- Email: [email protected]
Computational fluid dynamics analysis of spallation target for experimental accelerator-driven transmutation system
Nuclear Engineering and Design 235 (2005) 761–772
Computational fluid dynamics analysis of spallation target for experimental accelerator-driven transmutation system Nam-il Tak∗ , Hans-Joachim Neitzel, Xu Cheng Forschungszentrum Karlsruhe, Postfach 3640, D-76021 Karlsruhe, Germany Received 29 August 2004; received in revised form 2 November 2004; accepted 4 November 2004
Abstract One of the key milestones in the roadmap of the European accelerator-driven transmutation system (ADS) is the design and construction of the European experimental ADS (XADS). The window spallation target unit in the lead–bismuth eutectic (LBE) cooled reactor system is one of the basic options considered in the preliminary design study of XADS within the PDS-XADS project, partly financed by the European Commission in the frame of the 5th Framework Programme (PDS-XADS). This paper presents the computational fluid dynamics (CFD) analysis and the main results achieved for this option focusing on the coolability of the window. Steady-state as well as transient behavior, including beam interrupts and three major accident scenarios, has been analyzed using the CFD code CFX 5.6 with an advanced turbulence model. The required boundary conditions were provided by a one-dimensional system code. Based on the CFD analysis, the window geometry was modified in order to achieve sufficient cooling capability of the window under normal operating conditions. The transient behavior of the window temperature under beam trip conditions shows the importance of the beam interrupt duration to the thermal stress load of the window structural material. Further transient analysis of three major accidental scenarios, i.e., beam focusing, loss of heat sink, and beam intensity jump, indicates that the beam focusing accident gives the most serious safety concern. In this case, window failure occurs in less than 1 s after the start of the beam focusing. © 2004 Elsevier B.V. All rights reserved.
1. Introduction Transmutation of long-lived radio-nuclides using an accelerator-driven system (ADS) is a promising solution for reducing the long-term radiotoxicity of nuclear wastes (DOE, 1999; European Technical Working ∗ Corresponding author. Tel.: +49 7247 824773; fax: +49 7247 824837. E-mail address: [email protected] (N.-i. Tak).
0029-5493/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.nucengdes.2004.11.007
Group on ADS, 2001; OECD/NEA, 2002). The realization of an industrial scale ADS requires the operational experience of a small scale experimental ADS. One of the key milestones in the European ADS roadmap (European Technical Working Group on ADS, 2001) is the design and construction of the European experimental ADS (XADS). At the present stage, European research institutions, industrial partners, and universities are working together on the preliminary design study of XADS within the PDS-XADS project, partly
762
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
financed by the European Commission in the frame of the 5th Framework Programme (PDS-XADS). The main objectives of the PDS-XADS project are to select the most promising technical options of XADS, to address the critical points of the entire ADS system (i.e., accelerator, spallation target unit, sub-critical core), to identify further research and development (R&D) needs, and to consolidate the road map of the development of the European XADS. Additionally, the studies will allow to maintain a high level of expertise on nuclear technology in Europe. Three different reactor core concepts are being taken into consideration, i.e., a small core (50 MW) cooled by lead–bismuth eutectic (LBE), a large (80 MW) LBEcooled concept, and a large (80 MW) gas-cooled concept (Carluec, 2003). Related to the large core concepts, three different designs of spallation target have been investigated (Coors et al., 2004). These are the window target in the LBE-cooled concept, the window target in the gas-cooled concept, and the windowless target in the LBE-cooled concept. In the frame of the European PDS-XADS project, the Forschungszentrum Karlsruhe was engaged in the thermal–hydraulic design analysis of the window target unit for the large LBE-cooled concept. In the thermal–hydraulic point of view, the main issue of the window target is the cooling capability of the window. The window is a thin physical barrier to separate the vacuum space for the proton beam from the liquid spallation material LBE. Due to spallation reactions, the window material undergoes a high thermal load as well as a high irradiation load. Therefore, special attention has to be paid to the cooling capability of the window. The existing studies (Buono et al., 1998; Cheng and Slessarev, 2000; Tak et al., 2001; Smith et al., 2003) pointed out the importance of the window cooling to the design of a spallation target with a window, also called window target. This paper presents numerical studies using the CFD technology and main results achieved for the window target of the large LBE-cooled XADS. The CFX 5.6 code (ANSYS Inc., 2003) has been used, which provides the possibility to use various advanced turbulence models combined with an advanced wall treatment, e.g., the shear stress transport (SST) turbulence model with the automatic wall treatment. An adequate selection of turbulence models is of crucial importance to the prediction of the window temperature, because flow
stagnation occurs just below the window center, where the largest thermal load exists. It was reported by Vieser et al. (2002) that the SST turbulence model provides a better performance than the k–ε and the k–ω turbulence models on the prediction of heat transfer in stagnation flow region. The automatic wall treatment allows the application of coarser meshes near walls than typical low Reynolds number treatments. In the present study, therefore, the SST turbulence model with the automatic wall treatment was applied. Since, the previous CFD studies on spallation targets focused mainly on steady-state performance, there is less knowledge about detailed window behavior under normal and abnormal transients. The present study covers both steady-state and transient conditions, including beam interrupt, beam focusing, loss of heat sink, and beam intensity jump.
2. Physical model The mechanical design of the spallation target unit is presented in Fig. 1. The target unit is located in the center of the XADS core and consists of several co-axis cylinders. It is enclosed by the main shell, which is the physical boundary between the spallation material (LBE) and the reactor coolant. In the center of the target unit, there is an evacuated central beam tube. A 600 MeV proton beam enters the beam tube from the top, penetrates the beam window and impinges on the upward flowing LBE. This produces a large heat deposition in LBE as well as in the window. About 70% of the beam energy is deposited as heat in the lower part of the target, the so-called ‘active part’. The rest of energy is contained in the particles escaping the target unit or in the binding energy of the nucleus of the spallation material. Fig. 2 shows the geometry of the lower part of the target, which is the main concern of the present paper. The deposited heat is removed in a heat exchanger at the upper end inside the target unit. A LBE flow inside the target unit is driven by buoyancy force. Therefore, a sufficient height and a small pressure loss are required to guarantee a sufficiently strong natural circulation. The total height of the target unit is 7.8 m. The 5 mm thick guide tube is placed to separate the downward flowing cold LBE and the upward flowing hot LBE. The lower end of the guide tube is referred as funnel.
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
763
Fig. 2. Geometry of the lower part of the XADS target.
Fig. 1. Mechanical design of window target unit for LBE cooled XADS (Batta et al., 2003).
A proper design of the funnel is important, because it affects the natural circulation flow in the target and the injection flow to the window. In the initial stage of the target design, the thickness of the window was chosen as 3 mm. One of the challenging issues of the window target design is the selection of the window material. The reference material for the present study is a modified 9Cr1Mo ferritic–martensitic steel (T91). All other structural parts are made of stainless steel SS 316L. For the thermal–hydraulic design, some design criteria were proposed by the PDS-XADS working group and summarized as below: • The maximum temperature of the beam window should be less than 525 ◦ C.
764
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
Fig. 3. Heat deposition distribution in the window (data from Travleev and Broeders (2003)).
