Inherent Safety Features and Passive Prevention Approaches for Pb/Bi-cooled Accelerator-Driven Systems Johan Carlsson
Stockholm 2003 Doctoral Dissertation Royal Institute of Technology Department of Nuclear and Reactor Physics
Akademisk avhandling sob med tillst˚ and AB Kung Tekniska H¨ogskolan framl¨agges till offentlig granskning f¨ or avl¨ aggande AB teknisk doktorsexamen onsdagen den 7 maj 2003 kl 10.00 i Kollegiesalen, Administrationsbyggnaden, Kungl Tekniska H¨ogskolan, Valhallav¨ agen 79, Stockholm. ISBN 91-7283-470-6 TRITA-FYS 2003:7 ISSN 0280-316X ISRN KTH/FYS/--03/7--SE c Johan Carlsson, Thursday, March 21, 2003
Universitetsservice US AB, Stockholm 2003
Abstract This thesis is devoted to the investigation of passive safety and inherent features of subcritical nuclear transmutation systems - accelerator-driven systems. The general objective of this research has been to improve the safety performance and avoid elevated coolant temperatures in worst-case scenarios like unprotected loss-of-flow accidents, loss-of-heat-sink accidents, and a combination of both these accident initiators. The specific topics covered are emergency decay heat removal by reactor vessel auxiliary cooling systems, beam shut-off by a melt-rupture disc, safety aspects from locating heat-exchangers in the riser of a pool-type reactor system, and reduction of pressure resistance in the primary circuit by employing bypass routes. The initial part of the research was focused on reactor vessel auxiliary cooling systems. It was shown that an 80 MWth Pb/Bi-cooled accelerator-driven system of 8 m height and with a 6 m diameter vessel can be well cooled in the case of loss-of-flow accidents in which the accelerator proton beam is not switched off. However, after a loss-of-heat-sink accident the proton beam has to be switched off within 40 minutes in order to avoid fast creep of the vessel. If a melt-rupture disc is included in the wall of the beam pipe, which breaks at 150 K above the normal core outlet temperature or 5 minutes after loss-of-heat-sink accident initiation, the grace period until the beam has to be shut off is increased to 6 hours. For the same vessel geometry, but an operating power of 250 MWth the structural materials can still avoid fast creep in case the proton beam is shut off immediately. If beam-shut off is delayed, additional cooling methods are needed to increase the heat removal. Investigations were made on the filling of the gap between the guard and the reactor vessel with liquid metal coolant and using water spray cooling on the guard vessel surface. The second part of the thesis presents examinations regarding an acceleratordriven system also cooled with Pb/Bi but with heat-exchangers located in the risers of the reactor vessel. For a pool type design with gas lift pumps, this approach has advantages regarding heat-exchanger tube failures, particularly if water is used as the secondary fluid. This is because a leakage of water from the secondary circuit into the Pb/Bi-cooled primary circuit leads to upward sweeping of steam bubbles, which would collect in the gas plenum. In the case of heat-exchangers in the downcomer steam bubbles may be dragged into the ADS core and add reactivity. Moreover, bypass routes are employed to increase the flow speed in loss-of-flow events. It iii
is shown that the 200 MWth accelerator-driven system with heat-exchangers in the riser copes reasonably well with both a loss-of-flow accident with the beam on and also an unprotected loss-of-heat-sink accident. For a total-loss-of-power (station blackout) and an immediate beam-stop the core outlet temperature peaks at 680 K. After a combined loss-of-flow and loss-of-heat-sink accident the beam should be shut off within 4 minutes to avoid exceeding the ASME level D of 977 K, and within 8 minutes to avoid fast creep. Assuming the same core inlet temperature, both the reactor design with heat-exchanger in the risers and the downcomers have similar temperature evolutions after a total-loss-of-power accident. A large accelerator-driven system of 800 MWth with a 17 m tall vessel may eventually become a standard size. For this higher power ADS, the heat-exchangers have to be in the downcomers in order to cope with loss-of-flow accidents. For a loss-of-flow accident with the beam on, the long-term vessel temperature peaks at 996 K, which does not pose a threat of creep rupture for the vessel. However, the location of the heat-exchangers in the downcomers will probably require secondary coolant other than water, like for example oil (for temperatures not higher than 673 K) or Pb/Bi coolant.
