- Email: [email protected]
Natural convection test in Phenix reactor and associated CATHARE calculation
Nuclear Engineering and Design 253 (2012) 23–31
Contents lists available at SciVerse ScienceDirect
Nuclear Engineering and Design journal homepage: www.elsevier.com/locate/nucengdes
Natural convection test in Phenix reactor and associated CATHARE calculation D. Tenchine a,∗ , D. Pialla a , P. Gauthé b , A. Vasile b a b
CEA, DEN, DM2S, F-38054 Grenoble, France CEA, DEN, DER, F-13108 Saint Paul lez Durance, France
h i g h l i g h t s
Phenix reactor characteristics and instrumentation are briefly described. Phenix natural convection test scenario and main test results are presented. CATHARE modelling of Phenix primary circuit is depicted. Comparison of CATHARE simulation and Phenix data is presented and discussed.
a r t i c l e
i n f o
Article history: Received 15 May 2012 Received in revised form 20 July 2012 Accepted 2 August 2012
a b s t r a c t The Phenix sodium cooled fast reactor (SFR) started operation in 1973 and was stopped in 2009. Before the reactor was definitively stopped, ultimate tests were performed, including a natural convection test in the primary circuit. One objective of this natural convection test is the validation of plant dynamic codes as CATHARE code for future safety studies on SFRs. The paper firstly describes the Phenix pool type reactor primary circuit with the main components and the instrumentation used during the tests. The initial test conditions and the detailed transient scenario are presented: steam generators dry out, scram, stop of the primary pumps, development of natural convection in the primary circuit with two different phases. Then, CATHARE modelling of the Phenix primary circuit is described. The whole test is calculated for a total duration of 7 h in natural convection regime. The CATHARE calculations are compared to the Phenix measurements. A particular attention is paid to the significant decrease of the core power before the scram, due to the increase of temperature at the core inlet. Then, the computed evolution of main components inlet and outlet temperatures is compared to the reactor data. The need of coupling system code with CFD code to model the 3D behaviour of large pools during natural convection regime is pointed out. © 2012 Elsevier B.V. All rights reserved.
1. Introduction In the frame of Generation IV deployment, a sodium cooled fast reactor (SFR) prototype called Astrid is planned for 2020 (Carbonnier et al., 2007). One important safety challenge for Astrid and future reactors is the passive Decay Heat Removal (DHR) capability (Tenchine, 2010). In SFRs, passive DHR based on natural convection is possible (Hoffmann, 1989) thanks to sodium coolant characteristics. However, the reactor design must be adapted to facilitate natural convection and the efficiency and the reliability of natural convection must be checked. For the Astrid prototype, the CATHARE plant dynamics code is used to calculate the thermal hydraulic behaviour of the reactor during all transient situations. As the CATHARE code was initially
∗ Corresponding author. Tel.: +33 4 38 78 30 85; fax: +33 4 38 78 57 28. E-mail address: [email protected] (D. Tenchine). 0029-5493/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.nucengdes.2012.08.001
developed in the frame of pressurised water reactors (Bestion et al., 1999), a large amount of tests have been used for its validation in this domain of application. Then, the code was extended to various types of reactors (Geffraye et al., 2011) and specific adaptations have been developed for SFR applications (Tenchine et al., 2012b): implementation of sodium physical properties, adapted pressure drop and heat transfer correlations, fast reactor point kinetic model for neutronics. A validation process is currently underway with comparisons of code predictions to measured data from several reactor experiments, including the final tests in the Phenix reactor. Of course, the loss of flow situation and the transition to natural convection is one important case to be predicted by the CATHARE code. So, the ultimate natural convection test in Phenix reactor is quite valuable for the CATHARE code validation. The Phenix plant started in operation in 1973 and was stopped in 2009. It was previously decided to perform ultimate tests in Phenix reactor before its final shutdown. Among the ultimate tests, a natural convection test was selected as it is obviously one important
24
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
challenge for future reactors. The test was performed in June 2009. This test is used for an international benchmark exercise in the frame of an IAEA Coordinated Research Project and for the THINS (Thermal Hydraulics Innovative System) project on behalf of the 7th European Framework Program. The paper successively presents Phenix main features and instrumentation, the natural convection test conditions, some test results, CATHARE code main characteristics, the CATHARE modelling of Phenix reactor and CATHARE results compared to reactor data. 2. Presentation of Phenix reactor Phenix is a pool type sodium cooled fast reactor (Martin et al., 2009) composed of a main primary reactor vessel, three sodium secondary circuits, three tertiary water/steam circuits and one turbine, as depicted in Fig. 1. Phenix nominal power was 560 MWt/250 MWe. However, during the ultimate tests in 2009, one secondary circuit was out of operation and the reactor was operating at a reduced power. The reactor power was reduced to 120 MWt in order to ensure safe operation during the test. The reactor main vessel shown in Fig. 2 is 11.8 m in diameter and 9.8 m height, filled with 800 tons of sodium. A safety vessel surrounds the main vessel in case of sodium leakage. The space between the main vessel and the safety vessel is filled with nitrogen. The roof is a 0.06 m depth structure closing the top of the main vessel to ensure a tight barrier. Argon is filling the upper space between the sodium and the roof. A concrete structure is supporting the whole reactor vessel with the internal structures and the primary components. The main internal structures in the reactor vessel are the core support structure, the diagrid structure feeding the core, the above core structure, the handling machine and the internal vessel separating the hot pool and the cold pool. The main components are the three primary pumps and the six intermediate heat exchangers (IHX). The usual nominal core flow rate is 3000 kg/s at full power, but it was reduced to 1280 kg/s at 120 MWt before the natural convection test. The three primary pumps were in operation during the ultimate tests, but only four intermediate heat exchangers were in operation, the two other ones being replaced by inactive components called DOTE. The core is composed of three main regions: an inner core corresponding to a power of 60 MWt and a flow rate of 50% of the total core mass flow rate, an outer core corresponding to a power
of 50 MWt and a flow rate of 40% of the total core mass flow rate, a fertile zone corresponding to a power of 10 MWt and a flow rate of 10% of the total core mass flow rate (values given for the initial power of 120 MWt). Reflector and shielding zones are surrounding the main core region, with very low power release and very low flow rate. Six main control rods and one safety control rod allow the reactor power control. The intermediate heat exchangers are made of vertical straight tubes of 5.3 m long and 12 × 14 mm in diameter, tied to an upper plate and a lower plate. Primary sodium is flowing in the outer part of the tubes (shell side), with an inlet window in the hot pool and an outlet window in the cold pool. Three pipes are connecting the three primary pumps to the diagrid. A main vessel cooling circuit is arranged from the bottom to the top of the main vessel. A small flow rate of sodium (less than 10%) is taken at the core inlet to cool the main vessel and it is returned into the cold pool via a weir. At the core outlet, the flow rate through the above core structure is very low. The mean temperatures were approximately 360 ◦ C at the core inlet and 435 ◦ C at the core outlet for the 120 MWt initial state. Each secondary circuit depicted in Fig. 3 is composed mainly of two intermediate heat exchangers, a steam generator, a secondary pump and an expansion tank. Secondary sodium is distributed in the tubes of the IHX through an inlet collector. Then, an outlet collector collects the sodium from the tubes and connects the IHX to an outlet pipe. The two outlet pipes from the two IHX of the same circuit join together and sodium is flowing towards the steam generator. The steam generator is composed of three main parts: the evaporator, the heater and the super-heater. The steam generator sodium tubes are nearly 0.2 m in diameter with seven inserted water tubes. Sodium is flowing downwards as water is flowing upwards. The whole steam generator is surrounded by a casing which can be opened if required to ensure an upward natural convection air cooling in case of a steam generator dry out. From the steam generator outlet, secondary sodium is flowing upwards to the expansion tank where the secondary pump is located. Then, sodium is flowing back to the two IHXs through a main pipe which separates in two smaller pipes before reaching the two IHXs. The secondary flow rate is 690 kg/s per loop and the range of temperature in the secondary circuit is from 320 ◦ C to 520 ◦ C for 560 MWt. The water/steam tertiary circuit depicted in Fig. 1 is not described as CATHARE model uses boundary conditions measured on the secondary circuit.
