks NPP has been excellent since the beginning of their commercial operation and these PWRs are among the highest in the ranking of load factors. This achievement can be explained by several reasons but the importance of the safety related investigations cannot be overemphasized. Looking at the status of system code verification process of VVER-440/213 in the thermal hydraulic area, there is a great need for enlargement of existing data base with new experiments. Without the proper benchmarking against experimental data, a computer code cannot be used with any degree of confidence. With this purpose in mind, the OECD Nuclear Energy Agency and the International Atomic Energy Agency organized test series, such as the International Standard Problems (ISP-33), and Standard Problem Exercises (SPE-1 to 4) in the Finnish PACTEL and the Hungarian PMK-2 facilities, respectively. Results of these large 1

2 Table 1: Comparison of main parameters Parameter PACTEL PMK-2 VVER-440 Volumetric scaling ratio 1:305 1:2070 1:1 Vertical scaling factor 1:1 1:1 1:1 Number of primary loops Max. heating power 1 MW 0.7 MW 1375 MW Number of fuel rods Diameter of heater rods 9.1 mm 9.1 mm 9.1 mm Heated length of the core 2.42 m 2.50 m 2.42 m Axial power distribution ~cosine uniform cosine Axial peaking factor Number of SG Number of SG tubes Average SG tube length 8.8 m m 9.26 m Number of SG tube rows Pitch of the SG tubes 24 mm mm 24 mm Max. operating pressure 8.0 MPa 12.3 MPa 12.3 MPa Nominal prim. temperature 573 K 573 K 573 K Nominal. sec. pressure 4.65 MPa 4.65 MPa 4.65 MPa Nominal sec. temperature 533 K 533 K 533 K Accumulator pressure 5.5 MPa 5.5 MPa 5.5 MPa LPIS pressure 0.7 MPa 0.7 MPa 0.7 MPa HPIS pressure 8.0 MPa 12.0 MPa 12.0 MPa projects demonstrated the applicability of various computer codes for a few type of transients. (Purhonen et al., 1995), (IAEA 1987, 1988, 1991 and 1996). The principal aim of this paper is to contribute to validation process with two analyses using RELAP5 Mod 3.2 version of the code against a specific type of accidental situation, the total loss of feed water. Since the code was originally developed for Western-type reactors, it is worthwhile to access the capabilities in case of the VVER geometry. In particular, the code has been challenged by simulation of the heat transfer in horizontal tube bundle in the steam generators of both facilities, even if the progression of the transients are relatively low. The PACTEL loss of feed water experiment (LOF-01) was performed with low core power and with controlled primary and secondary pressures within a narrow range, while the water in the steam generator secondary side was let to evaporate. The primary side was operated in natural circulation mode and two out of the three loops of the PACTEL were valved off from the system. The main goal was to study the heat transfer process and give an opportunity for testing the steam generator model used in various computer codes. The experiment was performed in 1993 when the facility included 3 small-diameter steam generators. Circulating pumps were added later to the primary system and the steam generators were replaced by new largediameter types. A quick look description of the test was written by Kouhia (1998). The PMK-2 total loss of feed water test (TLFW) simulated a beyond design basis accident. This initiating event was selected, because loss of feed water accidents give a relatively high contribution to core melt frequency. Consequently, the computer code prediction needs a high degree of confidence. Furthermore, this experiment included accident management measures, e.g. primary and secondary side bleed and primary side feed. A thorough description on the findings of the TLFW experiment was worked out by Ézsöl et al. (1996). The main parameters of the PACTEL and the PMK-2 facilities and their reference reactor are compared in Table 1. The detailed analysis of the RELAP5 simulation will show later in this paper that it is possible to draw some fundamental conclusions from the comparison of the calculated and experimental results. However, despite the fact that the PACTEL and PMK-2 facilities meant to be simulating the same reference reactor (the VVER-440/213), there are dissimilarities both in the initial and boundary conditions of the two tests, which prevent any direct comparison of the transients. Therefore, the present study can be considered as a combination of two individual assessments. NOMENCLATURE BRU-A Steam generator atmospheric relief valve CL Cold Leg DC Downcomer ECC Emergency Core Cooling HA Hydro Accumulator HL Hot Leg HPIS High Pressure Injection System IAEA International Atomic Energy Agency ISP International Standard Problem LOF Loss Of Feed water LP Lower Plenum LPIS Low Pressure Injection System MV Motor Valve PACTEL PArallel Channel TEst Loop PMK Paks Model Circuit (in Hungarian) PRZ Pressurizer PV Pressure Valve PWR Pressurized Water Reactor SG Steam Generator SIT Safety Injection Tank SPE: Standard Problem Exercise TLFW: Total Loss of Feed Water UP: Upper Plenum VVER: Water cooled Water moderated Energetic Reactor (in Russian) dt: Time step )t: Requested time step (in RELAP) 2