iv
Contents Abstract
iii
Nomenclature
xv
Acknowledgements
xxi
1 Introduction
1
2 Review of passive prevention measures and inherent safety pects 2.1 Definitions of passive and inherent safety . . . . . . . . . . . . . 2.2 Generations of nuclear power reactors . . . . . . . . . . . . . . . 2.3 General safety strategies for nuclear power plants . . . . . . . . . 2.3.1 Defense in depth . . . . . . . . . . . . . . . . . . . . . . . 2.3.2 Redundancy, diversity and physical separation . . . . . . 2.3.3 Quality assurance and maintenance . . . . . . . . . . . . . 2.4 Design objectives for Generation IV reactors . . . . . . . . . . . . 2.4.1 Sustainability . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.2 Safety and reliability . . . . . . . . . . . . . . . . . . . . . 2.5 Safety issues specific for ADS . . . . . . . . . . . . . . . . . . . . 2.6 Choice of coolant . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7 Summary and review of inherent and passive safety strategies . . 3 Validation of the STAR-CD for natural convection 3.1 The PASCO experiment . . . . . . . . . . . . . . . . 3.2 STAR-CD calculations using the Two-Layer model . 3.3 Heat transfer correlations . . . . . . . . . . . . . . . 3.4 Convergence on a finer mesh . . . . . . . . . . . . . 3.5 Hand calculation . . . . . . . . . . . . . . . . . . . . 3.6 Conclusions from the validation calculations . . . . . v
flows . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
as. . . . . . . . . . . .
5 5 5 7 7 10 11 11 11 14 16 16 22
. . . . . .
25 25 28 29 31 31 32
4 Investigations of a 200 and 800 MWth ADS-HXR 33 4.1 Design and physical properties of the 200 MWth ADS-HXR . . . . 33 4.1.1 Heat-Exchangers located in the riser of the vessel . . . . . . 35 4.1.2 Bypass Routes . . . . . . . . . . . . . . . . . . . . . . . . . 36 4.1.3 Emergency Decay Heat Removal . . . . . . . . . . . . . . . 37 4.1.4 Accelerator Beam-stop Devices . . . . . . . . . . . . . . . . 37 4.2 Investigations of design basis accidents for the 200 MWth ADS-HXR 38 4.2.1 Total-loss-of-power accident . . . . . . . . . . . . . . . . . . 39 4.2.2 Loss-of-flow accident . . . . . . . . . . . . . . . . . . . . . . 41 4.2.3 Loss-of-heat-sink accident . . . . . . . . . . . . . . . . . . . 42 4.3 Design and physical properties of the 800 MWth ADS-HXR . . . . 43 4.4 Investigations of design basis accidents for the 800 MWth ADS-HXR 43 4.4.1 Total-loss-of-power accident . . . . . . . . . . . . . . . . . . 43 4.4.2 Loss-of-flow accident . . . . . . . . . . . . . . . . . . . . . . 44 4.4.3 Loss-of-heat-sink accident . . . . . . . . . . . . . . . . . . . 45 4.5 Computational set-up and assumptions . . . . . . . . . . . . . . . . 46 4.6 Conclusions on the 200 MWth and 800 MWth ADS-HXR . . . . . 47 5 Abstracts 5.1 Paper 5.2 Paper 5.3 Paper 5.4 Paper 5.5 Paper 5.6 Paper
of 1 2 3 4 5 6
appended papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
49 49 50 50 51 51 52
6 Conclusions
53
Appendices
55
A The Ansaldo design
55
B Theory and calculation methods B.1 Fluid dynamics models . . . . . . . . . . . . . . B.1.1 Mass and momentum conservation . . . B.1.2 Buoyancy driven flows . . . . . . . . . . B.2 Boundary layer models . . . . . . . . . . . . . . B.2.1 The law of the wall representation . . . B.2.2 The Two-Layer model . . . . . . . . . . B.3 Heat transfer models . . . . . . . . . . . . . . . B.4 Turbulence models . . . . . . . . . . . . . . . . B.4.1 Standard k − model equations (linear) B.4.2 The Two-Layer model . . . . . . . . . . B.5 Droplet model . . . . . . . . . . . . . . . . . . . vi
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
59 59 59 59 60 60 60 61 61 61 63 63