Fig. 1. Phenix power plant.
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
25
Fig. 2. Phenix primary circuit.
Fig. 3. Phenix secondary circuit.
3. Instrumentation in Phenix reactor The standard instrumentation available in the reactor and a special instrumentation for the final tests were used simultaneously. The standard instrumentation consists in:
- Primary and secondary pumps speed which can be used to estimate the flow rate via the pumps characteristics, when the pumps are operating. - Secondary mass flow rate on each secondary loop. - Primary pumps inlet temperature. - Fuel subassemblies outlet temperature measured a few centimetres above each fuel subassembly. - Intermediate heat exchangers inlet and outlet temperatures on the primary side and on the secondary side.
- Steam generator inlet and outlet temperatures on the secondary side and on the tertiary side. One can note that the core inlet temperature is not directly measured. It can only be estimated by the measurement of the primary pumps inlet temperatures. A special instrumentation is used for the ultimate tests as shown in Fig. 2: - A permanent pole of thermocouples is immersed in the hot pool and penetrates in the inter-wrapper region, at a radius corresponding to the core edge. - A special pole of thermocouples was specially immersed in the hot pool from the free surface to the core exit level, at an intermediate radius between the above core structure and the core edge.
26
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
- A permanent pole of thermocouples is placed in the cold pool along one intermediate heat exchanger and it measures the vertical temperature profile in the cold pool. - A new pole of thermocouples was built and installed in the cold pool for the ultimate tests; this pole was supported by one of the two DOTE devices and it measured the vertical temperature profile in the cold pool. All the thermocouples mounted on the previous poles are one millimetre in diameter and they are in direct contact with the sodium. Fig. 4. Core power during phase 1.
4. Scenario of the natural convection test The ultimate natural convection test in Phenix reactor was performed on June 22–23, 2009. The test scenario is described hereafter: - The reactor is initially stabilized at a power of 120 MWt and a core inlet temperature of 360 ◦ C, with the three primary pumps in operation but one secondary circuit not operating. - Manual dry out of the two steam generators (reference time 0) without scram to reduce the temperature difference between the primary and secondary sodium at the inlet of the intermediate heat exchanger. This first part of the test corresponds to an unprotected loss of heat sink transient. - Manual scram at 460 s when the previous difference of temperature reaches 15 ◦ C. - Manual trip on the three primary pumps 8 s after the scram, with pumps speed decreasing to zero on their own inertia. - Secondary pumps speed decreases automatically to 110 rpm in about one minute after the scram. - Onset and development of natural convection in the primary circuit. The natural convection test in the primary circuit can be divided into three phases:
concerned with this core inlet temperature increase, such as the thermal expansion of the diagrid and the relative thermal expansion of the control rods. The well-known negative reactivity effect induced by the increase of temperature at the core inlet is clearly confirmed by this Phenix test. This is an important safety feature of sodium cooled fast reactors. Just before the scram, the pump inlet temperature has increased from 360 ◦ C to 405 ◦ C and the core outlet temperatures have decreased of nearly 10 ◦ C due to the power decrease. Thirty seconds after the scram, the core power is about 5 MW. The three primary pumps are tripped 8 s after the scram and the pump speed decreases to half-speed in 30 s and reaches zero within 2 min. The core outlet temperatures decrease sharply at the scram to 410 ◦ C within 1 min. Then, the outlet temperatures increases to 448 ◦ C for the hot fuel assemblies and 437 ◦ C for the whole fuel assemblies in about four minutes corresponding to the onset of natural convection in the primary circuit. About 5 min after the scram, a slight decrease of the core outlet temperature occurs, due to the core power decrease. In the same period, the pump inlet temperature decreases slowly from 405 ◦ C to 395 ◦ C in about 20 min due to the thermal inertia of the hot and cold pools. In longer term of the second phase, as there is no real heat sink, except the reactor thermal inertia and the heat losses in
- Phase 1: loss of heat sink (total loss of steam generator feed water after dry out), scram and trip of primary pumps. - Phase 2: no significant heat sink in the secondary circuits, except the heat losses along the piping and through the casing of the steam generator. - Phase 3: natural convection with an efficient heat sink by opening the casing at the bottom and at the top of the steam generator, which induces air natural circulation in the casing. The duration of phase 1 is about 7 min, the duration of the phase 2 is about 3 h and the duration of phase 3 is about 4 h. So, the total duration of the natural convection regime in the primary circuit is around 7 h.
Fig. 5. Pump inlet temperature (short term).
5. Main results of the natural convection test The core power evolution during the first phase of the test is given in Fig. 4. The primary pump inlet temperature is shown in Fig. 5 and core outlet temperatures (average of hot central fuel assemblies called “PX avg hot FA” and average of all fuel assemblies called “PX avg FA”) are depicted in Fig. 6. The first main important physical phenomenon is the core power decrease from 120 MWt to about 50 MWt before the scram as shown in Fig. 4. This significant decrease of power is induced by the temperature increase at the primary pump inlet temperature shown in Fig. 5, and therefore at the core inlet, after the steam generators dry out. Several reactivity feedback effects are
Fig. 6. Core outlet temperatures (short term).
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
Fig. 7. Pump inlet temperature (long term).
Fig. 8. Core outlet temperatures (long term).
the secondary circuits, the temperatures have a slow evolution between t = 1500 s and t = 10,500 s as shown in Fig. 7 for the pump inlet and Fig. 8 for the core outlet. This global reactor vessel behaviour lasts about three hours, until the end of the second phase. For the primary pump inlet temperature, after reaching the minimal value due to the scram, the temperature increases slowly to 400 ◦ C. Regarding to the core outlet temperatures, there is a slight decrease of about 10 ◦ C during these 3 h. One can notice that a radial gradient of temperature of 10 ◦ C remains at the fuel assemblies outlet. At t = 10,500 s, the third phase starts with the opening of the steam generators casing: the natural circulation of air in the casing generates significant heat losses which cool efficiently the secondary sodium. The IHX secondary inlet temperature decreases from 400 ◦ C to 340 ◦ C in about four hours. Consequently, the pump inlet temperature depicted in Fig. 7 decreases from 400 ◦ C to 355 ◦ C within the same time, as the core outlet temperatures presented in Fig. 8 decreases as well, from 435 ◦ C to 395 ◦ C for the hot fuel assemblies and from 425 ◦ C to 390 ◦ C for the whole fuel assemblies.