3 Figure 2: The steam generator of the PACTEL Figure 1: The PACTEL facility MODELING OF THE VVER-440 Figure 3: Cross section of the SG model The Reference Reactor The VVER-440/213 PWRs of Russian design are located, for instance, at Loviisa, in Finland and at Paks, in Hungary. This type of reactor has a number of unique features, such as horizontal steam generators, combined safety injection, accumulator injection to upper plenum and downcomer with higher set-point pressure than the secondary pressure, loop-seal in both hot-leg and cold-leg, hexagonal fuel bundle arrangement. Consequently, compared with Western type PWRs, the system s response to the disturbances can be different. The following phenomena have also been reproduced in most of the PACTEL and PMK-2 experiments: Heat transfer in the horizontal tube bundle of the steam generator: because of the relatively large secondary side inventory, the temperature and density distribution is different and internal circulation may develop. Loop-seal effect: phase separation in the horizontal components is different, with the possible formation of loopseals in the hot-legs and cold-legs. Loop-seals in the hot-leg may reduce or completely stop primary coolant flow to the SG during natural circulation or in boiler-condenser mode. Different mixing and condensation effects, especially during ECC injection to the upper plenum. PACTEL Facility Description Since 1976, VTT Energy and the Department of Energy Technology of the Lappeenranta University of Technology (LUT) have co-operated in the field of nuclear reactor thermal hydraulics. During these years, VTT and LUT have built a series of experimental facilities, such as REWET-I, -II -III, and VEERA to simulate PWRs. The newest facility is the PACTEL (PArallel Channel TEst Loop). A schematic drawing is depicted in Figure 1. It is an out-of-pile experimental facility designed to simulate the major components of the primary loop of a 3

4 commercial PWR during postulated small- and medium-size break Loss-Of-Coolant Accidents (LOCAs), natural circulation, and operational transients. The PACTEL facility is a volumetrically scaled model of the VVER-440 type PWR in Loviisa, Finland. It has three separate primary loops, each of which simulates two loops in the reference reactor. In the core section of the facility, there are 144 full-length, electrically-heated fuel rod simulators arranged in three parallel channels. The facility has been designed using volumetric scaling and its scaling factor is 1:305. Instead of the reference reactor s six primary loops, the PACTEL has three, with horizontal steam generators in each loop. The hot- and cold legs include the loop-seals in the same way as in the reference reactor. The main components of the facility are the primary system, the secondary side of the steam generators and the emergency core cooling system (ECC). The reactor vessel is simulated by a U-tube construction including the downcomer, lower plenum, core and upper plenum. The maximum core power is 1 MW, i.e. approximately 22 % of the nominal scaled power of the VVER The maximum primary system pressure and temperature are 8.0 MPa and 300 o C. In the secondary side, these values are 4.65 MPa and 260 o C, respectively. Restructuring of the facility started in 1993, when the primary loops were equipped with main circulating pumps and the steam generators were replaced by larger diameter components. However, the present analysis is based on an experiment which was performed with using the old steam generators and the facility was operated in natural circulation. The steam generator of the PACTEL contains 38 horizontal heat exchange U-tubes. The average length of the tubes is 8.8 m. The diameter and the pitch of the tubes are the same as in the reference SG. The heat transfer area of the tube bundle and the volume of each SG is scaled down so that one SG in PACTEL corresponds to two SGs in the reference reactor. A schematic