B.5.1 Direct liquid-solid contact heat transfer B.5.2 Heat transfer by dispersed spray cooling B.6 Thermal radiation . . . . . . . . . . . . . . . . B.6.1 View factor calculation . . . . . . . . . B.6.2 Radiant fluxes . . . . . . . . . . . . . . B.6.3 Combined radiation and convection heat B.7 Computation methods . . . . . . . . . . . . . . B.7.1 Boundary and Initial Conditions . . . . B.7.2 Discretisation schemes . . . . . . . . . . B.7.3 Convergence and consistence . . . . . . B.7.4 Stability . . . . . . . . . . . . . . . . . . Papers
. . . . . . . . . . . . . . . . . . . . . . . . . transfer . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
63 64 65 65 66 66 67 67 67 70 70 79
vii
viii
List of Figures 1.1 2.1 2.2 2.3 2.4
2.5 3.1 3.2 3.3 3.4 3.5 3.6 3.7 4.1 4.2 4.3 4.4
Spent and processed fuel in the EU: cumulative and change per year from 145 reactors (127 GWe) currently in operation [8]. . . . . . . The evolution of nuclear power described in generations. Additions in brackets by the author. [17] . . . . . . . . . . . . . . . . . . . . The relation between physical barriers and levels of protection in defense in depth [23]. Additions in brackets by the author. . . . . . The radiotoxic inventory of fission products and transuranics of an LWR without reprocessing, Sv/g. [30] . . . . . . . . . . . . . . . . The dynamic process of lead reducing iron oxide film and iron reforming oxide constitutes the self-healing protective oxide film formation. [45] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Schematic view of PRISM’s RVACS [55]. . . . . . . . . . . . . . . . The PASsive COntainment cooling (PASCO) test channel [58]. . . Measured temperature distribution at the PASCO channel outlet. X is the width and Y is the depth of the channel [58]. . . . . . . . . . Measured velocity distribution at the PASCO channel outlet. X is the width and Y is the depth of the channel [58]. . . . . . . . . . . Temperature distribution at the PASCO channel at outlet calculated with STAR-CD using the Two-Layer model. . . . . . . . . . . . . . Velocity distribution at the PASCO channel at outlet calculated with STAR-CD using the Two-Layer model. . . . . . . . . . . . . . . . . Temperature distribution at the PASCO channel at outlet calculated with STAR-CD using the ANL heat transfer correlation. . . . . . . Velocity distribution at the PASCO channel at outlet calculated with STAR-CD using the ANL heat transfer correlation. . . . . . . . . . Drawing of the ADS-HXR from the side and from the top. . . . . . Melt-rupture disc included in the accelerator beam pipe [6]. . . . . Flow field of the ADS-HXR one hour after accident initiation. . . . Temperature and velocity evolution at core outlet and in the reactor vessel wall after a TLOP accident for an ADS-HXR. . . . . . . . . ix
2 6 9 13
19 21 26 27 27 29 29 30 30 34 38 39 40
4.5
Temperature and velocity evolution at core outlet after a TLOP accident with a 4 minutes delayed beam-stop for ADS-HXR. . . . . 4.6 Temperature and velocity evolution at core outlet and in the reactor vessel wall after a LOF accident for ADS-HXR. . . . . . . . . . . . 4.7 Temperature and velocity evolution at core outlet and in the reactor vessel wall after LOHS accident for ADS-HXR. . . . . . . . . . . . 4.8 Comparison of the temperature at core outlet after a TLOP accident for an 800 MWth ADS-HXR and 800 MWth ADS-HXD. . . . . . . 4.9 Comparison of the temperature at core outlet after a LOF accident for an 800 MWth ADS-HXR and 800 MWth ADS-HXD. . . . . . . 4.10 Comparison of temperature at core outlet after a LOHS accident for 800 MWth ADS-HXR and ADS-HXD. . . . . . . . . . . . . . . . .