• The maximum temperature of the window outer surface contacting LBE should not exceed 500 ◦ C, to avoid a significant corrosion damage of the window. • The temperature upper limit for the other structural parts is 450 ◦ C. • The maximum local flow velocity of LBE should be less than 2 m/s, to minimize the erosion damage of the structural material. The footprint of the proton beam has a circular shape with the radius of 8 cm and its intensity follows an elliptical radial distribution as 2 1/2 3I0 r Φ(r) = 1− (1) r0 2πr02 where I0 is the total proton beam current, r0 the radius of the beam, and r the radial distance from the axis of the beam. The heat deposition profile in the target unit was evaluated from the neutronic analysis by Travleev and Broeders (2003) using the MCNPX 2.4.0 code. Their results are summarized in Batta et al. (2003). In the MCNPX calculation, fine meshes were adopted near the window to satisfy the CFD requirements. Fig. 3 shows the heat deposition in the window. The radial distribution is close to that defined by Eq. (1). The maximum heat deposition density in the window reaches 0.14 kW/cm3 /mA on the center of the window outer surface, where flow stagnation is expected. Fig. 4 shows the heat deposition distribution in LBE. The maximum heat deposition density, 0.15 kW/cm3 /mA, is located ∼2 cm below the window center. The heat deposition
Fig. 4. Heat deposition distribution in LBE (data from Travleev and Broeders (2003)).
density is dramatically decreased outside the beam radius. During XADS operations, the beam intensity is increased with the fuel burnup to compensate the increasing sub-criticality of the reactor core. In the present work, the maximum beam current (=6 mA) from the XADS accelerator has been considered as the reference value for the target design. With the beam current 6 mA, the total heat deposition in the entire target and in the window is 2.8 MW and 39 kW, respectively. In case of the 3 mm thick window, the maximum heat flux at the window center, where local flow stagnation is expected, can be roughly estimated as 2.5 MW/m2 (=140 MW/m3 /mA × 6 mA × 0.003 m). Cooling of the window becomes, therefore, one of the key issues of the target design. 3. Numerical approach The present CFD analysis has been performed for the lower part of the target, shown in Fig. 2, in order to focus on the window cooling as much as possible with reasonable computational efforts. Boundary conditions of the lower part of the target are provided by one-dimensional system code, HERETA (Neitzel and Knebel, 2002). 3.1. Computational model An axi-symmetrical two-dimensional computation domain was chosen due to symmetric conditions. Structured grids were used for the present CFD anal-
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
765
The second order advection scheme was adopted for all the calculations. The second order backward Euler scheme, which is an implicit time stepping scheme with second order accuracy, was applied for transient simulations. 3.2. Boundary conditions
Fig. 5. Computation meshes near the window region.
ysis. As mentioned before, the SST turbulence model with the automatic wall treatment was adopted. The total number of grids used for the analysis is about 120,000. Fine meshes were applied to regions close to walls, particularly around the window in order that viscous sublayer integration is sufficiently accurate. The computation meshes near the window region are shown in Fig. 5. Fig. 6 shows the y+ distribution along the outer surface of the window. A grid sensitivity study confirmed that the grid size used for the present analysis is sufficiently fine.
Fig. 6. y+ profile along the outer surface of the window.
Since, the target is cooled by natural circulation, thermal–hydraulic conditions of incoming flow to the computational domain have to be calculated in advance. These boundary conditions were provided by the HERETA code. Under steady-state conditions, the predicted inlet flow rate and the inlet LBE temperature by the HERETA code are 192 kg/s and 233 ◦ C, respectively. Uniform velocity and temperature profiles at the inlet cross section were assumed. All external walls, except the main shell, were considered adiabatic. For the main shell, the temperature profile determined by the HERETA code was applied as fixed temperature boundary conditions. The predicted temperature of the main shell by HERETA is in the range of 256–275 ◦ C. 4. Results and discussions 4.1. Steady-state performance 4.1.1. Results of initial design At first, a CFX 5.6 calculation was performed for the initial design having a 3 mm thick window. Fig. 7 shows the predicted velocity profile in the lower part of the target. As expected, flow stagnation near the center of the window can be clearly seen. Flow separation occurs both in the downcomer and in the riser. In the downcomer, flow separation starts earlier than expected. Additional CFX 5.6 calculations by neglecting the buoyancy effect showed that this earlier separation is caused by the buoyancy effect. The predicted maximum LBE velocity is 1.28 m/s, which is considerably below the design limit 2 m/s. The corresponding temperature on the window surface is presented in Fig. 8. The maximum window temperature occurs in the center of the inner surface and is as high as 596 ◦ C. The maximum temperature drop across the window thickness is 123 ◦ C. The maximum temperature of the guide tube is 450 ◦ C, which is much lower than that of the window. Since, the pre-
766
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
Fig. 8. Window surface temperature of the initial design.
shows the CFX 5.6 results with a reduced window thickness. The definition of δ is shown in Fig. 10. The window thickness is gradually reduced in the direction to the window center without changing the thickness of the cylindrical part of the beam tube. By a reduction in the window thickness of 1 mm in the center, the maximum window temperature is reduced by 86 ◦ C and kept below the design limit. For the window design with 2 mm thickness in the center, the maximum temperature difference across the window thickness is 57 ◦ C, as shown in Fig. 11, which is less than one half of the initial design. The structural analysis by Zucchini and Turroni (2004) confirmed that the maximum stress for this modified window design is in the allowable range. Therefore, all the remaining calculations in this paper are made for the modified window design with 2 mm thickness in the center.
Fig. 7. Velocity profile of the initial design.
dicted maximum window temperature exceeds the design limit (525 ◦ C) significantly, modification of the initial design is inevitable. 4.1.2. Results of modified design One of the proposals for the modification to the initial design is to reduce the window thickness. Fig. 9
Fig. 9. Effect of window thickness on maximum window temperature.
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
767
Fig. 12. Effects of funnel size on mass flow rate and average LBE speed in funnel.
Fig. 10. Definition of δ for window thickness variation.
The maximum temperature of the guide tube is not changed by the modification of the window thickness. It should be noted that the predicted maximum temperature of the guide tube, 450 ◦ C, exactly corresponds to the design limit for the guide tube. However, further analysis showed that the possible ways to reduce the guide tube temperature would be the reduction in the tube wall thickness or the replacement of the present guide tube material SS 316L by another material of a higher thermal conductivity.
Fig. 11. Window temperature for modified design (2 mm thickness in center).
4.1.3. Effects of funnel size A series of HERETA calculations showed that the size of the funnel is an important design parameter affecting the natural circulation flow around the target. As shown in Fig. 12, in case of a wider funnel, the LBE flow rate is increased due to a smaller flow resistance. But the injection speed to the window is decreased due to the increase in the flow area. Both effects contribute oppositely to the window cooling. Therefore, detailed analysis is necessary to achieve an optimum size of the funnel. Fig. 13 shows the results of CFX 5.6 calculations for three different funnel sizes. When the funnel size is reduced by 20 mm from the reference funnel size (=140 mm), the maximum window temperature is reduced by 7 ◦ C. However, the temperature of the guide tube increases dramatically, because the heat generation rate in the guide tube is larger than that of
Fig. 13. Effects of funnel size on maximum temperatures of window and guide tube.