27
including thermodynamic and transport properties of sodium (liquid and vapor), heat transfer correlations adapted to sodium and pressure drop correlations for fuel pins with helical wires. The neutronics point kinetics model has been adapted to take into account specific neutronic feedback effects due to temperature variation, such as fuel, clad, wrapper tube, diagrid and control rod relative expansion. These new options are used for the Phenix test calculation. The core power can be computed by this neutronics point kinetics model or it can be issued as an input data from the reactor core power measurement during the test. Several Superphenix tests have already been used for CATHARE validation in forced or mixed convection regimes. The calculation of the natural convection test in Phenix reactor is a key milestone regarding to the whole qualification process of CATHARE code for SFR calculations. CATHARE qualification will include the calculation of other Phenix final tests performed in 2009 (Vasile et al., 2010). Moreover, Monju transient tests are also calculated to qualify the code on a loop type reactor (Pialla et al., 2009), complementary to the pool type concept. The CATHARE modeling of Phenix primary circuit (Pialla et al., 2011) is presented in Fig. 9. The CATHARE input deck is briefly described hereafter:
- The core is modeled with four parallel 1D pipes, one for the inner fuel core, one for the outer fuel core, one for the blanket zone and one for the reflector and shielding region. The specific Pontier correlation (Tenchine et al., 2012b) is used for the friction in the fuel pin zone with helical wires. - The core outlet region is modeled by a 0D module with two connections to take into account the hydraulic path of the hot sodium through the upper core structure and the direct path to the IHX primary inlet window. - The hot pool is modeled with several connected 0D volumes to roughly evaluate thermal stratification during the transient test. Heat transfer with the cold pool through the internal structure is taken into account. - The pressure of the cover gas is regulated. - The hydraulic path through the upper core structure is modeled by a 1D axial pipe directly connecting the hot core outlet region to the upper part of the hot pool. - The IHX is modeled with two 1D pipes, one simulating the primary side (shell side) and the other one the secondary side (tube side), with heat transfer between both sides. - The cold pool is modeled with several connected 0D volumes to be able to evaluate thermal stratification. The cold pool is divided into three angular sectors to deal with asymmetrical conditions. - The reactor vessel cooling system is modeled by two 1D pipes and a 0D volume, with vessel external heat transfer based on a correlation deduced from previous reactor measurements. - The diagrid and the lower plenum are modeled with 0D volumes. - The primary pumps are modeled by three 1D pipes and pump characteristics are input data.
6. CATHARE modelling of Phenix primary circuit The CATHARE system code has been developed in collaboration between CEA, EDF, AREVA and IRSN for more than 30 years. CATHARE is the reference code in France for the pressurized water reactor safety analysis. It has also been used for other light water reactor concepts and for experimental reactors. In the frame of the Generation IV, CATHARE code has already integrated new developments in order to calculate super critical light water reactors, gas cooled reactors, heavy liquid metal reactors (lead or lead–bismuth) and sodium cooled fast reactors (Geffraye et al., 2011). For SFR calculations, several developments were added to the standard version of CATHARE code (Tenchine et al., 2012b),
With the use of 0D volumes at the top of the hot pool, the cold pool and the reactor vessel cooling system, the code computes the three free levels in the upper part of the reactor vessel. The secondary circuits of the reactor are not included in the present calculation, except the secondary side of the IHX modeled by a 1D pipe. Boundary conditions from the secondary circuits are input data given at the IHX secondary inlet point thanks to the reactor measurements, namely secondary flow rate and temperature as a function of time. The other boundary conditions are the core power evolution and the primary pumps speed evolution after the scram.
28
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
Fig. 9. CATHARE modelling of Phenix primary circuit.
7. CATHARE results 7.1. Calculation of the initial steady state Table 1 gives the comparison between Phenix measurements and CATHARE computation for the initial steady state at 120 MWt. The small differences between the plant data and the code prediction remain below the sensors uncertainties. Therefore, we can point out that there is a good agreement between Phenix data and CATHARE code predictions for the initial steady state. The same comparison has been done successfully at full power of 560 MWt. For the steady-state calculation, the secondary flow rate is calculated with a set point on the primary pump inlet temperature. Due to the IHX efficiency at reduced power, the IHX outlet secondary temperature is closed to the IHX inlet primary temperature. 7.2. Calculation of the transient test Phase 1 starts with the steam generators dry out and it corresponds to an unprotected loss of the heat sink until the scram. The reactivity feedback effects conducting to the core power sharp decrease have been already depicted in Section 5. Fig. 10 shows the evolution of the core power for two CATHARE calculations:
- A first CATHARE calculation with a limited input deck called ‘CATHARE small’, modelling only the diagrid, the core channels, one mixing area at the core outlet, with the neutronics point kinetic model and given evolution of primary pumps mass flow rate and pump outlet temperature. The code prediction of the core power evolution is consistent with the Phenix data, as the core inlet temperature is an input to the code. - A second CATHARE calculation with an extended input deck called ‘CATHARE full’ modelling the whole primary circuit, with the same neutronics point kinetic model. In this case, the decrease of the core power predicted by the code is underestimated (60 MW instead of 50 MW), due to an underestimation of the core inlet temperature rise. In fact, the increase of the core inlet temperature is shortly delayed. These comparative results show that the neutronics point kinetic model in CATHARE code can predict the feedback effects during an unprotected loss of the heat sink. It also shows the difficulty for system codes to precisely calculate thermal inertia and real flow paths in a complicated reactor configuration. After the scram (phase 2 and phase 3), the residual core power is an input data for CATHARE issued from Phenix measurement.
Parameter
Phenix data
CATHARE results
Core power Primary pump speed Core inlet temperature Primary IHX inlet temperature Primary pump inlet temperature Secondary IHX mass flow rate Secondary IHX inlet temperature Secondary IHX outlet temperature Hot pool level Reactor cooling system level
120 MW 350 rpm 358 ◦ C 435 ◦ C 360 ◦ C 375 kg/s 307 ◦ C 430 ◦ C 1.88 m 1.65 m
120 MW 350 rpm 359 ◦ C 431 ◦ C 360 ◦ C 374 kg/s 307 ◦ C 430 ◦ C 1.88 m 1.65 m
Core power (MW)
140
Table 1 120 MW initial steady state (input data are in italic).
120 100 80 60
PHENIX CATHARE small
40 20
CATHARE full
0 0
100
200
300
400
Time (s) Fig. 10. Core power evolution during phase 1.
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
29
Fig. 11. Primary pump inlet temperature.
Fig. 12. Core outlet temperatures.
The comparison between CATHARE computation and Phenix data at short term and at longer term are depicted, for the primary pump inlet temperature in Fig. 11, for the core outlet temperatures in Fig. 12, for the IHX primary inlet and outlet temperatures in Fig. 13. Through the comparison between CATHARE results and Phenix data, we can point out the following comments. Concerning the primary pump inlet, as already noticed the main event during phase 1 is the heating of the lower part of the reactor vessel due to the dry out of the steam generators. The increase of the pump inlet temperature is slightly under-estimated as the system code modelling does not take into account the possible direct hydraulic path between the IHX outlet and pump inlet. Then during phase 2, the temperature decrease at the pump inlet after the scram is slightly over-estimated, probably due to some uncertainty on the cold pool thermal inertia and possible local buoyancy effects. Later in phase 2, the increase of the pump inlet temperature is correctly predicted. During phase 3, a delay is observed in the simulation before the cooling induced by the opening of the steam generators
casing. One can assume the system code does not take into account possible direct paths between the IHX outlet and the pump inlet with large buoyancy influence. Then, the slope of the temperature decrease is well predicted by the computation. Concerning the core outlet temperatures, one can observe an initial computed temperature lower than the Phenix one, as the modelling assumption is a full mixing of the whole core outlet fluid (including blanket and shielding regions) in a single volume, whereas in the reactor there is a significant radial gradient of temperature. During phase 1, the temperature decrease induced by the reactivity feedback effect is over-estimated by the computation as it is applied to the whole core. After the scram and the primary pumps trip, the increase of temperature due to the progressive onset of natural convection is slower than in the reactor, mainly for the same reason as explained just above. One can note the good agreement on the stabilized temperature at about 1200 s, when natural convection is established. Then, during phase 2 and phase 3, the code prediction is quite closed to the reactor core outlet average
Fig. 13. IHX inlet and outlet temperatures.