representation of the PACTEL SG is shown in Figure 2. Figure 3 illustrates a cross-sectional view of the SG and it also indicates how the heat exchange tubes are arranged into nine layers. Description of the PMK-2 Facility The PMK-2 is a full-pressure, integral-type model of the four VVER-440 type pressurized water reactors used in the Paks NPP in Hungary. The facility (Figure 4) is located in the KFKI Atomic Energy Research Institute in Budapest in Hungary. The aspect ratio for power and volumes is 1:2070. Relative elevations are preserved the same as in the reference reactor, except for the pressurizer and lower plenum, in order to maintain the driving head for natural circulation. The primary and secondary sides of the PMK-2 correspond to six loops in the NPP. The facility is equipped with similar safety systems, e.g. two hydro-accumulators (safety injection tanks, SITs), high and low pressure injection systems (HPIS and LPIS). The SIT-1 injects water to the top of the downcomer, while the SIT-2 is connected to the upper plenum. The LPIS is attached to the downcomer head. The setpoint values of these components are identical to those of the reference plant. The Figure 4: Axonometric view of the PMK-2 reactor vessel is simulated with a U-tube construction, consisting of an external downcomer. The core itself comprises 19 directly heated fuel rod simulators embedded in a hexagonal ceramic shroud. The diameter and pitch of the heater rods are 9.1 mm and 12.2 mm, respectively and the active length is 2.5 m. The nominal power of 664 kw is distributed uniformly along the height of the rods. The steam generator model is a full height, vertically divided section of the horizontal VVER-440 steam generator, with 82 tubes bent in serpentine shape. The ratio between the volumes of steam and water is kept in the SG secondary side. The PV23 valve acts as a steam relief valve, which opens at certain secondary pressures or by operator intervention. Steam output from the SG can be controlled by the PV22 valve, which is used before the transient initiation. A simplified flow diagram of the PMK-2 is shown in Figure 5 with the location of some selected measurement. 4

5 Table 2: Initial parameters of LOF-01 Parameter Value Core power Primary pressure Secondary pressure SG sec. side level Core outlet temperature Loop mass flow rate 75 ± 5 kw 7.4 ± 0.03 MPa 4.1 ± 0.02 MPa ± m 273 ± 3 o C ± kg/s Figure 5: Simplified flow diagram of the PMK-2 with the location of selected measurements Figure 6: Primary and secondary pressures PROFILE OF THE EXPERIMENTS The PACTEL LOF-01 Experiment Experiment LOF-01 was performed in the PACTEL to study the behavior of a VVER reactor geometry during loss of feed water transient. LOF-01 was a one-loop test where the core power used throughout the experiment was lower than the scaled nominal value. The test data base is useful for verification of the system components, particularly the SG model used in RELAP5 calculation, even if the experiment cannot be considered as a realistic plant transient. The main initial parameters are summarized in Table 2. LOF-01 was started from steady-state conditions where the water level in the steam generator secondary side was above the heat exchange tubes. The test was performed with only one primary loop of the facility in operation (Loop 3), which was realized by closing the primary isolation valves in the other two loops The feed water injection to the steam generator secondary side was closed at the beginning of the test. The experiment was continued until the heat transfer to the secondary side ceased and the primary system pressure started to rise (Figure 6). The heating power in this test was 75 kw, corresponding to 1.7 % power in the reference reactor. Figure 7: Level in the SG sec. side In the experiment, the main primary system parameters remained unchanged until the uppermost layer of heat exchange tubes in the SG sec. side were no longer covered by water. This happened when the collapsed liquid level was about 0.2 m from the bottom of SG (Figure 7). Soon after this superheating of the steam was observed in the sec. side. Uncovery of the tube layers affected the flow rate in the primary side: the loop flow rate 5