768
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
4.2. Transient performance under beam trip conditions
Fig. 14. Temperature profile on inner surface of window at BOC and EOC conditions.
the original design. Furthermore, it is deduced from a sensitivity study that a reduction in the guide tube thickness cannot reduce the maximum guide tube temperature down to the design limit. When the funnel size is increased by 20 mm, the maximum window temperature is increased by 21 ◦ C, and exceeds the design limit. Based on the above results, the reference funnel size is, therefore, considered as the optimum value from the thermal–hydraulic point of view. 4.1.4. Effects of burnup The beam intensity, and subsequently, the spallation heat source, is varied with the fuel burnup. The neutronic calculation (Batta et al., 2003) shows that during operation of a 1-batch-cycle of XADS the beam current varies from 2.47 mA at keff ≈ 0.97 (typical at the beginning of cycle, BOC) to 5.1 mA at keff ≈ 0.94 (typical at the end of cycle, EOC). The total amount of heat deposition in the target is changed accordingly from 1.2 to 2.4 MW. The increase in the heat deposition leads to the change of the steady-state LBE flow rate from 149 kg/s at BOC to 183 kg/s at EOC. The LBE temperature at the inlet of the present computational domain is changed from 214 ◦ C at BOC to 228 ◦ C at EOC. Certainly, the window temperature also varies with the fuel burnup. Based on the boundary conditions provided by the HERETA code, CFX 5.6 calculations were performed for both BOC and EOC conditions. The results are shown in Fig. 14. It can be seen that the maximum window temperature is far below the design limit during the entire fuel cycle (∼3 years).
The operating experience of existing accelerators shows that beam trips occur very frequently (Bauer et al., 2000; Pierini, 2003). Beam trips produce thermal cycle and cause additional stress, which would become one of the main issues of structural material failure. Improved accelerator performance is required for XADS. In the technical specification of the accelerator for XADS, the frequency of beam trips with a time duration longer than 1 s has to be kept below five interrupts per year (Mueller, 2003; Pierini, 2003). No special requirements are defined for the beam trips with a beam interrupt duration shorter than 1 s. The present analysis focuses on beam trips with an interrupt duration shorter than 1 s. Beam trips with four different interrupt durations, i.e., 0.1, 0.3, 0.5, and 1 s, were investigated. The HERETA calculation showed that the mass flow rate of LBE and the LBE temperature at the inlet of the computational domain are hardly changed during these beam trips with a short interrupt duration. In case of the beam trip with a interrupt duration of 1 s, the change of the LBE mass flow rate is less than 2% and the change of the LBE temperature at the exit of the heat exchanger is less than 1 ◦ C. Therefore, in the CFX 5.6 calculations the inlet boundary conditions were kept unchanged. Fig. 15 presents the predicted temperature at the central point of the inner and outer surface of the window. For a better understanding, the case of beam shut off was also simulated and presented. For all cases shown in Fig. 15, the beam interrupt starts at the time point t = 0 s, and the beam power recovers after the corresponding trip periods, i.e., beam interrupt duration. As expected, beam trips with a smaller trip period result in a smaller temperature drop. The window undergoes a temperature drop up to 207 ◦ C for the trip period of 1 s, whereas the maximum temperature drop is only 32 ◦ C for the trip period of 0.1 s. Fig. 15 also shows that the steepest slope is located between 0 and 0.5 s after the beam interrupt, regardless of the investigated interrupt duration. Therefore, additional CFX 5.6 calculations were performed with a much finer time step for the first 0.5 s after the beam interrupt, in order to more accurately evaluate the maximum temperature change rate. As indicated in Fig. 16,
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
769
Regarding to the maximum temperature change rate, beam trips with a trip period shorter than 1 s could also be crucial event affecting the integrity of the window although no attention was paid to these trips in the technical specification of the XADS accelerator. Moreover, it is expected that beam trips with a trip period shorter than 0.1 s do not have enough time to reach the maximum temperature change rate of 412 ◦ C/s. It is thus concluded that related to both the maximum temperature drop and the temperature change rate, beam trips with a trip period shorter than 0.1 s are less critical than those with a trip period longer than 0.1 s, and are not further analyzed in this work. 4.3. Safety analysis
the predicted maximum temperature change rate is as high as 412 ◦ C/s, at the center of the outer surface of the window. This maximum temperature change rate occurs at 0.1 s after the beam interrupt.
Three kinds of abnormal accidents, i.e., beam focusing, loss of heat sink, and beam intensity jump, are analyzed in the present study. The main purpose of this analysis is to provide basic knowledge to the design of safety systems, to avoid window failure. Although the melting temperature of the window material (T91) is far above 1000 ◦ C, window failure might occur prior to its melting point due to both mechanical and thermal loads. The results of the tensile test of fresh T91 (Groeschel, 2004) and the stress analysis (Zucchini and Turroni, 2004) indicate that the window could fail at about 700 ◦ C under normal operation conditions. More data under irradiation with LBE environment are required for a more accurate prediction of window failure. In the present work, however, 700 ◦ C is used as a rough estimation for the temperature, at which window failure occurs.
Fig. 16. Temperature change rate under beam trip conditions.
4.3.1. Beam focusing In order to expand the proton beam onto the target, the reference accelerator for XADS uses the so-called raster scanning method deflecting a pencil-like beam with fast magnets in a designed pattern to paint the target area (Mueller, 2003). Therefore, a focused beam accident will occur by failure of rastering magnets. There are four rastering magnets for beam scanning. One pair of magnets covers each axis of the footprint (i.e., circle with a radius of 8 cm). Therefore, if one of the rastering magnets fails, the proton beam will cover only the half of the entire footprint with the same beam power. In this case, the beam intensity becomes double for one half of the footprint.
Fig. 15. (a) Temperature behavior under beam trip conditions (center of inner surface of the window). (b) Temperature behavior under beam trip conditions (center of outer surface of the window).
770
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
Fig. 17. Window temperature behavior under beam focusing conditions.
An accident initiated by the failure of one rastering magnet was simulated. To simplify CFX 5.6 analysis, the computational model and boundary conditions used for the reference steady-state analysis were unchanged except the beam current, which was increased to 12 mA from the reference value (=6 mA). These assumptions, in particular, axi-symmetrical twodimensional assumption, are not realistic and will give conservative results. However, this conservative analysis would provide a basic understanding of the target performance under beam focusing conditions. As shown in Fig. 17, the maximum window temperature arrives 700 ◦ C in about 0.8 s after the failure of one magnet. The window failure will occur even at an earlier time point since the thermal stress is increased with the double beam intensity. The result of the tensile test indicates that the window will not fail in the first 0.1 s, because the maximum window temperature is lower than 550 ◦ C. Therefore, it can be concluded that window failure will occur at a time point between 0.1 and 0.8 s after the failure of one rastering magnet. 4.3.2. Loss of heat sink Under normal operation conditions, the spallation heat is transferred to the secondary oil loop through the heat exchanger in the target unit. The oil loop itself is also cooled by another cooling loop. Loss of heat sink accident is initiated, in case one of the cooling loops is inactive. In the present work, the worst event, i.e., complete loss of heat sink, has been considered. Fig. 18 shows the mass flow rate and the LBE temperature at the exit of the heat exchanger under a complete loss of heat sink case.
Fig. 18. HERETA results under loss of heat sink accident.
The LBE flow rate is slowly reduced. It is still kept 78% of the initial value after 400 s. However, the LBE temperature at the exit of the heat exchanger is sharply increased after the loss of heat sink and reaches as high as 700 ◦ C in about 400 s. For more detailed analysis of window temperature using CFX 5.6 steady-state assumption was made due to the slow change of the boundary conditions. CFX analysis was carried out for three time points (=100, 200, 300 s) based on the boundary conditions supplied by HERETA. Fig. 19 shows the maximum window temperature at 100, 200, and 300 s after the loss of heat sink. It can be seen that the window failure occurs in about 200 s. 4.3.3. Beam intensity jump The maximum beam current of the XDAS accelerator is 6 mA, which gives a sufficient margin to the maximum needs of the XADS operation. As pointed
Fig. 19. Maximum window temperature vs. time under loss of heat sink accident.