30
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
- In the second phase (natural convection with a weak heat sink), the modification of the lateral heat exchanges has some influence on the mass flow rate in the core. - In the third phase (natural convection with an efficient heat sink), the modification of the lateral heat exchanges has low impact on the mass flow rate in the core. A modification in the modelling of the cold plenum, with a direct path from the IHX outlet window to the primary pump suction head, has been tested:
Fig. 14. Computed core mass flow rate.
temperature. The change of slope due to the opening of the steam generators casing is well predicted. Concerning the IHX inlet and outlet temperatures, one can notice a rather good agreement with measured data until the scram, with just a slight under-estimation of the temperature level. At the scram, the system code does not predict the fall of the IHX outlet temperature which may be due to local buoyancy effects in the vicinity of the measurement. This possible explanation must be confirmed by future 3D computations, as shown by some participants of an IAEA Coordinated Research Project on Phenix natural convection test (Monti et al., 2012). During phase 2, the code prediction of the IHX inlet and outlet temperatures is correct. Then during phase 3, the decrease of temperature induced by the opening of the steam generators casing is under-estimated at the IHX inlet and over-estimated at the IHX outlet. It could mean that the code over-predicts the heat transfer through IHX during phase 3. But, possible thermal stratification and local buoyancy influence in the vicinity of the measurements or in the IHX itself (or both) could explain this discrepancy. Future 3D computations coupling CATHARE system code and TRIO U code (Tenchine et al., 2012a,b) will give more information on this point. The total core mass flow rate calculated with CATHARE is plotted in Fig. 14. Unfortunately, there is no reactor measurement of the core mass flow rate. One can see that the computed minimum value is around 15 kg/s after the primary pumps trip. Then, the flow rate increases progressively up to 40 kg/s during phase 1. After the opening of the steam generators casing, the core mass flow rate increases rapidly to 60 kg/s and then stabilizes at about 65 kg/s. This leads to an estimation of the natural circulation mass flow rate in the reactor vessel of 2–3% of the nominal mass flow rate.
7.3. Sensitivity calculations CATHARE code has shown it capability to simulate the different phases of this complicated natural convection test. Some discrepancies have been identified and discussed. To go further in the analysis, several sensitivity calculations have been performed and the main conclusions are pointed out hereafter. A modification of the friction law for the fuel pin area has been tested, with use of Rehme correlation instead of Pontier correlation. Rehme correlation leads to a smaller friction factor than Pontier correlation. There is no significant influence of this parameter. A modification of the heat exchange between the hot pool and the cold pool, and between the cold pool and the reactor vessel cooling system, has been tested:
- In the first phase, the primary pump inlet temperature increase is slightly better predicted. - In the second phase, the slow warm up of the lower part of the cold plenum is still under-predicted. - In the third phase, the global cooling of the reactor vessel remains nearly the same. Up to now, no key parameter has emerged from these sensitivity tests. The key point seems to be the difficulty for system code to represent the complicated 3D phenomena involved in the pools during such transient situations. The buoyancy effects play a key role with possible thermal stratification, recirculation and evolution of the flow paths, especially in the mixed and natural convection regimes. The system code limitations are well known by the thermal–hydraulic code users. A possible track to overtake these limitations is the coupling of system code and CFD code, such as the CATHARE – Trio U coupling, which is in progress at CEA. 8. Conclusion Before Phenix reactor was ultimately shutdown in 2009, ultimate tests were performed to provide valuable data for the development of future sodium cooled fast reactors as the so-called Astrid prototype in France. A special instrumentation was installed in the reactor before the tests, especially new poles of thermocouples in the hot and cold pools. The paper deals with the ultimate natural convection test in the primary circuit. The natural convection test consisted in a dry out of the steam generators (unprotected loss of heat sink) followed by a scram a few minutes later, immediately followed by a trip of the three primary pumps. The test duration after the pump trip was about seven hours. The test is used for the qualification of the CATHARE system code, developed at CEA in collaboration with AREVA, EDF and IRSN, and now applied to Astrid project. The test has shown the very important decrease of the core power from 120 MWt to about 50 MWt before the scram and the pump trip, thanks to the reactivity feedback effects induced by the increase of the core inlet temperature after the steam generator dry out. After the primary pumps trip, natural convection is established within about five minutes. The first phase without significant heat sink except thermal inertia and secondary heat losses lasted about 3 h and it shows a slight decrease of the core outlet temperature. The second phase with natural air cooling in the steam generators casing lasted about 4 h and it showed a higher decrease of the core outlet temperature and an increase of natural convection flow rate. The CATHARE code can predict the initial state of the Phenix reactor. The neutronics point kinetic model is able to estimate the neutronic feedback effects and the core power decrease before the scram, provided that the core inlet temperature is correctly calculated. The global prediction of CATHARE code is consistent with the Phenix measurements for the whole natural convection test. But it is pointed out that system codes have some difficulty to take into account the modification of the hydraulic paths in mixed or natural convection, the buoyancy effects like thermal stratification and
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
some 3D local phenomena. This kind of physical problem requires new approaches such as the coupling of system code and CFD code, as CATHARE and Trio-U coupling in progress at CEA. Acknowledgments The authors would like to acknowledge the Phenix reactor staff who prepared and performed the ultimate tests, including the natural convection test. Authors also acknowledge IAEA which retained the Phenix natural convection test for a Common Research Project. The last acknowledgement is for the THINS project on behalf of the 7th European Framework Program for the support in the test analysis. References Bestion, D., Barre, F., Faydide, B., 1999. Methodology, status and plans for development and assessment of CATHARE code. In: Proceedings of the International Conference on OECD/CSNI, Annapolis, USA, November 5–8. Carbonnier, J.L., Delbecq, J.M., Assedo, R., 2007. French program for sodium fast reactors. In: Proceedings of the International Congress on Advances in Nuclear Power Plants (ICAPP), Nice, France, May 14–18.
31
Geffraye, G., et al., 2011. CATHARE 2 V2.5 2: a single version for various applications. Nucl. Eng. Des. 241, 4456–4463. Hoffmann, H., 1989. Thermohydraulic investigations of decay heat removal systems by natural convection for liquid–metal fast breeder reactors. Nucl. Technol. 88 (1), 75–86. Martin, L., et al., 2009. Phenix: 35 Years Operation 1974–2009. In: Revue Générale du Nucléaire. Janvier-Février. Monti, S., et al., 2012. Overall results of lessons learned from the IAEA/CRP on sodium natural circulation test performed during Phenix end-of-life experiments. In: International Congress on Advances in Nuclear Power Plants (ICAPP’12), Chicago, USA, June 24–28. Pialla, D., et al., 2009. Monju calculations with French CATHARE system code. In: Proceeding of the Atomic Energy Society of Japan (AESJ) Annual Meeting, March 23–25. Pialla, D., et al., 2011. Natural convection test in Phenix reactor and associated CATHARE calculation. In: Proceeding of the 14th International Topical Meeting on Nuclear Reactor Thermal Hydraulics (NURETH- 14), Toronto, Canada, September 25–29. Tenchine, D., 2010. Some thermal hydraulic challenges in sodium cooled fast reactors. Nucl. Eng. Des. 240, 1195–1217. Tenchine, D., et al., 2012a. Status of TRIO U code for sodium cooled fast reactors. Nucl. Eng. Des. 242, 307–315. Tenchine, D., et al., 2012b. Status of CATHARE code for sodium cooled fast reactors. Nucl. Eng. Des. 245, 140–152. Vasile, A., et al., 2010. The Phenix final tests. In: Proceedings of the European Nuclear Conference (ENC), Barcelona, Spain, May 30–June 2.