6 Table 4: Occurrences in PMK-2 test TLFW Occurrence Time Initiation by SG isolation Loss of feed water 0 s 0 s Scram 6 s PR21=12.0 MPa Sec. bleed 127 s PR81=4.96 MPa Primary feed 513 s LE71=8.33 m Pump trip 1523 s PR21=7.6 MPa Figure 8: Mass flow rate in the primary side decreased when one row of tubes became free of water, as illustrated in Figure 8. This can be attributed to increasing SG outlet temperatures, and, hence to a smaller temperature difference between heat source (core simulator) and sink (SG). At approx s the temperature distribution in the upper tubes became close to uniform, which indicated that the heat transfer from the primary to secondary side was lost. Further evaporation of the secondary water and the increasing primary fluid temperature resulted in boiling of the primary side. The test was terminated at 7.6 MPa primary pressure when the water level in the SG sec. side was as low as 0.08 m. Table 3: Initial conditions of test TLFW Parameter Code Value Primary bleed 1523 s PR21=7.6 MPa Pump stopped 1673 s Pump trip s Test terminated 7886 s power used in this test followed the decay heat curve after the scram. The test included the modeling of the pump trip and coast-down. In order to promote the implementation of the accident management measures in the Paks NPP, primary bleedand-feed (using the pressurizer safety valve and the high pressure injection system, respectively) and bleed of the sec. side (by opening the SG relief valve) were modeled. The primary feed was assured by only 1 operable HPIS and it was triggered from the signal of "PRZ level 1 m below nominal". The primary and secondary bleed valves were modeled with orifices of 1 mm and 4 mm of diameter, respectively. These valves remained open after their actuation. Primary pressure PR ± 0.05 MPa Loop flow FL ± 0.06 kg/s Core inlet temp. TE ± 1 K Core power PW ± 3 kw Level in PRZ LE ± 0.02 m SG sec. pressure PR ± 0.02 MPa Level in SG secondary LE ± 0.05 m Feed water flow FL ± 0.02 kg/s Feed water temp. TE ± 1 K The PMK-2 TLFW Experiment The total loss of feed water transient started from the nominal operating parameters of the reference plant at time=0, when the feed water is lost. The initial conditions and the main occurrences of the PMK-2 TLFW test are listed in Tables 3 and 4, respectively. As a conservative assumption, the hydroaccumulators were considered to be inoperable. The core input Figure 9: Primary pressure and PRZ level The transient can be characterized by the variation of pressure in the upper plenum. As it is shown in Figure 9, PR21 decreases rapidly after the scram and this trend slows down until the secondary bleed (opening of PV23, t=127 s). Due to the heat sink and the lower secondary pressure, a further decrease was observed in PR21. The first characteristic point can be seen at the actuation of high pressure injection system (primary feed, t=513 s). Initially, the PRZ level decreases due to the lower 6

7 RELAP5 MODEL OF THE PACTEL AND THE PMK-2 Figure 10: Pressure in the SG secondary side coolant temperature. After the start-up of the HPIS, the PRZ level starts to rise again. Reaching the nominal level in the PRZ, the safety valve PV71 opens and the primary bleed begins at 1523 s. The fast depressurization continues until approximately 2100 s, when two-phase flow starts to be discharged. The maximal level in the PRZ is approached at 4000 s, followed by a nearly stabilized state. In this phase, the mass flow at the safety valve is roughly compensated by the primary feed. The pressure in the steam generator secondary side is depicted in Figure 10. The test results show that the SG is effective for the whole duration of the transient. Changing of the level in the primary side takes place only in the PRZ. The PACTEL Nodalization The nodalization scheme used in the calculations of the PACTEL LOF-01 test is illustrated in Figure 11. The core section was divided into four parallel channels, three of them were used to simulate the fuel rod bundles, while the fourth one is the bypass flow path between and around the fuel channels. The hot and cold legs were modeled with eleven and twelve nodes and one valve component, respectively. The pressurizer line, which is connected to one of the hot legs of the facility, was modeled with six nodes and one valve component. The pressurizer model included six nodes and a description of the pressurizer heaters as well as pressurizer spray system. The present steam generator model is based on former VVER-440 reactors studies developed by Karppinen and Tuunanen (1993) in the VTT. The six uppermost row of heat exchange tubes in the PACTEL steam generator were modeled separately, while three remaining rows were lumped together. The primary side hot and cold collectors were modeled with eight nodes each. In the secondary side the bypass flow path was divided into seven nodes. Heat structures were used to model the heat exchange tubes, pressurizer heaters, heater rods as well as all pipe walls, valves, and flanges in the primary and secondary side of the facility. In the simulation of the heat losses to the environment, a constant heat transfer coefficient together with a constant boundary volume temperature were specified as boundary condition. Figure 11: Nodalization scheme of the PACTEL facility 7