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
out above, the required beam current for the XADS operation is 2.47 mA at BOC and 5.1 mA at EOC. Safety analysis should also consider the scenario that during the normal operating condition, beam current is suddenly increased from the normal operating value to its maximum value, i.e., 6 mA. This scenario is called beam intensity jump in this paper. Since, the target is cooled by natural circulation, beam intensity jump may cause temperature peak in the window and could damage the window. To a beam intensity jump scenario much attention has been paid in safety analysis of ADS (Cheng et al., 2004; Chen et al., 2004). However, the existing studies did not analyze the behavior of the window in detail, although the integrity of the window is very important for the calculation of the reactivity feedback. The present analysis concentrates on thermal behavior of the window after beam intensity jump. Results of beam intensity jump events depend on the time when beam intensity jumps occur. The most probable beam intensity jump occurs during the normal operation period with full reactor power. Fig. 20 shows the mass flow rate and the LBE temperature at the exit of the heat exchanger when the beam current is suddenly increased from 3 to 6 mA. The mass flow rate reaches a small peak after ∼25 s and stabilizes to the steady-state value. It never drops down below the initial flow rate. The LBE temperature at the exit of the heat exchanger is increased by ∼20 ◦ C. The temperature peak in the target occurs because the natural circulation flow is not established sufficiently in the early stage of the beam intensity jump transient. In order to minimize the computing expenditure, one steady-state calculation was performed using CFX 5.6
Fig. 20. HERETA results under a beam jump from 3 to 6 mA.
771
with the most conservative assumptions to investigate the window temperature under a beam intensity jump during normal operation conditions with full reactor power. For this calculation, the cooling conditions are kept to BOC conditions but the beam current is changed from 2.47 to 6 mA. With these assumptions, CFX 5.6 predicted the maximum window temperature as high as 550 ◦ C only. Therefore, window failure is not expected for beam intensity jump events during normal operating period with full reactor power.
5. Summary and conclusions The window target unit for the large LBE-cooled concept is one of the basic options in the proposed spallation target designs for XADS. In the present work, CFD studies were made for this option with the emphasis on the cooling capability of the window. Steadystate as well as transient behavior was analyzed using the CFX 5.6 code with the advanced turbulence model. The results achieved are summarized as follows: • The maximum steady-state temperature of the initially proposed 3 mm thick window exceeds the design limit by 71 ◦ C for the reference condition of 6 mA beam current. By a reduction in the thickness of 1 mm in the center, the maximum window temperature is reduced to 510 ◦ C and is kept below the design limit. • The reference funnel size (=140 mm) is proven as an optimum value in terms of the cooling capability of the window and the guide tube. • For the proposed reference target design, the maximum window temperature is far below the design limit during the entire fuel cycle (∼3 years). Under steady-state operating conditions, the performance of the proposed reference design was also confirmed by the stress analysis. • At beam trip transient, a temperature drop of about 200 ◦ C is obtained in case of a trip period of 1 s. For all beam trips analyzed, the maximum temperature change rate, as high as 412 ◦ C/s, occurs at about 0.1 s after the start of the beam interrupt. Therefore, beam trips with a trip period less than 1 s could be crucial events affecting the integrity of the window, and need to be considered in the design of the XADS accelerator.
772
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
• Among all three accidental transients analyzed, i.e., beam focusing, loss of heat sink, and beam intensity jump, the scenario of beam focusing shows the most serious safety concern. In case of beam focusing, window failure occurs in less than 1 s after the start of the rastering magnet failure. In case of loss of heat sink, about 200 s time is remaining for activating safety systems, to avoid window failure. No window failure is expected under beam intensity jump transients. Nowadays, due to the fast development of computer hardware as well as software, more and more applications of CFD codes to nuclear engineering and design are envisaged. Although a CFD code is believed to be the best tool for the design and analysis of spallation targets, it should be noted that the validation of CFD codes is still not completed in specific areas such as turbulent heat transfer in liquid metals. Acknowledgements This study was carried out in the framework of the PDS-XADS project partially financed by the 5th Framework Programme of the European Commission (Contract number FIKW-CT2001-00179/key action: Nuclear Fission). References ANSYS Incorporated, 2003. CFX 5.6 Manuals. Batta, A., et al., 2003. Window target unit for the XADS lead–bismuth cooled primary system. In: Proceedings of the International Workshop on P&T and ADS Development 2003, Mol, Belgium, October 6–8. Bauer, G., et al., 2000. Description of SINQ and boundary conditions for MEGAPIE. In: Proceedings of the 1st MEGAPIE General Meeting, CEA, Cadarache, June 14–15. Buono, S., et al., 1998. Numerical studies related to the design of the beam target of the energy amplifier prototype. HLMC’98, vol. 1. Carluec, B., 2003. The European project PDS-XADS, “Preliminary design studies of an experimental accelerator-driven system”.
In: Proceedings of the International Workshop on P&T and ADS Development 2003, Mol, Belgium, October 6–8. Chen, X.-N., et al., 2004. Comparative transient analyses of ADS with conventional fast reactor fuel and ADT with advanced fertile free fuel. In: Proceedings of the OECD/NEA Fourth International Workshop on Utilization and Reliability of High Power Proton Accelerators, Daejeon, Korea, May 16–19. Cheng, X., Cahalan, J.E., Finck, P.J., 2004. Safety analysis of an accelerator-driven test facility. Nucl. Eng. Des. 229, 289– 306. Cheng, X., Slessarev, I., 2000. Thermal–hydraulic investigations on liquid metal target systems. Nucl. Eng. Des. 202, 297–310. Coors, D., et al., 2004. Target units for XADS primary system. EURADWASTE’04—Sixth European Commission Conference on the Management and Disposal of Radioactive Waste, Luxembourg, March 29–31. DOE, 1999. A report to congress—a roadmap for developing accelerator transmutation of waste (ATW) technology. DOE/RW-0519. European Technical Working Group on ADS, 2001. A European roadmap for developing accelerator-driven systems (ADS) for nuclear waste incineration. Groeschel, F., 2004. Private communications. Paul Scherrer Institut. Mueller, A.C., 2003. The PDS-XDS reference accelerator. In: Proceedings of the International Workshop on P&T and ADS Development 2003, Mol, Belgium, October 6–8. Neitzel, H.-J., Knebel, J.U., 2002. Auslegung eines geschlossenen 4 MW-Targetmoduls mit W¨armeabfuhrsystem f¨ur eine ADSAnordnung. Scientific Report FZKA 6687, Forschungszentrum Karlsruhe. OECD/NEA, 2002. Accelerator-driven system (ADS) and fast reactors (FR) in advanced nuclear fuel cycles—a comparative study. OECD Nuclear Energy Agency. Pierini, P., 2003. Reliability study of the PDS-XADS accelerator. In: Proceedings of the International Workshop on P&T and ADS Development 2003, Mol, Belgium, October 6–8. Smith, B.L., Dury, T.V., Maciocco, L., Roubin, P., Tak, N.I., 2003. A benchmark study based on a representative design of the MEGAPIE spallation source target. In: Proceedings of the 10th International Topical Meeting on Nuclear Reactor Thermal–Hydraulics (NURETH-10), Seoul, Korea, October 5–9. Tak, N.I., Song, T.Y., Park, W.S., 2001. Numerical studies on thermal–hydraulics of HYPER target. AccApp/ADTTA’01, Reno, Nevada. Travleev, A., Broeders, C.H.M., 2003. Private communications. Forschungszentrum Karlsruhe. Vieser, W., Esch, T., Menter, F., 2002. Heat transfer predictions using advanced two equation turbulence models. CFX Tech. Memorandum, CFX-VAL10/0602. Zucchini, A., Turroni, P., 2004. Private communications. ENEA.