41 42 43 44 45 46
A.1 Schematic view of the Ansaldo design. The numbers in the figure represent: 1. Core 2. Reactor Vessel 3. Rotating Plug 4. Above Core Structure (ACS) 5. Target Unit 6. Sub-Assembly (SA) Transfer Machine 7. Intermediate HX 8. SA Handling Channel 9. SA Basket 10. Cover Gas Cooler. . . . . . . . . . . . . . . . . . . . . . . . . . A.2 Schematic view of Ansaldo’s RVACS [13]. . . . . . . . . . . . . . .
55 57
B.1 The heat transfer efficiency as a function of water spray mass flux [78]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.2 The heat transfer efficiency as a function of the Weber number [78]. B.3 Radiative heat transfer at patch i. [65]. . . . . . . . . . . . . . . .
64 65 66
x
List of Tables I II III IV V
Levels of defense in depth Event categories in the French-German safety approach Hand calculation to verify the STAR-CD predictions. Critical temperature limitations for structural materials and protective oxide layers Steam generator tube rupture events
xi
Publications included in the thesis and author’s contribution to those 1. Carlsson, J., Wider, H., ”Emergency Decay Heat Removal from an AcceleratorDriven System”, Nuclear Technology, 140, No. 1, pp. 28-40, (Oct. 2002) Main part of this research 2. Carlsson, J., Wider, H., ”Comparison of Safety Performance of Pb/Bi-cooled Accelerator-Driven Systems for different heat-exchanger locations and power levels”, submitted to Nuclear Technology Main part of this research 3. Wider, H.U., and Karlsson*, J., ”Passive Safety Approaches in Lead/BismuthCooled Accelerator-Driven Systems, Annual Meeting on Nuclear Technology 2000, Bonn, Germany Thermal hydraulics calculations 4. Wider, H.U., and Karlsson*, J., ”Safety Aspects of Heavy Metal-Cooled Accelerator-Driven Waste Burners, Journal de Physique IV, 9, pp. 127-135, (1999) Thermal hydraulics calculations 5. Karlsson*, J., ”Decay Heat Removal by Natural Convection and Thermal Radiation from the Reactor Vessel”, Proc. of Accelerator Driven Transmutation Technologies and Applications ’99, Prague, Czech Republic, (June 1999) Main part of this research 6. Wallenius, J., Tuˇ cek, K., Carlsson, J., Gudowski, W., ”Application of burnable absorbers in an accelerator-driven system”, Nuclear Science and Eng., 137, No. 1, pp. 96-106, (Jan. 2002) Calculation of temperature and velocity profiles in pin bundles of the SingSing Core using STAR-CD * Until the autumn of 2002 the author’s surname was spelled as Karlsson in some official documents and as Carlsson in others. Since the surname was spelled Karlsson in the passport the author chose to use that spelling while he was abroad.
xii
Other publications of relevance published during Ph.D. studies • Carlsson, J. ”Decay Heat Removal from the Guard Vessel by Thermal Radiation and Natural Convection”, Licentiate Dissertation, Royal Institute of Technology, Stockholm, Sweden, TRITA-Fys 3072, (Dec. 2000) • Wider, H.U., Carlsson, J., Jones, A.V., ”Choices for Deployment of AcceleratorDrive Waste Burners”,Nuclear Applications in the New Millennium, AccAPPADTTA’01, Reno, Nevada, USA, (2001) • Wider, H.U., Carlsson, J., ”An Efficient Beam Interruption Device - Crucial to Avoid Structural Material Problems During an ADS Accident”, Nuclear Applications in the New Millennium, AccAPP-ADTTA’01, Reno, Nevada, USA, (2001) • Wider, H.U., Carlsson, J., Horvath, G., Mueller, K., Jones, A.V., ”Beam shut-Off in ADS Accidents - an Essential Requirement, Int. Conf. on BackEnd of the Fuel Cycle: From Research to Solutions”, GLOBAL2001, Paris, France, (2001) • Wider, H.U., Karlsson, J., Jones, A.V., ”Lead/Bismuth-Cooled, Thorium Based ADS and Critical Systems Meet Sceptic’s Criteria”, The 10th International Conference on Emerging Nuclear Energy Systems ICENES2000, Petten, Netherlands, Sept. 24-28, 2000
xiii
xiv
Nomenclature List of abbreviations ABWR ACS ADS ADS-HXD ADS-HXR AISI ALARA ALMR ALWR ANL APS APWR ASCLIM ASME ATW BOL BWR CFD DOE EOL EPR FDM FP FR FVM FZK HTR HX