Contents lists available at SciVerse ScienceDirect
Nuclear Engineering and Design journal homepage: www.elsevier.com/locate/nucengdes
Natural convection test in Phenix reactor and associated CATHARE calculation D. Tenchine a,∗ , D. Pialla a , P. Gauthé b , A. Vasile b a b
CEA, DEN, DM2S, F-38054 Grenoble, France CEA, DEN, DER, F-13108 Saint Paul lez Durance, France
h i g h l i g h t s
Phenix reactor characteristics and instrumentation are briefly described. Phenix natural convection test scenario and main test results are presented. CATHARE modelling of Phenix primary circuit is depicted. Comparison of CATHARE simulation and Phenix data is presented and discussed.
a r t i c l e
i n f o
Article history: Received 15 May 2012 Received in revised form 20 July 2012 Accepted 2 August 2012
a b s t r a c t The Phenix sodium cooled fast reactor (SFR) started operation in 1973 and was stopped in 2009. Before the reactor was definitively stopped, ultimate tests were performed, including a natural convection test in the primary circuit. One objective of this natural convection test is the validation of plant dynamic codes as CATHARE code for future safety studies on SFRs. The paper firstly describes the Phenix pool type reactor primary circuit with the main components and the instrumentation used during the tests. The initial test conditions and the detailed transient scenario are presented: steam generators dry out, scram, stop of the primary pumps, development of natural convection in the primary circuit with two different phases. Then, CATHARE modelling of the Phenix primary circuit is described. The whole test is calculated for a total duration of 7 h in natural convection regime. The CATHARE calculations are compared to the Phenix measurements. A particular attention is paid to the significant decrease of the core power before the scram, due to the increase of temperature at the core inlet. Then, the computed evolution of main components inlet and outlet temperatures is compared to the reactor data. The need of coupling system code with CFD code to model the 3D behaviour of large pools during natural convection regime is pointed out. © 2012 Elsevier B.V. All rights reserved.
1. Introduction In the frame of Generation IV deployment, a sodium cooled fast reactor (SFR) prototype called Astrid is planned for 2020 (Carbonnier et al., 2007). One important safety challenge for Astrid and future reactors is the passive Decay Heat Removal (DHR) capability (Tenchine, 2010). In SFRs, passive DHR based on natural convection is possible (Hoffmann, 1989) thanks to sodium coolant characteristics. However, the reactor design must be adapted to facilitate natural convection and the efficiency and the reliability of natural convection must be checked. For the Astrid prototype, the CATHARE plant dynamics code is used to calculate the thermal hydraulic behaviour of the reactor during all transient situations. As the CATHARE code was initially
∗ Corresponding author. Tel.: +33 4 38 78 30 85; fax: +33 4 38 78 57 28. E-mail address: [email protected] (D. Tenchine). 0029-5493/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.nucengdes.2012.08.001
developed in the frame of pressurised water reactors (Bestion et al., 1999), a large amount of tests have been used for its validation in this domain of application. Then, the code was extended to various types of reactors (Geffraye et al., 2011) and specific adaptations have been developed for SFR applications (Tenchine et al., 2012b): implementation of sodium physical properties, adapted pressure drop and heat transfer correlations, fast reactor point kinetic model for neutronics. A validation process is currently underway with comparisons of code predictions to measured data from several reactor experiments, including the final tests in the Phenix reactor. Of course, the loss of flow situation and the transition to natural convection is one important case to be predicted by the CATHARE code. So, the ultimate natural convection test in Phenix reactor is quite valuable for the CATHARE code validation. The Phenix plant started in operation in 1973 and was stopped in 2009. It was previously decided to perform ultimate tests in Phenix reactor before its final shutdown. Among the ultimate tests, a natural convection test was selected as it is obviously one important
24
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
challenge for future reactors. The test was performed in June 2009. This test is used for an international benchmark exercise in the frame of an IAEA Coordinated Research Project and for the THINS (Thermal Hydraulics Innovative System) project on behalf of the 7th European Framework Program. The paper successively presents Phenix main features and instrumentation, the natural convection test conditions, some test results, CATHARE code main characteristics, the CATHARE modelling of Phenix reactor and CATHARE results compared to reactor data. 2. Presentation of Phenix reactor Phenix is a pool type sodium cooled fast reactor (Martin et al., 2009) composed of a main primary reactor vessel, three sodium secondary circuits, three tertiary water/steam circuits and one turbine, as depicted in Fig. 1. Phenix nominal power was 560 MWt/250 MWe. However, during the ultimate tests in 2009, one secondary circuit was out of operation and the reactor was operating at a reduced power. The reactor power was reduced to 120 MWt in order to ensure safe operation during the test. The reactor main vessel shown in Fig. 2 is 11.8 m in diameter and 9.8 m height, filled with 800 tons of sodium. A safety vessel surrounds the main vessel in case of sodium leakage. The space between the main vessel and the safety vessel is filled with nitrogen. The roof is a 0.06 m depth structure closing the top of the main vessel to ensure a tight barrier. Argon is filling the upper space between the sodium and the roof. A concrete structure is supporting the whole reactor vessel with the internal structures and the primary components. The main internal structures in the reactor vessel are the core support structure, the diagrid structure feeding the core, the above core structure, the handling machine and the internal vessel separating the hot pool and the cold pool. The main components are the three primary pumps and the six intermediate heat exchangers (IHX). The usual nominal core flow rate is 3000 kg/s at full power, but it was reduced to 1280 kg/s at 120 MWt before the natural convection test. The three primary pumps were in operation during the ultimate tests, but only four intermediate heat exchangers were in operation, the two other ones being replaced by inactive components called DOTE. The core is composed of three main regions: an inner core corresponding to a power of 60 MWt and a flow rate of 50% of the total core mass flow rate, an outer core corresponding to a power
of 50 MWt and a flow rate of 40% of the total core mass flow rate, a fertile zone corresponding to a power of 10 MWt and a flow rate of 10% of the total core mass flow rate (values given for the initial power of 120 MWt). Reflector and shielding zones are surrounding the main core region, with very low power release and very low flow rate. Six main control rods and one safety control rod allow the reactor power control. The intermediate heat exchangers are made of vertical straight tubes of 5.3 m long and 12 × 14 mm in diameter, tied to an upper plate and a lower plate. Primary sodium is flowing in the outer part of the tubes (shell side), with an inlet window in the hot pool and an outlet window in the cold pool. Three pipes are connecting the three primary pumps to the diagrid. A main vessel cooling circuit is arranged from the bottom to the top of the main vessel. A small flow rate of sodium (less than 10%) is taken at the core inlet to cool the main vessel and it is returned into the cold pool via a weir. At the core outlet, the flow rate through the above core structure is very low. The mean temperatures were approximately 360 ◦ C at the core inlet and 435 ◦ C at the core outlet for the 120 MWt initial state. Each secondary circuit depicted in Fig. 3 is composed mainly of two intermediate heat exchangers, a steam generator, a secondary pump and an expansion tank. Secondary sodium is distributed in the tubes of the IHX through an inlet collector. Then, an outlet collector collects the sodium from the tubes and connects the IHX to an outlet pipe. The two outlet pipes from the two IHX of the same circuit join together and sodium is flowing towards the steam generator. The steam generator is composed of three main parts: the evaporator, the heater and the super-heater. The steam generator sodium tubes are nearly 0.2 m in diameter with seven inserted water tubes. Sodium is flowing downwards as water is flowing upwards. The whole steam generator is surrounded by a casing which can be opened if required to ensure an upward natural convection air cooling in case of a steam generator dry out. From the steam generator outlet, secondary sodium is flowing upwards to the expansion tank where the secondary pump is located. Then, sodium is flowing back to the two IHXs through a main pipe which separates in two smaller pipes before reaching the two IHXs. The secondary flow rate is 690 kg/s per loop and the range of temperature in the secondary circuit is from 320 ◦ C to 520 ◦ C for 560 MWt. The water/steam tertiary circuit depicted in Fig. 1 is not described as CATHARE model uses boundary conditions measured on the secondary circuit.