Computational fluid dynamics analysis of spallation target for experimental accelerator-driven transmutation system Nam-il Tak∗ , Hans-Joachim Neitzel, Xu Cheng Forschungszentrum Karlsruhe, Postfach 3640, D-76021 Karlsruhe, Germany Received 29 August 2004; received in revised form 2 November 2004; accepted 4 November 2004
Abstract One of the key milestones in the roadmap of the European accelerator-driven transmutation system (ADS) is the design and construction of the European experimental ADS (XADS). The window spallation target unit in the lead–bismuth eutectic (LBE) cooled reactor system is one of the basic options considered in the preliminary design study of XADS within the PDS-XADS project, partly financed by the European Commission in the frame of the 5th Framework Programme (PDS-XADS). This paper presents the computational fluid dynamics (CFD) analysis and the main results achieved for this option focusing on the coolability of the window. Steady-state as well as transient behavior, including beam interrupts and three major accident scenarios, has been analyzed using the CFD code CFX 5.6 with an advanced turbulence model. The required boundary conditions were provided by a one-dimensional system code. Based on the CFD analysis, the window geometry was modified in order to achieve sufficient cooling capability of the window under normal operating conditions. The transient behavior of the window temperature under beam trip conditions shows the importance of the beam interrupt duration to the thermal stress load of the window structural material. Further transient analysis of three major accidental scenarios, i.e., beam focusing, loss of heat sink, and beam intensity jump, indicates that the beam focusing accident gives the most serious safety concern. In this case, window failure occurs in less than 1 s after the start of the beam focusing. © 2004 Elsevier B.V. All rights reserved.
1. Introduction Transmutation of long-lived radio-nuclides using an accelerator-driven system (ADS) is a promising solution for reducing the long-term radiotoxicity of nuclear wastes (DOE, 1999; European Technical Working ∗ Corresponding author. Tel.: +49 7247 824773; fax: +49 7247 824837. E-mail address: [email protected] (N.-i. Tak).
0029-5493/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.nucengdes.2004.11.007
Group on ADS, 2001; OECD/NEA, 2002). The realization of an industrial scale ADS requires the operational experience of a small scale experimental ADS. One of the key milestones in the European ADS roadmap (European Technical Working Group on ADS, 2001) is the design and construction of the European experimental ADS (XADS). At the present stage, European research institutions, industrial partners, and universities are working together on the preliminary design study of XADS within the PDS-XADS project, partly
762
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
financed by the European Commission in the frame of the 5th Framework Programme (PDS-XADS). The main objectives of the PDS-XADS project are to select the most promising technical options of XADS, to address the critical points of the entire ADS system (i.e., accelerator, spallation target unit, sub-critical core), to identify further research and development (R&D) needs, and to consolidate the road map of the development of the European XADS. Additionally, the studies will allow to maintain a high level of expertise on nuclear technology in Europe. Three different reactor core concepts are being taken into consideration, i.e., a small core (50 MW) cooled by lead–bismuth eutectic (LBE), a large (80 MW) LBEcooled concept, and a large (80 MW) gas-cooled concept (Carluec, 2003). Related to the large core concepts, three different designs of spallation target have been investigated (Coors et al., 2004). These are the window target in the LBE-cooled concept, the window target in the gas-cooled concept, and the windowless target in the LBE-cooled concept. In the frame of the European PDS-XADS project, the Forschungszentrum Karlsruhe was engaged in the thermal–hydraulic design analysis of the window target unit for the large LBE-cooled concept. In the thermal–hydraulic point of view, the main issue of the window target is the cooling capability of the window. The window is a thin physical barrier to separate the vacuum space for the proton beam from the liquid spallation material LBE. Due to spallation reactions, the window material undergoes a high thermal load as well as a high irradiation load. Therefore, special attention has to be paid to the cooling capability of the window. The existing studies (Buono et al., 1998; Cheng and Slessarev, 2000; Tak et al., 2001; Smith et al., 2003) pointed out the importance of the window cooling to the design of a spallation target with a window, also called window target. This paper presents numerical studies using the CFD technology and main results achieved for the window target of the large LBE-cooled XADS. The CFX 5.6 code (ANSYS Inc., 2003) has been used, which provides the possibility to use various advanced turbulence models combined with an advanced wall treatment, e.g., the shear stress transport (SST) turbulence model with the automatic wall treatment. An adequate selection of turbulence models is of crucial importance to the prediction of the window temperature, because flow
stagnation occurs just below the window center, where the largest thermal load exists. It was reported by Vieser et al. (2002) that the SST turbulence model provides a better performance than the k–ε and the k–ω turbulence models on the prediction of heat transfer in stagnation flow region. The automatic wall treatment allows the application of coarser meshes near walls than typical low Reynolds number treatments. In the present study, therefore, the SST turbulence model with the automatic wall treatment was applied. Since, the previous CFD studies on spallation targets focused mainly on steady-state performance, there is less knowledge about detailed window behavior under normal and abnormal transients. The present study covers both steady-state and transient conditions, including beam interrupt, beam focusing, loss of heat sink, and beam intensity jump.
2. Physical model The mechanical design of the spallation target unit is presented in Fig. 1. The target unit is located in the center of the XADS core and consists of several co-axis cylinders. It is enclosed by the main shell, which is the physical boundary between the spallation material (LBE) and the reactor coolant. In the center of the target unit, there is an evacuated central beam tube. A 600 MeV proton beam enters the beam tube from the top, penetrates the beam window and impinges on the upward flowing LBE. This produces a large heat deposition in LBE as well as in the window. About 70% of the beam energy is deposited as heat in the lower part of the target, the so-called ‘active part’. The rest of energy is contained in the particles escaping the target unit or in the binding energy of the nucleus of the spallation material. Fig. 2 shows the geometry of the lower part of the target, which is the main concern of the present paper. The deposited heat is removed in a heat exchanger at the upper end inside the target unit. A LBE flow inside the target unit is driven by buoyancy force. Therefore, a sufficient height and a small pressure loss are required to guarantee a sufficiently strong natural circulation. The total height of the target unit is 7.8 m. The 5 mm thick guide tube is placed to separate the downward flowing cold LBE and the upward flowing hot LBE. The lower end of the guide tube is referred as funnel.
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
763
Fig. 2. Geometry of the lower part of the XADS target.
Fig. 1. Mechanical design of window target unit for LBE cooled XADS (Batta et al., 2003).
A proper design of the funnel is important, because it affects the natural circulation flow in the target and the injection flow to the window. In the initial stage of the target design, the thickness of the window was chosen as 3 mm. One of the challenging issues of the window target design is the selection of the window material. The reference material for the present study is a modified 9Cr1Mo ferritic–martensitic steel (T91). All other structural parts are made of stainless steel SS 316L. For the thermal–hydraulic design, some design criteria were proposed by the PDS-XADS working group and summarized as below: • The maximum temperature of the beam window should be less than 525 ◦ C.
764
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
Fig. 3. Heat deposition distribution in the window (data from Travleev and Broeders (2003)).