8 Code Input Description of the PMK-2 The basic structure of the nodalization scheme (Figure 12) was adopted from the former IAEA Standard Problem Exercise calculations, with a few modifications according to the needs of the TLFW test. (E.g. the steam line was removed and the pressurizer surge line was connected to the bottom of the hot leg). The model includes 93 volumes and 102 junctions The main features of the input can be summarized as follows: most of the hydrodynamic volumes were modeled with branches. Crossflow junctions are applied at connecting the following parts: downcomer to the lower plenum, upper plenum U-tubes, upper plenum to the hot leg, SIT connections, cold leg connection to the downcomer top, valve connections to the steam dome, the pressurizer safety valve and finally the steam generator hot and cold collector connections to the heat exchange tubes. The core is axially subdivided into 10 parts and rod bundle interphase friction model is used in the volumes of the heated section. Counter-current flow limitation (CCFL) model is turned on only in the core and the downcomer. Choking option is applied in the vicinity of the pressurizer safety valve (PV71), in order to simulate the primary bleed process. In fact, the pressurizer safety valve behaves like a break in the primary side during the bleed phase. Consequently, it is necessary to assure the adequate value for the valve discharge coefficient to match the correct amount of ejected primary coolant. Figure 12: Nodalization scheme of the PMK-2 facility The steam generator model includes basically three horizontal layers for the heat exchange area with three vertical volumes in each layer. The steam dome contains a separator component near the initial secondary water level. The original PMK-2 steam generator does not contain any steam separator. This arrangement helped to intensify the internal circulation. Based on the experiences of earlier calculations, void fraction limitation quantities of 0.35 and 0.55 gave the best results for the vapor exit junction and the liquid fall back junction, respectively. The current RELAP5 model allows the code user to have full control over appropriate simulation of the pump. The steadystate loop flow rate is achieved by regulation of the pump speed. In the transient the by-pass line is closed by the MV11 valve and the pump is running at constant speed until the pump trip. When the pump is tripped, the PV11 valve starts to close while the MV11 valve is gradually opening in a prescribed manner. Simulation of the pump coast-down is terminated by closing of the MV12 valve. The nominal length of this process is 150 s. Properties of the heat structures are calculated in the heater rod simulators, the steam generator tubes and all the pipe walls along the facility. For heat losses, convective boundary conditions were considered, with a constant heat transfer coefficient of 5 W/m 2 K. The safety injection tanks and the pressurizer spray line were modeled in the input, however, these components had no specific role during the transient. 8

9 Figure 13: Primary and secondary pressures ANALYSIS OF THE RELAP5 CODE CALCULATIONS PACTEL Transient LOF-01 Calculation of the experiment started with running the code with using steady-state option. Auxiliary components were added to maintain the required initial conditions. Level control systems provided the adequate initial water level in the pressurizer and in the SG secondary side. It took approximately 250 s process time to reach steadystate when the code stopped the calculation and all the parameters showed satisfactory matching with the measured initial data set. The primary pressure was controlled by switching the pressurizer medium power heater (4 kw) on and off between 7.3 MPa and 7.45 MPa, respectively. The alternating operation of the heater, which was maintained during the whole experiment, can be seen in Figure 13. Difficulties arose only in the simulation of the loop flow rate in the primary side. Lack of the main circulating pump disallowed the code user to maneuver with pump speed, therefore, the approximation of the mass flow rate took a few successive steps. The transient was