Fig. 1. Phenix power plant.
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
25
Fig. 2. Phenix primary circuit.
Fig. 3. Phenix secondary circuit.
3. Instrumentation in Phenix reactor The standard instrumentation available in the reactor and a special instrumentation for the final tests were used simultaneously. The standard instrumentation consists in:
- Primary and secondary pumps speed which can be used to estimate the flow rate via the pumps characteristics, when the pumps are operating. - Secondary mass flow rate on each secondary loop. - Primary pumps inlet temperature. - Fuel subassemblies outlet temperature measured a few centimetres above each fuel subassembly. - Intermediate heat exchangers inlet and outlet temperatures on the primary side and on the secondary side.
- Steam generator inlet and outlet temperatures on the secondary side and on the tertiary side. One can note that the core inlet temperature is not directly measured. It can only be estimated by the measurement of the primary pumps inlet temperatures. A special instrumentation is used for the ultimate tests as shown in Fig. 2: - A permanent pole of thermocouples is immersed in the hot pool and penetrates in the inter-wrapper region, at a radius corresponding to the core edge. - A special pole of thermocouples was specially immersed in the hot pool from the free surface to the core exit level, at an intermediate radius between the above core structure and the core edge.
26
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
- A permanent pole of thermocouples is placed in the cold pool along one intermediate heat exchanger and it measures the vertical temperature profile in the cold pool. - A new pole of thermocouples was built and installed in the cold pool for the ultimate tests; this pole was supported by one of the two DOTE devices and it measured the vertical temperature profile in the cold pool. All the thermocouples mounted on the previous poles are one millimetre in diameter and they are in direct contact with the sodium. Fig. 4. Core power during phase 1.
4. Scenario of the natural convection test The ultimate natural convection test in Phenix reactor was performed on June 22–23, 2009. The test scenario is described hereafter: - The reactor is initially stabilized at a power of 120 MWt and a core inlet temperature of 360 ◦ C, with the three primary pumps in operation but one secondary circuit not operating. - Manual dry out of the two steam generators (reference time 0) without scram to reduce the temperature difference between the primary and secondary sodium at the inlet of the intermediate heat exchanger. This first part of the test corresponds to an unprotected loss of heat sink transient. - Manual scram at 460 s when the previous difference of temperature reaches 15 ◦ C. - Manual trip on the three primary pumps 8 s after the scram, with pumps speed decreasing to zero on their own inertia. - Secondary pumps speed decreases automatically to 110 rpm in about one minute after the scram. - Onset and development of natural convection in the primary circuit. The natural convection test in the primary circuit can be divided into three phases:
concerned with this core inlet temperature increase, such as the thermal expansion of the diagrid and the relative thermal expansion of the control rods. The well-known negative reactivity effect induced by the increase of temperature at the core inlet is clearly confirmed by this Phenix test. This is an important safety feature of sodium cooled fast reactors. Just before the scram, the pump inlet temperature has increased from 360 ◦ C to 405 ◦ C and the core outlet temperatures have decreased of nearly 10 ◦ C due to the power decrease. Thirty seconds after the scram, the core power is about 5 MW. The three primary pumps are tripped 8 s after the scram and the pump speed decreases to half-speed in 30 s and reaches zero within 2 min. The core outlet temperatures decrease sharply at the scram to 410 ◦ C within 1 min. Then, the outlet temperatures increases to 448 ◦ C for the hot fuel assemblies and 437 ◦ C for the whole fuel assemblies in about four minutes corresponding to the onset of natural convection in the primary circuit. About 5 min after the scram, a slight decrease of the core outlet temperature occurs, due to the core power decrease. In the same period, the pump inlet temperature decreases slowly from 405 ◦ C to 395 ◦ C in about 20 min due to the thermal inertia of the hot and cold pools. In longer term of the second phase, as there is no real heat sink, except the reactor thermal inertia and the heat losses in
- Phase 1: loss of heat sink (total loss of steam generator feed water after dry out), scram and trip of primary pumps. - Phase 2: no significant heat sink in the secondary circuits, except the heat losses along the piping and through the casing of the steam generator. - Phase 3: natural convection with an efficient heat sink by opening the casing at the bottom and at the top of the steam generator, which induces air natural circulation in the casing. The duration of phase 1 is about 7 min, the duration of the phase 2 is about 3 h and the duration of phase 3 is about 4 h. So, the total duration of the natural convection regime in the primary circuit is around 7 h.
Fig. 5. Pump inlet temperature (short term).
5. Main results of the natural convection test The core power evolution during the first phase of the test is given in Fig. 4. The primary pump inlet temperature is shown in Fig. 5 and core outlet temperatures (average of hot central fuel assemblies called “PX avg hot FA” and average of all fuel assemblies called “PX avg FA”) are depicted in Fig. 6. The first main important physical phenomenon is the core power decrease from 120 MWt to about 50 MWt before the scram as shown in Fig. 4. This significant decrease of power is induced by the temperature increase at the primary pump inlet temperature shown in Fig. 5, and therefore at the core inlet, after the steam generators dry out. Several reactivity feedback effects are
Fig. 6. Core outlet temperatures (short term).
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
Fig. 7. Pump inlet temperature (long term).
Fig. 8. Core outlet temperatures (long term).
the secondary circuits, the temperatures have a slow evolution between t = 1500 s and t = 10,500 s as shown in Fig. 7 for the pump inlet and Fig. 8 for the core outlet. This global reactor vessel behaviour lasts about three hours, until the end of the second phase. For the primary pump inlet temperature, after reaching the minimal value due to the scram, the temperature increases slowly to 400 ◦ C. Regarding to the core outlet temperatures, there is a slight decrease of about 10 ◦ C during these 3 h. One can notice that a radial gradient of temperature of 10 ◦ C remains at the fuel assemblies outlet. At t = 10,500 s, the third phase starts with the opening of the steam generators casing: the natural circulation of air in the casing generates significant heat losses which cool efficiently the secondary sodium. The IHX secondary inlet temperature decreases from 400 ◦ C to 340 ◦ C in about four hours. Consequently, the pump inlet temperature depicted in Fig. 7 decreases from 400 ◦ C to 355 ◦ C within the same time, as the core outlet temperatures presented in Fig. 8 decreases as well, from 435 ◦ C to 395 ◦ C for the hot fuel assemblies and from 425 ◦ C to 390 ◦ C for the whole fuel assemblies.