• The maximum temperature of the window outer surface contacting LBE should not exceed 500 ◦ C, to avoid a significant corrosion damage of the window. • The temperature upper limit for the other structural parts is 450 ◦ C. • The maximum local flow velocity of LBE should be less than 2 m/s, to minimize the erosion damage of the structural material. The footprint of the proton beam has a circular shape with the radius of 8 cm and its intensity follows an elliptical radial distribution as 2 1/2 3I0 r Φ(r) = 1− (1) r0 2πr02 where I0 is the total proton beam current, r0 the radius of the beam, and r the radial distance from the axis of the beam. The heat deposition profile in the target unit was evaluated from the neutronic analysis by Travleev and Broeders (2003) using the MCNPX 2.4.0 code. Their results are summarized in Batta et al. (2003). In the MCNPX calculation, fine meshes were adopted near the window to satisfy the CFD requirements. Fig. 3 shows the heat deposition in the window. The radial distribution is close to that defined by Eq. (1). The maximum heat deposition density in the window reaches 0.14 kW/cm3 /mA on the center of the window outer surface, where flow stagnation is expected. Fig. 4 shows the heat deposition distribution in LBE. The maximum heat deposition density, 0.15 kW/cm3 /mA, is located ∼2 cm below the window center. The heat deposition
Fig. 4. Heat deposition distribution in LBE (data from Travleev and Broeders (2003)).
density is dramatically decreased outside the beam radius. During XADS operations, the beam intensity is increased with the fuel burnup to compensate the increasing sub-criticality of the reactor core. In the present work, the maximum beam current (=6 mA) from the XADS accelerator has been considered as the reference value for the target design. With the beam current 6 mA, the total heat deposition in the entire target and in the window is 2.8 MW and 39 kW, respectively. In case of the 3 mm thick window, the maximum heat flux at the window center, where local flow stagnation is expected, can be roughly estimated as 2.5 MW/m2 (=140 MW/m3 /mA × 6 mA × 0.003 m). Cooling of the window becomes, therefore, one of the key issues of the target design. 3. Numerical approach The present CFD analysis has been performed for the lower part of the target, shown in Fig. 2, in order to focus on the window cooling as much as possible with reasonable computational efforts. Boundary conditions of the lower part of the target are provided by one-dimensional system code, HERETA (Neitzel and Knebel, 2002). 3.1. Computational model An axi-symmetrical two-dimensional computation domain was chosen due to symmetric conditions. Structured grids were used for the present CFD anal-
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
765
The second order advection scheme was adopted for all the calculations. The second order backward Euler scheme, which is an implicit time stepping scheme with second order accuracy, was applied for transient simulations. 3.2. Boundary conditions
Fig. 5. Computation meshes near the window region.
ysis. As mentioned before, the SST turbulence model with the automatic wall treatment was adopted. The total number of grids used for the analysis is about 120,000. Fine meshes were applied to regions close to walls, particularly around the window in order that viscous sublayer integration is sufficiently accurate. The computation meshes near the window region are shown in Fig. 5. Fig. 6 shows the y+ distribution along the outer surface of the window. A grid sensitivity study confirmed that the grid size used for the present analysis is sufficiently fine.
Fig. 6. y+ profile along the outer surface of the window.
Since, the target is cooled by natural circulation, thermal–hydraulic conditions of incoming flow to the computational domain have to be calculated in advance. These boundary conditions were provided by the HERETA code. Under steady-state conditions, the predicted inlet flow rate and the inlet LBE temperature by the HERETA code are 192 kg/s and 233 ◦ C, respectively. Uniform velocity and temperature profiles at the inlet cross section were assumed. All external walls, except the main shell, were considered adiabatic. For the main shell, the temperature profile determined by the HERETA code was applied as fixed temperature boundary conditions. The predicted temperature of the main shell by HERETA is in the range of 256–275 ◦ C. 4. Results and discussions 4.1. Steady-state performance 4.1.1. Results of initial design At first, a CFX 5.6 calculation was performed for the initial design having a 3 mm thick window. Fig. 7 shows the predicted velocity profile in the lower part of the target. As expected, flow stagnation near the center of the window can be clearly seen. Flow separation occurs both in the downcomer and in the riser. In the downcomer, flow separation starts earlier than expected. Additional CFX 5.6 calculations by neglecting the buoyancy effect showed that this earlier separation is caused by the buoyancy effect. The predicted maximum LBE velocity is 1.28 m/s, which is considerably below the design limit 2 m/s. The corresponding temperature on the window surface is presented in Fig. 8. The maximum window temperature occurs in the center of the inner surface and is as high as 596 ◦ C. The maximum temperature drop across the window thickness is 123 ◦ C. The maximum temperature of the guide tube is 450 ◦ C, which is much lower than that of the window. Since, the pre-
766
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
Fig. 8. Window surface temperature of the initial design.
shows the CFX 5.6 results with a reduced window thickness. The definition of δ is shown in Fig. 10. The window thickness is gradually reduced in the direction to the window center without changing the thickness of the cylindrical part of the beam tube. By a reduction in the window thickness of 1 mm in the center, the maximum window temperature is reduced by 86 ◦ C and kept below the design limit. For the window design with 2 mm thickness in the center, the maximum temperature difference across the window thickness is 57 ◦ C, as shown in Fig. 11, which is less than one half of the initial design. The structural analysis by Zucchini and Turroni (2004) confirmed that the maximum stress for this modified window design is in the allowable range. Therefore, all the remaining calculations in this paper are made for the modified window design with 2 mm thickness in the center.
Fig. 7. Velocity profile of the initial design.
dicted maximum window temperature exceeds the design limit (525 ◦ C) significantly, modification of the initial design is inevitable. 4.1.2. Results of modified design One of the proposals for the modification to the initial design is to reduce the window thickness. Fig. 9
Fig. 9. Effect of window thickness on maximum window temperature.
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
767
Fig. 12. Effects of funnel size on mass flow rate and average LBE speed in funnel.
Fig. 10. Definition of δ for window thickness variation.
The maximum temperature of the guide tube is not changed by the modification of the window thickness. It should be noted that the predicted maximum temperature of the guide tube, 450 ◦ C, exactly corresponds to the design limit for the guide tube. However, further analysis showed that the possible ways to reduce the guide tube temperature would be the reduction in the tube wall thickness or the replacement of the present guide tube material SS 316L by another material of a higher thermal conductivity.
Fig. 11. Window temperature for modified design (2 mm thickness in center).
4.1.3. Effects of funnel size A series of HERETA calculations showed that the size of the funnel is an important design parameter affecting the natural circulation flow around the target. As shown in Fig. 12, in case of a wider funnel, the LBE flow rate is increased due to a smaller flow resistance. But the injection speed to the window is decreased due to the increase in the flow area. Both effects contribute oppositely to the window cooling. Therefore, detailed analysis is necessary to achieve an optimum size of the funnel. Fig. 13 shows the results of CFX 5.6 calculations for three different funnel sizes. When the funnel size is reduced by 20 mm from the reference funnel size (=140 mm), the maximum window temperature is reduced by 7 ◦ C. However, the temperature of the guide tube increases dramatically, because the heat generation rate in the guide tube is larger than that of
Fig. 13. Effects of funnel size on maximum temperatures of window and guide tube.
768
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
4.2. Transient performance under beam trip conditions
Fig. 14. Temperature profile on inner surface of window at BOC and EOC conditions.
the original design. Furthermore, it is deduced from a sensitivity study that a reduction in the guide tube thickness cannot reduce the maximum guide tube temperature down to the design limit. When the funnel size is increased by 20 mm, the maximum window temperature is increased by 21 ◦ C, and exceeds the design limit. Based on the above results, the reference funnel size is, therefore, considered as the optimum value from the thermal–hydraulic point of view. 4.1.4. Effects of burnup The beam intensity, and subsequently, the spallation heat source, is varied with the fuel burnup. The neutronic calculation (Batta et al., 2003) shows that during operation of a 1-batch-cycle of XADS the beam current varies from 2.47 mA at keff ≈ 0.97 (typical at the beginning of cycle, BOC) to 5.1 mA at keff ≈ 0.94 (typical at the end of cycle, EOC). The total amount of heat deposition in the target is changed accordingly from 1.2 to 2.4 MW. The increase in the heat deposition leads to the change of the steady-state LBE flow rate from 149 kg/s at BOC to 183 kg/s at EOC. The LBE temperature at the inlet of the present computational domain is changed from 214 ◦ C at BOC to 228 ◦ C at EOC. Certainly, the window temperature also varies with the fuel burnup. Based on the boundary conditions provided by the HERETA code, CFX 5.6 calculations were performed for both BOC and EOC conditions. The results are shown in Fig. 14. It can be seen that the maximum window temperature is far below the design limit during the entire fuel cycle (∼3 years).