initiated by eliminating the auxiliary components and cutting off the feed water injection to the SG secondary side. The first transient calculations resulted with Figure 14: Level in SG secondary side Figure 15: Time steps and Courant-limits deviation of the parameters from the measured values. Both primary and secondary temperatures started to rise within the first few hundred seconds, while the steam generator level was dropping sharply (Figure 14). The required time step was 0.4 s at this stage of the test runs. The output listing indicated that the code accepted this time step value and the Courant-limit was exceeded only a few cases, so the vast majority of the advancements were not necessary to be repeated (Figure 15). Several attempts were made in order to recover from errors. For instance, a test run was performed with a totally isolated steam generator secondary side. According to the calculations, the collapsed level was still dropping, despite the fact that there were no in and outflows. To investigate the reason of this contradiction, the calculated (numeric) mass errors have been checked. For the primary side this value was in the order of magnitude of 10-6, which is acceptable. However, the mass error in the secondary side exceeded the total system mass of the steam generator. This has been a clear sign of numerical problems. To examine the sensitivity of the results, several new calculations were performed with various requested time steps ()t). The conclusions can be summarized as follows: )t $ 0.40 s: numerical errors occur and the agreement is generally wrong, 0.40 s $ )t $ 0.08 s: the mass error is smaller but still relatively high. The calculated results show the same trend as in the previous case. With the required time step as small as 0.04 s, the whole transient is perfectly well simulated. Magnitude of the mass errors are negligible and basically most of the discrepancies are within the range of measurement uncertainties. None of the advancements were repeated. This value of time step is used in further discussion of the results. The time variation of the primary side mass flow rate is depicted in Figure 16. It became slightly underestimated shortly after the transient initiation, however, the oscillating nature of this parameter is very well repeated. The wavelength is close to the same value which was observed in the experiment. The main parameters in the primary and secondary side remained nearly stationary until the uppermost layer of the heat exchange tubes in the steam generator secondary side were uncovered, i.e. the 9

10 Figure 16: Mass flow rate in the primary loop swell level dropped below these tubes. This took place when the collapsed level was approximately 20 centimeters from the bottom of the SG. Superheating of the steam started at about 3000 s of transient time. The oscillation phenomenon is due to the decreasing water level. When the level reached a heat exchanger tube row in the SG, it also affected the mass flow rate in the primary side. The flow almost regained its original value after each period and this trend continued until the end of the transient. Figure 17 shows the behavior of the fluid temperature in the hot-leg. RELAP5 predicted the phenomenon very well. The characteristic break points are clearly visible on the experimental curve by a closer examination of the plots. This can be attributed to multi-layer structure of the SG model used in the calculation. There is a good correspondence between the elevation of the layers and uncovery of the actual heat exchange tube row. Degradation of the heat exchange in the SG resulted with increasing temperatures in the primary circuit. Matching of the calculated and test data is close to the accuracy of the measurement. One of the most significant shortcomings of the former code calculations using earlier versions (MOD3) of RELAP5 was that superheating of the steam could not be repeated in the simulation (Tuunanen, 1993). The present detailed nodalization is based on Figure 18: Temperatures in the steam line his work, with minor changes in the input deck. The new version enabled the code user to increase the accuracy of the calculations. This is particularly valid for behavior of the SG. As it is illustrated in Figure 18, the steam temperature is fairly well simulated in the steam line, except the first few hundred seconds at the beginning of superheating. At the end of the experiment the steam line temperature was approx. 