- Advanced Boiling Water Reactor - Above Core Structure - Accelerator-Driven System - Accelerator-Driven System with Heat-eXchanger in the Downcomer - Accelerator-Driven System with Heat-eXchanger in the Riser - American Iron and Steel Institute - As Low As Reasonably Achievable - Advanced Liquid Metal Reactor - Advanced Light Water Reactor - Argonne National Laboratory - Automatic Protective Systems - Advanced Pressurized Water Reactor - ASsessment of Computational fluid dynamics Codes for heavy LIquid Metal coolants - American Society of Mechanical Engineers - Accelerator Transmutation of Waste - Beginning Of Life - Boiling Water Reactor - Computational Fluid Dynamics - (US) Department of Energy - End Of Life - European Pressurized Reactor - Finite Difference Method - Fission Products - Fast Reactor - Finite Volume Method - ForschungsZentrum Karlsruhe - High Temperature Reactor - Heat-eXchanger xv
IAEA IHX IRIS LLFP LOCA LOD LOF LOHS LWR MA MARS NPP NRC PASCO PBMR PCC PMR PRISM PSA PWR R&D RNG RRC RVACS SSC TLOP TMI TRU TVD
-
International Atomic Energy Agency Intermediate Heat-eXchanger International Reactor Innovative & Secure Long-Lived Fission Products Loss-Of-Coolant Accident Line Of Defense Loss-Of-Flow Loss-Of-Heat-Sink Light Water Reactor Minor Actinides Monotone Advection and Reconstruction Scheme Nuclear Power Plant (US) Nuclear Regulatory Commission PASsive COntainment Cooling facility Pebble-Bed Modular Reactor Plant Condition Category Prismatic fuel Modular Reactor Power Reactor Inherently Safe Module Probabilistic Safety Analysis Pressurized Water Reactor Research and Development ReNormalisation Group Risk Reduction Category Reactor Vessel Auxiliary Cooling System Sing-Sing Core Total-Loss-Of-Power Three Mile Island Transuranium Total Variation Diminishing
Elements, isotopes and chemical symbols Am - Americium Bi - Bismuth CO - Carbon Monoxide CO2 - Carbon Dioxide Cs - Cesium Cu - Curium F e3 O4 - Iron oxide He - Helium I - Iodine Na - Sodium N Ox - Nitrogen oxides Np - Neptunium Pb - Lead xvi
P bO P b/Bi Po Pu SO2 Tc U U O2
-
Lead Oxide Lead Bismuth eutectic Polonium Plutonium Sulfur dioxide Technetium Uranium Uranium Dioxide
Greek αr αs βT δij r φ φw κ µ µt ν ρ ρ0 ρr σ σ σφ σφ,t τij τw
-
surface absorption thermal diffusivity of solid thermal expansion coefficient the Kronecker delta turbulent energy dissipation heat transfer efficiency (droplet), see Eq ?? surface emissivity temperature in fluid wall temperature empirical coefficient molecular viscosity [N s/m2 ] turbulent viscosity [N s/m2 ] kinematic viscosity fluid density [kg/m3 ] fluid density at reference temperature T0 [kg/m3 ] surface reflectivity droplet surface tension [N/m] Stefan-Boltzmann’s constant molecular Prandtl number turbulent Prandtl number stress tensor components wall shear stress
Other As (t) Aµ Cµ cp c0p Dm E fµ Fh,j gc h
-
instantaneous droplet spreading area [mm2 ] constant empirical coefficient, usually taken as constant specific heat [J/kgK] reference specific heat at temperature T0 molecular diffusivity for constituent m empirical coefficient coefficient diffusional thermal energy flux in direction xj dimensional constant in W e heat transfer coefficient [W/m2 K] xvii
hf g hm,t ht h0t Ht Ii Ji k k ks md mm N Nu P p Pr qc qr qs,i qw Q Re RY → − S sh si sij sm SΦ t T Tw Tf Ts,0 Ti ∆Tsub Tsat Tl xi ui u0j u+ u
-
latent heat of vaporization [J/kg] thermal enthalpy for constituent m static enthalpy, cv T − c0v T0 static enthalpy fluctuations heat transfer coefficient [W/m2 K] total incident radiation flux to patch i total emitted radiation flux from patch i conductivity [W/mK] turbulent kinetic energy conductivity solid [W/mK] mass of droplet [g] mass of constituent m number of droplets in a square plane Nusselt’s number, N u = hL/k sublayer resistance factor, see Eq B.2 the pressure [P a] Prandtl number, P r = v/α convective heat flux [W/m2 ] radiative heat flux [W/m2 ] heat flux from wall to droplet [W/m2 ] total wall heat flux [W/m2 ] energy transferred to spray [J] Reynolds number, V Lρ/µ Reactor year surface vector energy source momentum source component the rate of strain tensor mass source component source coefficient time [s] temperature [K] wall temperature [K] fluid temperature [K] initial solid temperature [K] temperature at the solid-liquid interface [K] subcooling temperature, Tsat − Tl [K] temperature at saturation [K] liquid temperature [K] Cartesian co-ordinate absolute fluid velocity [m/s] absolute velocity fluctuation (u − uw )/uτ tangential fluid velocity xviii