The operating experience of existing accelerators shows that beam trips occur very frequently (Bauer et al., 2000; Pierini, 2003). Beam trips produce thermal cycle and cause additional stress, which would become one of the main issues of structural material failure. Improved accelerator performance is required for XADS. In the technical specification of the accelerator for XADS, the frequency of beam trips with a time duration longer than 1 s has to be kept below five interrupts per year (Mueller, 2003; Pierini, 2003). No special requirements are defined for the beam trips with a beam interrupt duration shorter than 1 s. The present analysis focuses on beam trips with an interrupt duration shorter than 1 s. Beam trips with four different interrupt durations, i.e., 0.1, 0.3, 0.5, and 1 s, were investigated. The HERETA calculation showed that the mass flow rate of LBE and the LBE temperature at the inlet of the computational domain are hardly changed during these beam trips with a short interrupt duration. In case of the beam trip with a interrupt duration of 1 s, the change of the LBE mass flow rate is less than 2% and the change of the LBE temperature at the exit of the heat exchanger is less than 1 ◦ C. Therefore, in the CFX 5.6 calculations the inlet boundary conditions were kept unchanged. Fig. 15 presents the predicted temperature at the central point of the inner and outer surface of the window. For a better understanding, the case of beam shut off was also simulated and presented. For all cases shown in Fig. 15, the beam interrupt starts at the time point t = 0 s, and the beam power recovers after the corresponding trip periods, i.e., beam interrupt duration. As expected, beam trips with a smaller trip period result in a smaller temperature drop. The window undergoes a temperature drop up to 207 ◦ C for the trip period of 1 s, whereas the maximum temperature drop is only 32 ◦ C for the trip period of 0.1 s. Fig. 15 also shows that the steepest slope is located between 0 and 0.5 s after the beam interrupt, regardless of the investigated interrupt duration. Therefore, additional CFX 5.6 calculations were performed with a much finer time step for the first 0.5 s after the beam interrupt, in order to more accurately evaluate the maximum temperature change rate. As indicated in Fig. 16,
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
769
Regarding to the maximum temperature change rate, beam trips with a trip period shorter than 1 s could also be crucial event affecting the integrity of the window although no attention was paid to these trips in the technical specification of the XADS accelerator. Moreover, it is expected that beam trips with a trip period shorter than 0.1 s do not have enough time to reach the maximum temperature change rate of 412 ◦ C/s. It is thus concluded that related to both the maximum temperature drop and the temperature change rate, beam trips with a trip period shorter than 0.1 s are less critical than those with a trip period longer than 0.1 s, and are not further analyzed in this work. 4.3. Safety analysis
the predicted maximum temperature change rate is as high as 412 ◦ C/s, at the center of the outer surface of the window. This maximum temperature change rate occurs at 0.1 s after the beam interrupt.
Three kinds of abnormal accidents, i.e., beam focusing, loss of heat sink, and beam intensity jump, are analyzed in the present study. The main purpose of this analysis is to provide basic knowledge to the design of safety systems, to avoid window failure. Although the melting temperature of the window material (T91) is far above 1000 ◦ C, window failure might occur prior to its melting point due to both mechanical and thermal loads. The results of the tensile test of fresh T91 (Groeschel, 2004) and the stress analysis (Zucchini and Turroni, 2004) indicate that the window could fail at about 700 ◦ C under normal operation conditions. More data under irradiation with LBE environment are required for a more accurate prediction of window failure. In the present work, however, 700 ◦ C is used as a rough estimation for the temperature, at which window failure occurs.
Fig. 16. Temperature change rate under beam trip conditions.
4.3.1. Beam focusing In order to expand the proton beam onto the target, the reference accelerator for XADS uses the so-called raster scanning method deflecting a pencil-like beam with fast magnets in a designed pattern to paint the target area (Mueller, 2003). Therefore, a focused beam accident will occur by failure of rastering magnets. There are four rastering magnets for beam scanning. One pair of magnets covers each axis of the footprint (i.e., circle with a radius of 8 cm). Therefore, if one of the rastering magnets fails, the proton beam will cover only the half of the entire footprint with the same beam power. In this case, the beam intensity becomes double for one half of the footprint.
Fig. 15. (a) Temperature behavior under beam trip conditions (center of inner surface of the window). (b) Temperature behavior under beam trip conditions (center of outer surface of the window).
770
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
Fig. 17. Window temperature behavior under beam focusing conditions.
An accident initiated by the failure of one rastering magnet was simulated. To simplify CFX 5.6 analysis, the computational model and boundary conditions used for the reference steady-state analysis were unchanged except the beam current, which was increased to 12 mA from the reference value (=6 mA). These assumptions, in particular, axi-symmetrical twodimensional assumption, are not realistic and will give conservative results. However, this conservative analysis would provide a basic understanding of the target performance under beam focusing conditions. As shown in Fig. 17, the maximum window temperature arrives 700 ◦ C in about 0.8 s after the failure of one magnet. The window failure will occur even at an earlier time point since the thermal stress is increased with the double beam intensity. The result of the tensile test indicates that the window will not fail in the first 0.1 s, because the maximum window temperature is lower than 550 ◦ C. Therefore, it can be concluded that window failure will occur at a time point between 0.1 and 0.8 s after the failure of one rastering magnet. 4.3.2. Loss of heat sink Under normal operation conditions, the spallation heat is transferred to the secondary oil loop through the heat exchanger in the target unit. The oil loop itself is also cooled by another cooling loop. Loss of heat sink accident is initiated, in case one of the cooling loops is inactive. In the present work, the worst event, i.e., complete loss of heat sink, has been considered. Fig. 18 shows the mass flow rate and the LBE temperature at the exit of the heat exchanger under a complete loss of heat sink case.
Fig. 18. HERETA results under loss of heat sink accident.
The LBE flow rate is slowly reduced. It is still kept 78% of the initial value after 400 s. However, the LBE temperature at the exit of the heat exchanger is sharply increased after the loss of heat sink and reaches as high as 700 ◦ C in about 400 s. For more detailed analysis of window temperature using CFX 5.6 steady-state assumption was made due to the slow change of the boundary conditions. CFX analysis was carried out for three time points (=100, 200, 300 s) based on the boundary conditions supplied by HERETA. Fig. 19 shows the maximum window temperature at 100, 200, and 300 s after the loss of heat sink. It can be seen that the window failure occurs in about 200 s. 4.3.3. Beam intensity jump The maximum beam current of the XDAS accelerator is 6 mA, which gives a sufficient margin to the maximum needs of the XADS operation. As pointed
Fig. 19. Maximum window temperature vs. time under loss of heat sink accident.