25 o C higher than the secondary side saturation temperature. A test run was performed with using essentially the same input deck but the default (vertical) pipe flow correlations were replaced by horizontal convective and boiling correlations for the heat structures in the SG heat exchange tubes. The results of this calculation were practically identical to those of with using vertical option. The analysis of other parameters not plotted here confirms, that the MOD 3.2 version of RELAP5 code is capable of predicting the main events and trends in both sides of the facility with a satisfactory accuracy. Nevertheless, the analyst should pay special attention on the sophistication of the nodalization scheme. The analysis should always be extended with a detailed sensitivity study, in which the user carefully examines all the meaningful (but sometimes virtually hidden) information, such as time step issues discussed above. Figure 17: Fluid temperature in the hot leg RELAP5 Results of Experiment TLFW The initial data set and set-point values were derived from the nominal operating parameters of the VVER-440 plant. The strategies used to achieve steady-state with the code were similar to those of applied in the PACTEL analysis. Auxiliary components were connected to primary and secondary sides to stabilize the initial parameters at the same values as it was observed in the test. At the end of the search run, which took roughly 200 s, the code calculated basically an equal data set as of the initial conditions of TLFW. The requested time step was 0.1 s for the whole duration of the transient calculation. The transient started by the isolation of the SG secondary side. Due to the scram at 6 s of transient time depressurization 10

11 Figure 19: Pressures in the primary side of the primary system started rapidly. Figure 19 illustrates the time variation of measured and calculated values in the primary side. The set-point of the SG relief valve was reached at 127 s. At this time the valve PV23 opened and remained in open position for the rest of the transient. Bleeding of the secondary side resulted in a fast decreasing of pressure (Figure 20). While RELAP5 predicted the SG pressure at a high accuracy, the pressure in the upper plenum was well simulated only in the first part, until about 4000 s. The lack of slight increase of the pressure can be explained by the higher steam losses through the PRZ safety valve. Figure 21: Collapsed level in the pressurizer level in the PRZ fairly well. In the period, while there is one phase flow through the safety valve, is excellently simulated. The two phase flow was somewhat underestimated, resulting only about 0.1 m higher level in the PRZ. In the calculations homogeneous critical flow model was used with a discharge coefficient of Figure 22 shows the history of the temperature of the heater rods at the elevation of m, near the exit of the core. This parameter was also simulated with a high degree of accuracy. Both timing and the magnitude of the minor heat-excursion at 1800 s were well predicted. Figure 20: Pressures in the SG secondary Figure 22: Rod surface temp. at top of core By examination of Figure 21 the following events have particular interest. There was a continuous dropping of the collapsed level in the pressurizer until 513 s. At this time the primary side feed was actuated. Injection of cold water by HPIS started to fill up the pressurizer. The collapsed level reached the nominal value at 1523 s. At this moment the primary side bleed process was initiated by opening the pressurizer safety valve PV71. The fill-up process was terminated at approximately 2100 s, when the collapsed level reached the elevation of valve PV71. It can be concluded that RELAP5 predicted the variation of the CONCLUSIONS The analyses with RELAP5/MOD3.2 computer code showed that code is capable of predicting the key events in the experiment. It can be stated generally, that main trends have been repeated in the calculations and the qualitative agreement is good. Minor discrepancies were found in both experiments but these can be reduced by refinement of the input deck or the nodalization. The sensitivity of the results for the requested time step value should be underlined in the analysis of the PACTEL 11