uw uτ − → u r y+ + ym y We
- wall velocity - (τw /ρ)1/2 → → -− u −− uc , relative velocity between the fluid and the local coordinate velocity - ρCm u1/4 k 1/2 y/µ - point where change from linear to logarithmic modelling take place - normal distance from the wall to the center of the cell - Weber number, W e = DV 2 ρ/(gc σ)
xix
xx
Acknowledgements Grateful thoughts and many thanks to: Hartmut Wider for untiring and encouraging support, for his enthusiasm to science, optimism, and his great ability to answer complex questions. Waclaw Gudowski for accepting me as a Ph.D. student, for flexible guidance, and straight answers and advices. The Reactor Physics Group at KTH for friendship, stimulating discussions, and great, ”strong” Friday coffee times. Friends at JRC Ispra and JRC Petten for great company, nice food, and for inspiring working environment. My parents for their love and support. Ludmila, last but not the least. Special thanks to: The Swedish Nuclear Fuel and Waste Management Company, Svensk K¨arnbr¨anslehantering AB, and the European Commission for funding part of my studies within the frame of the Marie Curie fellowship program.
xxi
xxii
Chapter 1
Introduction The main mission of accelerator-driven systems (ADS) is the reduction radiotoxicity of the nuclear waste. Together with appropriate reprocessing, the ADS can transmute the long-lived waste from conventional nuclear power plants (NPP) into shorter-lived and less hazardous isotopes. The remaining radiotoxic inventory that needs geological disposal can thus be reduced to about one hundredth [1] and the effective radioactive dose from the geological waste repositories can be reduced to one thousandth compared to the case of no transmutation [2]. After reprocessing, the plutonium (Pu) and the minor actinides (MA) will be incinerated in ADSs. After transmutation a large fraction of the Pu inventory will be fissioned, and thus the risk for nuclear proliferation will be diminished significantly. Figure 1.1 presents a possible nuclear power park including ADSs. It should be noted is that a considerable fraction of Pu could be efficiently burnt in water reactors using thorium or inert matrix fuels [3, 4, 5]. Since the core of an ADS is designed to be subcritical, an external neutron source is required to sustain its operation. This neutron source is a beam of high energy protons, which impacts on a lead/bismuth (Pb/Bi) target and releases neutrons through spallation reactions. The neutrons are thereafter scattered and diffuse into the core where they sustain the chain reaction through fissions. Due to the subcriticality and Pb/Bi cooling, uncontrolled large power excursions cannot develop [6, 7]. Nevertheless, for larger inadvertent reactivity insertions the accelerator beam should be shut off shortly after accident initiation since negative reactivity feedbacks have a small effect on the power level due to the subcriticality of the core. Reactivity insertions considerably smaller than the subcriticality margin affect the power level very little [6, 7]. Efficient transmutation of MAs and higher Pu isotopes requires a fast neutron spectrum [8]. This is the reason why mainly gas and liquid metals coolants are considered for an ADS. A basic comparison, from the point of view of different safety aspects, is performed between the two most probable ADS coolants today, 1
Figure 1.1. Spent and processed fuel in the EU: cumulative and change per year from 145 reactors (127 GWe) currently in operation [8].