27
including thermodynamic and transport properties of sodium (liquid and vapor), heat transfer correlations adapted to sodium and pressure drop correlations for fuel pins with helical wires. The neutronics point kinetics model has been adapted to take into account specific neutronic feedback effects due to temperature variation, such as fuel, clad, wrapper tube, diagrid and control rod relative expansion. These new options are used for the Phenix test calculation. The core power can be computed by this neutronics point kinetics model or it can be issued as an input data from the reactor core power measurement during the test. Several Superphenix tests have already been used for CATHARE validation in forced or mixed convection regimes. The calculation of the natural convection test in Phenix reactor is a key milestone regarding to the whole qualification process of CATHARE code for SFR calculations. CATHARE qualification will include the calculation of other Phenix final tests performed in 2009 (Vasile et al., 2010). Moreover, Monju transient tests are also calculated to qualify the code on a loop type reactor (Pialla et al., 2009), complementary to the pool type concept. The CATHARE modeling of Phenix primary circuit (Pialla et al., 2011) is presented in Fig. 9. The CATHARE input deck is briefly described hereafter:
- The core is modeled with four parallel 1D pipes, one for the inner fuel core, one for the outer fuel core, one for the blanket zone and one for the reflector and shielding region. The specific Pontier correlation (Tenchine et al., 2012b) is used for the friction in the fuel pin zone with helical wires. - The core outlet region is modeled by a 0D module with two connections to take into account the hydraulic path of the hot sodium through the upper core structure and the direct path to the IHX primary inlet window. - The hot pool is modeled with several connected 0D volumes to roughly evaluate thermal stratification during the transient test. Heat transfer with the cold pool through the internal structure is taken into account. - The pressure of the cover gas is regulated. - The hydraulic path through the upper core structure is modeled by a 1D axial pipe directly connecting the hot core outlet region to the upper part of the hot pool. - The IHX is modeled with two 1D pipes, one simulating the primary side (shell side) and the other one the secondary side (tube side), with heat transfer between both sides. - The cold pool is modeled with several connected 0D volumes to be able to evaluate thermal stratification. The cold pool is divided into three angular sectors to deal with asymmetrical conditions. - The reactor vessel cooling system is modeled by two 1D pipes and a 0D volume, with vessel external heat transfer based on a correlation deduced from previous reactor measurements. - The diagrid and the lower plenum are modeled with 0D volumes. - The primary pumps are modeled by three 1D pipes and pump characteristics are input data.
6. CATHARE modelling of Phenix primary circuit The CATHARE system code has been developed in collaboration between CEA, EDF, AREVA and IRSN for more than 30 years. CATHARE is the reference code in France for the pressurized water reactor safety analysis. It has also been used for other light water reactor concepts and for experimental reactors. In the frame of the Generation IV, CATHARE code has already integrated new developments in order to calculate super critical light water reactors, gas cooled reactors, heavy liquid metal reactors (lead or lead–bismuth) and sodium cooled fast reactors (Geffraye et al., 2011). For SFR calculations, several developments were added to the standard version of CATHARE code (Tenchine et al., 2012b),
With the use of 0D volumes at the top of the hot pool, the cold pool and the reactor vessel cooling system, the code computes the three free levels in the upper part of the reactor vessel. The secondary circuits of the reactor are not included in the present calculation, except the secondary side of the IHX modeled by a 1D pipe. Boundary conditions from the secondary circuits are input data given at the IHX secondary inlet point thanks to the reactor measurements, namely secondary flow rate and temperature as a function of time. The other boundary conditions are the core power evolution and the primary pumps speed evolution after the scram.
28
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
Fig. 9. CATHARE modelling of Phenix primary circuit.
7. CATHARE results 7.1. Calculation of the initial steady state Table 1 gives the comparison between Phenix measurements and CATHARE computation for the initial steady state at 120 MWt. The small differences between the plant data and the code prediction remain below the sensors uncertainties. Therefore, we can point out that there is a good agreement between Phenix data and CATHARE code predictions for the initial steady state. The same comparison has been done successfully at full power of 560 MWt. For the steady-state calculation, the secondary flow rate is calculated with a set point on the primary pump inlet temperature. Due to the IHX efficiency at reduced power, the IHX outlet secondary temperature is closed to the IHX inlet primary temperature. 7.2. Calculation of the transient test Phase 1 starts with the steam generators dry out and it corresponds to an unprotected loss of the heat sink until the scram. The reactivity feedback effects conducting to the core power sharp decrease have been already depicted in Section 5. Fig. 10 shows the evolution of the core power for two CATHARE calculations:
- A first CATHARE calculation with a limited input deck called ‘CATHARE small’, modelling only the diagrid, the core channels, one mixing area at the core outlet, with the neutronics point kinetic model and given evolution of primary pumps mass flow rate and pump outlet temperature. The code prediction of the core power evolution is consistent with the Phenix data, as the core inlet temperature is an input to the code. - A second CATHARE calculation with an extended input deck called ‘CATHARE full’ modelling the whole primary circuit, with the same neutronics point kinetic model. In this case, the decrease of the core power predicted by the code is underestimated (60 MW instead of 50 MW), due to an underestimation of the core inlet temperature rise. In fact, the increase of the core inlet temperature is shortly delayed. These comparative results show that the neutronics point kinetic model in CATHARE code can predict the feedback effects during an unprotected loss of the heat sink. It also shows the difficulty for system codes to precisely calculate thermal inertia and real flow paths in a complicated reactor configuration. After the scram (phase 2 and phase 3), the residual core power is an input data for CATHARE issued from Phenix measurement.
Parameter
Phenix data
CATHARE results
Core power Primary pump speed Core inlet temperature Primary IHX inlet temperature Primary pump inlet temperature Secondary IHX mass flow rate Secondary IHX inlet temperature Secondary IHX outlet temperature Hot pool level Reactor cooling system level
120 MW 350 rpm 358 ◦ C 435 ◦ C 360 ◦ C 375 kg/s 307 ◦ C 430 ◦ C 1.88 m 1.65 m
120 MW 350 rpm 359 ◦ C 431 ◦ C 360 ◦ C 374 kg/s 307 ◦ C 430 ◦ C 1.88 m 1.65 m
Core power (MW)
140
Table 1 120 MW initial steady state (input data are in italic).
120 100 80 60
PHENIX CATHARE small
40 20
CATHARE full
0 0
100
200
300
400
Time (s) Fig. 10. Core power evolution during phase 1.
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
29
Fig. 11. Primary pump inlet temperature.
Fig. 12. Core outlet temperatures.
The comparison between CATHARE computation and Phenix data at short term and at longer term are depicted, for the primary pump inlet temperature in Fig. 11, for the core outlet temperatures in Fig. 12, for the IHX primary inlet and outlet temperatures in Fig. 13. Through the comparison between CATHARE results and Phenix data, we can point out the following comments. Concerning the primary pump inlet, as already noticed the main event during phase 1 is the heating of the lower part of the reactor vessel due to the dry out of the steam generators. The increase of the pump inlet temperature is slightly under-estimated as the system code modelling does not take into account the possible direct hydraulic path between the IHX outlet and pump inlet. Then during phase 2, the temperature decrease at the pump inlet after the scram is slightly over-estimated, probably due to some uncertainty on the cold pool thermal inertia and possible local buoyancy effects. Later in phase 2, the increase of the pump inlet temperature is correctly predicted. During phase 3, a delay is observed in the simulation before the cooling induced by the opening of the steam generators
casing. One can assume the system code does not take into account possible direct paths between the IHX outlet and the pump inlet with large buoyancy influence. Then, the slope of the temperature decrease is well predicted by the computation. Concerning the core outlet temperatures, one can observe an initial computed temperature lower than the Phenix one, as the modelling assumption is a full mixing of the whole core outlet fluid (including blanket and shielding regions) in a single volume, whereas in the reactor there is a significant radial gradient of temperature. During phase 1, the temperature decrease induced by the reactivity feedback effect is over-estimated by the computation as it is applied to the whole core. After the scram and the primary pumps trip, the increase of temperature due to the progressive onset of natural convection is slower than in the reactor, mainly for the same reason as explained just above. One can note the good agreement on the stabilized temperature at about 1200 s, when natural convection is established. Then, during phase 2 and phase 3, the code prediction is quite closed to the reactor core outlet average
Fig. 13. IHX inlet and outlet temperatures.