12 experiment. The code user has always an opportunity to improve the modeling capabilities in a calculation of an experiment. However, time step issues become more difficult in the case of a real plant analysis when the comparison data base is limited or even nonexistent. The results obtained by this simulation has confirmed that the built-in automatic time step selection method of RELAP5 should be used with reservations and the value of mass error carries and important piece of information. Comparison of the results of LOF-01 showed that using the 3.2 version of the code improved the prediction of steam superheating. Temperatures were well simulated in both circuits of the PACTEL. Calculated mass flow rates indicated that RELAP5 repeated the natural circulation phenomenon fairly well. Concerning the PMK-2 experiment, it can be concluded that similar time step problems did not occur, even if the TLFW transient was more sophisticated. Matching of the calculated and measured conditions was excellent in the pressurizer. From thermal hydraulic aspects, very complex processes took place in this limited volume: feeding with cold ECC water and bleeding by the safety valve. The inventory loss through the bleed valve challenged the code because it behaved as a small break in the top of PRZ. RELAP5 calculated this process with a good accuracy. As a common conclusion for the steam generator models used in both calculations, it can be summarized as follows: simulation of the horizontal SGs with the code include some problems which are not present in the case of vertical SGs. In the primary side, it is not possible to model all heat exchange tubes separately. Hence, several tubes must be combined in the model. Normally this is done so that the tube bundle is divided into 3-5 parallel channels, each of which corresponds to several rows of horizontal U-tubes. Determination of the adequate flow rates from these combined heat exchange tubes is somewhat difficult. In the secondary side, problems arise due to the large coolant volume and due to the three-dimensional structure of the SG. More detailed discussion of the horizontal SG modeling can be found in Tuunanen s work (1993) and Hyvärinen s Doctoral Thesis (1996). REFERENCES Bánáti, J., 1995, "Assessment of RELAP5/MOD3.1 Code for the IAEA SPE-4 Experiment", Proceedings of the International Symposium on Validation of System Transient Analysis Codes, ASME & JSME Fluids Engineering Conference, Hilton Head, SC, USA. FED-Vol. 223, pp Bánáti, J., 1995, "Simulation of a Beyond Design-Basis- Accident with RELAP5/MOD3.1", Proceedings of the 7th International Topical Meeting on Nuclear Thermal Hydraulics (NURETH-7), Saratoga Springs, NY, USA, pp Bánáti, J. and Ézsöl, Gy., 1997, "Simulation of Some Emergency Operating Procedures for VVER-440 Reactors", Proceedings of the ASME International Mechanical Engineering Congress and Exposition, (IMECE-97), Dallas, TX, USA, NE- Vol. 21, pp Bánáti, J. and Ézsöl, Gy., 1997, "Modeling of Accident Management Procedures for VVER-440-Type Reactors", Proceedings the 8th International Topical Meeting on Nuclear Thermal Hydraulics (NURETH-8), Kyoto, Japan, pp Ézsöl, Gy., Perneczky, L. and Szabados, L.,1996, "Bleed and Feed Studies to Assess the Development of Emergency Operating Procedures for VVER-Type Reactors", Proceedings of the 4th International Conference on Nuclear Engineering (ICONE-4), New Orleans, USA, Vol. 3, pp Hyvärinen, J., 1996, "On the Fundamentals of Nuclear Reactor Safety Assessment, Inherent Threats and Their Implications", Doctor of Technology Thesis, Lappeenranta, Finland, STUK-A135 IAEA, 1987, "Simulation of a Loss of Coolant Accident", (SPE-1) IAEA TECDOC-425, Vienna, Austria IAEA, 1988, "Simulation of a Loss of Coolant Accident with Hydroaccumulator Injection", (SPE-2) IAEA TECDOC- 477, Vienna, Austria IAEA, 1991, "Simulation of a Loss of Coolant Accident with Rupture in the Steam Generator Hot Collector", (SPE-3) IAEA TECDOC-586, Vienna, Austria IAEA,1995, "Simulation of a Loss of Coolant Accident without High Pressure Injection but with Secondary Side Bleed and Feed. Results of the 4th Standard Problem Exercise", IAEA- TECDOC-848, Vienna, Austria Kouhia, J., 1998, "Loss of Feed Water Test LOF-01, Quick Look Report", VTT Energy Research Report, Finland, TEKOJA 7/97 Purhonen, H., Kouhia, J. and Kalli, H., 1995, "ISP33 Standard Problem on the PACTEL Facility", Proceedings of the 7th International Meeting on Nuclear reactor Thermal Hydraulics, (NURETH-7), Saratoga Springs, NY, USA, pp Tuomisto, H., et al. (editors), 1995, Proceedings of the Third International Seminar on Horizontal Staem Generators, Lappeenranta University of Technology, Finland, Research Papers, Vol. 43. Tuunanen, J., Nakada, K., 1993, "Analysis of PACTEL Loss of Secondary Side Feed Water with RELAP5/MOD3 Code", ASME Conference on "Transient Phenomena in Nuclear Reactor Systems", Atlanta, Georgia, USA, HTD-Vol. 245, NE- Vol. 11, pp