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
out above, the required beam current for the XADS operation is 2.47 mA at BOC and 5.1 mA at EOC. Safety analysis should also consider the scenario that during the normal operating condition, beam current is suddenly increased from the normal operating value to its maximum value, i.e., 6 mA. This scenario is called beam intensity jump in this paper. Since, the target is cooled by natural circulation, beam intensity jump may cause temperature peak in the window and could damage the window. To a beam intensity jump scenario much attention has been paid in safety analysis of ADS (Cheng et al., 2004; Chen et al., 2004). However, the existing studies did not analyze the behavior of the window in detail, although the integrity of the window is very important for the calculation of the reactivity feedback. The present analysis concentrates on thermal behavior of the window after beam intensity jump. Results of beam intensity jump events depend on the time when beam intensity jumps occur. The most probable beam intensity jump occurs during the normal operation period with full reactor power. Fig. 20 shows the mass flow rate and the LBE temperature at the exit of the heat exchanger when the beam current is suddenly increased from 3 to 6 mA. The mass flow rate reaches a small peak after ∼25 s and stabilizes to the steady-state value. It never drops down below the initial flow rate. The LBE temperature at the exit of the heat exchanger is increased by ∼20 ◦ C. The temperature peak in the target occurs because the natural circulation flow is not established sufficiently in the early stage of the beam intensity jump transient. In order to minimize the computing expenditure, one steady-state calculation was performed using CFX 5.6
Fig. 20. HERETA results under a beam jump from 3 to 6 mA.
771
with the most conservative assumptions to investigate the window temperature under a beam intensity jump during normal operation conditions with full reactor power. For this calculation, the cooling conditions are kept to BOC conditions but the beam current is changed from 2.47 to 6 mA. With these assumptions, CFX 5.6 predicted the maximum window temperature as high as 550 ◦ C only. Therefore, window failure is not expected for beam intensity jump events during normal operating period with full reactor power.
5. Summary and conclusions The window target unit for the large LBE-cooled concept is one of the basic options in the proposed spallation target designs for XADS. In the present work, CFD studies were made for this option with the emphasis on the cooling capability of the window. Steadystate as well as transient behavior was analyzed using the CFX 5.6 code with the advanced turbulence model. The results achieved are summarized as follows: • The maximum steady-state temperature of the initially proposed 3 mm thick window exceeds the design limit by 71 ◦ C for the reference condition of 6 mA beam current. By a reduction in the thickness of 1 mm in the center, the maximum window temperature is reduced to 510 ◦ C and is kept below the design limit. • The reference funnel size (=140 mm) is proven as an optimum value in terms of the cooling capability of the window and the guide tube. • For the proposed reference target design, the maximum window temperature is far below the design limit during the entire fuel cycle (∼3 years). Under steady-state operating conditions, the performance of the proposed reference design was also confirmed by the stress analysis. • At beam trip transient, a temperature drop of about 200 ◦ C is obtained in case of a trip period of 1 s. For all beam trips analyzed, the maximum temperature change rate, as high as 412 ◦ C/s, occurs at about 0.1 s after the start of the beam interrupt. Therefore, beam trips with a trip period less than 1 s could be crucial events affecting the integrity of the window, and need to be considered in the design of the XADS accelerator.
772
N.-i. Tak et al. / Nuclear Engineering and Design 235 (2005) 761–772
• Among all three accidental transients analyzed, i.e., beam focusing, loss of heat sink, and beam intensity jump, the scenario of beam focusing shows the most serious safety concern. In case of beam focusing, window failure occurs in less than 1 s after the start of the rastering magnet failure. In case of loss of heat sink, about 200 s time is remaining for activating safety systems, to avoid window failure. No window failure is expected under beam intensity jump transients. Nowadays, due to the fast development of computer hardware as well as software, more and more applications of CFD codes to nuclear engineering and design are envisaged. Although a CFD code is believed to be the best tool for the design and analysis of spallation targets, it should be noted that the validation of CFD codes is still not completed in specific areas such as turbulent heat transfer in liquid metals. Acknowledgements This study was carried out in the framework of the PDS-XADS project partially financed by the 5th Framework Programme of the European Commission (Contract number FIKW-CT2001-00179/key action: Nuclear Fission). References ANSYS Incorporated, 2003. CFX 5.6 Manuals. Batta, A., et al., 2003. Window target unit for the XADS lead–bismuth cooled primary system. In: Proceedings of the International Workshop on P&T and ADS Development 2003, Mol, Belgium, October 6–8. Bauer, G., et al., 2000. Description of SINQ and boundary conditions for MEGAPIE. In: Proceedings of the 1st MEGAPIE General Meeting, CEA, Cadarache, June 14–15. Buono, S., et al., 1998. Numerical studies related to the design of the beam target of the energy amplifier prototype. HLMC’98, vol. 1. Carluec, B., 2003. The European project PDS-XADS, “Preliminary design studies of an experimental accelerator-driven system”.
In: Proceedings of the International Workshop on P&T and ADS Development 2003, Mol, Belgium, October 6–8. Chen, X.-N., et al., 2004. Comparative transient analyses of ADS with conventional fast reactor fuel and ADT with advanced fertile free fuel. In: Proceedings of the OECD/NEA Fourth International Workshop on Utilization and Reliability of High Power Proton Accelerators, Daejeon, Korea, May 16–19. Cheng, X., Cahalan, J.E., Finck, P.J., 2004. Safety analysis of an accelerator-driven test facility. Nucl. Eng. Des. 229, 289– 306. Cheng, X., Slessarev, I., 2000. Thermal–hydraulic investigations on liquid metal target systems. Nucl. Eng. Des. 202, 297–310. Coors, D., et al., 2004. Target units for XADS primary system. EURADWASTE’04—Sixth European Commission Conference on the Management and Disposal of Radioactive Waste, Luxembourg, March 29–31. DOE, 1999. A report to congress—a roadmap for developing accelerator transmutation of waste (ATW) technology. DOE/RW-0519. European Technical Working Group on ADS, 2001. A European roadmap for developing accelerator-driven systems (ADS) for nuclear waste incineration. Groeschel, F., 2004. Private communications. Paul Scherrer Institut. Mueller, A.C., 2003. The PDS-XDS reference accelerator. In: Proceedings of the International Workshop on P&T and ADS Development 2003, Mol, Belgium, October 6–8. Neitzel, H.-J., Knebel, J.U., 2002. Auslegung eines geschlossenen 4 MW-Targetmoduls mit W¨armeabfuhrsystem f¨ur eine ADSAnordnung. Scientific Report FZKA 6687, Forschungszentrum Karlsruhe. OECD/NEA, 2002. Accelerator-driven system (ADS) and fast reactors (FR) in advanced nuclear fuel cycles—a comparative study. OECD Nuclear Energy Agency. Pierini, P., 2003. Reliability study of the PDS-XADS accelerator. In: Proceedings of the International Workshop on P&T and ADS Development 2003, Mol, Belgium, October 6–8. Smith, B.L., Dury, T.V., Maciocco, L., Roubin, P., Tak, N.I., 2003. A benchmark study based on a representative design of the MEGAPIE spallation source target. In: Proceedings of the 10th International Topical Meeting on Nuclear Reactor Thermal–Hydraulics (NURETH-10), Seoul, Korea, October 5–9. Tak, N.I., Song, T.Y., Park, W.S., 2001. Numerical studies on thermal–hydraulics of HYPER target. AccApp/ADTTA’01, Reno, Nevada. Travleev, A., Broeders, C.H.M., 2003. Private communications. Forschungszentrum Karlsruhe. Vieser, W., Esch, T., Menter, F., 2002. Heat transfer predictions using advanced two equation turbulence models. CFX Tech. Memorandum, CFX-VAL10/0602. Zucchini, A., Turroni, P., 2004. Private communications. ENEA.