30
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
- In the second phase (natural convection with a weak heat sink), the modification of the lateral heat exchanges has some influence on the mass flow rate in the core. - In the third phase (natural convection with an efficient heat sink), the modification of the lateral heat exchanges has low impact on the mass flow rate in the core. A modification in the modelling of the cold plenum, with a direct path from the IHX outlet window to the primary pump suction head, has been tested:
Fig. 14. Computed core mass flow rate.
temperature. The change of slope due to the opening of the steam generators casing is well predicted. Concerning the IHX inlet and outlet temperatures, one can notice a rather good agreement with measured data until the scram, with just a slight under-estimation of the temperature level. At the scram, the system code does not predict the fall of the IHX outlet temperature which may be due to local buoyancy effects in the vicinity of the measurement. This possible explanation must be confirmed by future 3D computations, as shown by some participants of an IAEA Coordinated Research Project on Phenix natural convection test (Monti et al., 2012). During phase 2, the code prediction of the IHX inlet and outlet temperatures is correct. Then during phase 3, the decrease of temperature induced by the opening of the steam generators casing is under-estimated at the IHX inlet and over-estimated at the IHX outlet. It could mean that the code over-predicts the heat transfer through IHX during phase 3. But, possible thermal stratification and local buoyancy influence in the vicinity of the measurements or in the IHX itself (or both) could explain this discrepancy. Future 3D computations coupling CATHARE system code and TRIO U code (Tenchine et al., 2012a,b) will give more information on this point. The total core mass flow rate calculated with CATHARE is plotted in Fig. 14. Unfortunately, there is no reactor measurement of the core mass flow rate. One can see that the computed minimum value is around 15 kg/s after the primary pumps trip. Then, the flow rate increases progressively up to 40 kg/s during phase 1. After the opening of the steam generators casing, the core mass flow rate increases rapidly to 60 kg/s and then stabilizes at about 65 kg/s. This leads to an estimation of the natural circulation mass flow rate in the reactor vessel of 2–3% of the nominal mass flow rate.
7.3. Sensitivity calculations CATHARE code has shown it capability to simulate the different phases of this complicated natural convection test. Some discrepancies have been identified and discussed. To go further in the analysis, several sensitivity calculations have been performed and the main conclusions are pointed out hereafter. A modification of the friction law for the fuel pin area has been tested, with use of Rehme correlation instead of Pontier correlation. Rehme correlation leads to a smaller friction factor than Pontier correlation. There is no significant influence of this parameter. A modification of the heat exchange between the hot pool and the cold pool, and between the cold pool and the reactor vessel cooling system, has been tested:
- In the first phase, the primary pump inlet temperature increase is slightly better predicted. - In the second phase, the slow warm up of the lower part of the cold plenum is still under-predicted. - In the third phase, the global cooling of the reactor vessel remains nearly the same. Up to now, no key parameter has emerged from these sensitivity tests. The key point seems to be the difficulty for system code to represent the complicated 3D phenomena involved in the pools during such transient situations. The buoyancy effects play a key role with possible thermal stratification, recirculation and evolution of the flow paths, especially in the mixed and natural convection regimes. The system code limitations are well known by the thermal–hydraulic code users. A possible track to overtake these limitations is the coupling of system code and CFD code, such as the CATHARE – Trio U coupling, which is in progress at CEA. 8. Conclusion Before Phenix reactor was ultimately shutdown in 2009, ultimate tests were performed to provide valuable data for the development of future sodium cooled fast reactors as the so-called Astrid prototype in France. A special instrumentation was installed in the reactor before the tests, especially new poles of thermocouples in the hot and cold pools. The paper deals with the ultimate natural convection test in the primary circuit. The natural convection test consisted in a dry out of the steam generators (unprotected loss of heat sink) followed by a scram a few minutes later, immediately followed by a trip of the three primary pumps. The test duration after the pump trip was about seven hours. The test is used for the qualification of the CATHARE system code, developed at CEA in collaboration with AREVA, EDF and IRSN, and now applied to Astrid project. The test has shown the very important decrease of the core power from 120 MWt to about 50 MWt before the scram and the pump trip, thanks to the reactivity feedback effects induced by the increase of the core inlet temperature after the steam generator dry out. After the primary pumps trip, natural convection is established within about five minutes. The first phase without significant heat sink except thermal inertia and secondary heat losses lasted about 3 h and it shows a slight decrease of the core outlet temperature. The second phase with natural air cooling in the steam generators casing lasted about 4 h and it showed a higher decrease of the core outlet temperature and an increase of natural convection flow rate. The CATHARE code can predict the initial state of the Phenix reactor. The neutronics point kinetic model is able to estimate the neutronic feedback effects and the core power decrease before the scram, provided that the core inlet temperature is correctly calculated. The global prediction of CATHARE code is consistent with the Phenix measurements for the whole natural convection test. But it is pointed out that system codes have some difficulty to take into account the modification of the hydraulic paths in mixed or natural convection, the buoyancy effects like thermal stratification and
D. Tenchine et al. / Nuclear Engineering and Design 253 (2012) 23–31
some 3D local phenomena. This kind of physical problem requires new approaches such as the coupling of system code and CFD code, as CATHARE and Trio-U coupling in progress at CEA. Acknowledgments The authors would like to acknowledge the Phenix reactor staff who prepared and performed the ultimate tests, including the natural convection test. Authors also acknowledge IAEA which retained the Phenix natural convection test for a Common Research Project. The last acknowledgement is for the THINS project on behalf of the 7th European Framework Program for the support in the test analysis. References Bestion, D., Barre, F., Faydide, B., 1999. Methodology, status and plans for development and assessment of CATHARE code. In: Proceedings of the International Conference on OECD/CSNI, Annapolis, USA, November 5–8. Carbonnier, J.L., Delbecq, J.M., Assedo, R., 2007. French program for sodium fast reactors. In: Proceedings of the International Congress on Advances in Nuclear Power Plants (ICAPP), Nice, France, May 14–18.
31
Geffraye, G., et al., 2011. CATHARE 2 V2.5 2: a single version for various applications. Nucl. Eng. Des. 241, 4456–4463. Hoffmann, H., 1989. Thermohydraulic investigations of decay heat removal systems by natural convection for liquid–metal fast breeder reactors. Nucl. Technol. 88 (1), 75–86. Martin, L., et al., 2009. Phenix: 35 Years Operation 1974–2009. In: Revue Générale du Nucléaire. Janvier-Février. Monti, S., et al., 2012. Overall results of lessons learned from the IAEA/CRP on sodium natural circulation test performed during Phenix end-of-life experiments. In: International Congress on Advances in Nuclear Power Plants (ICAPP’12), Chicago, USA, June 24–28. Pialla, D., et al., 2009. Monju calculations with French CATHARE system code. In: Proceeding of the Atomic Energy Society of Japan (AESJ) Annual Meeting, March 23–25. Pialla, D., et al., 2011. Natural convection test in Phenix reactor and associated CATHARE calculation. In: Proceeding of the 14th International Topical Meeting on Nuclear Reactor Thermal Hydraulics (NURETH- 14), Toronto, Canada, September 25–29. Tenchine, D., 2010. Some thermal hydraulic challenges in sodium cooled fast reactors. Nucl. Eng. Des. 240, 1195–1217. Tenchine, D., et al., 2012a. Status of TRIO U code for sodium cooled fast reactors. Nucl. Eng. Des. 242, 307–315. Tenchine, D., et al., 2012b. Status of CATHARE code for sodium cooled fast reactors. Nucl. Eng. Des. 245, 140–152. Vasile, A., et al., 2010. The Phenix final tests. In: Proceedings of the European Nuclear Conference (ENC), Barcelona, Spain, May 30–June 2.