i.e. Pb/Bi eutectic and He gas. However, the research of this dissertation has been focused on a Pb/Bi-cooled reactor. Safe operation and sustainable energy generation are crucial development goals for ADSs. Important safety trends for the next generation of reactors are the enhancement of their inherent and passive safety. Enhanced inherent safety means that certain accident initiators have a lower probability of occurrence, e.g. due to a low pressure coolant. Passive systems are available even if reactor operators and/or the active safety systems are not responding, since they rely only on physical laws, e.g. natural convection. An example of a passive safety system investigated in this thesis is the reactor vessel auxiliary cooling systems (RVACS) [9, 10, 11, 12], which is based on heat removal by thermal radiation and natural air circulation around the vessel outside. The combination of passive and active systems leads to a high reliability of the whole plant, and thus, leads to enhanced safety. An attempt to improve the safety of a Pb/Bi-cooled ADS with water in the secondary cooling circuit, has been made by positioning the heat-exchangers (HX) in the risers of a simplified pool-type design. In other approaches the HXs are placed in the downcomer of the reactor vessel [13, 14]. By this relocation of HXs it can be avoided that steam bubbles are dragged into the core after a HX tube rupture. This work starts with a literature survey regarding passive systems and inherent safety features, see Section 2. In Section 3, the fluid dynamics code STAR-CD is validated regarding natural air convection cooling which is compared to experimental results of the Passive Containment Cooling facility (PASCO), see Section 3.1. In Section 4 both a 200 and an 800 MWth ADS concept is presented. This 2
section includes studies of RVACS, bypass routes, a beam-stop device, and the HXs positioned in the riser. Appendix A presents the Ansaldo ADS demo design, and App. B the theory and calculational methods.
3
4
Chapter 2
Review of passive prevention measures and inherent safety aspects 2.1
Definitions of passive and inherent safety
According to the International Atomic Energy Agency (IAEA) [15] the definitions of inherent and passive safety are: • Inherent safety characteristic: Safety achieved by the elimination of a specified hazard by means of the choice of material and design concept. (Comment by author: a specific accidental situation becomes impossible by nature, e.g. a low pressure coolant) • Passive component: A component which does not need any external input to operate. (Comment by author: safety mechanisms are activated by laws of nature and prevents/mitigates the accident development, e.g. RVACS) • Passive system: Either a system which is composed entirely of passive components and structures or a system which uses active components in a very limited way to initiate subsequent passive operation.
2.2
Generations of nuclear power reactors
New nuclear reactors are often times categorized as evolutionary or innovative/ revolutionary. The evolutionary designs have only limited conceptual changes and their improvements are based on previous experiences [16], whereas the innovative 5
designs have entirely different arrangements of the core and its cooling systems. Often the latter concepts aim at the elimination of a core melt scenario [16]. The US Department of Energy (DOE) categorizes the different nuclear reactor designs into generations. Each new generation was or is motivated by evolutionary improvements or innovations in reactor technology. Figure 2.1 illustrates the evolution of nuclear power. A brief description about which reactors belong to a certain generation of nuclear power plants is summarized below.
Figure 2.1. The evolution of nuclear power described in generations. Additions in brackets by the author. [17]
• Generation I: These designs were prototype commercial reactors of the 1950’s and 1960’s, e.g. Shippingport, Dresden, Magnox. Most of the Generation I plants are shut down today [18]. • Generation II: The second generation of reactors were deployed in the 1970’s and 1980’s and are currently in commercial use. They are the vast majority of the operating reactors today [18]. These include light-water reactors (LWR), i.e. boiling water reactors (BWR) and pressurized water reactors (PWR), as well as the CANDU heavy-water reactors and the advanced gas cooled reactor (AGR). • Generation III: The Generation III are also referred to as advanced design nuclear power plants, which include the advanced boiling water reactor (ABWR), the European pressurized reactor (EPR), the System 80+ advanced pressurized water reactor (APWR), and the AP600 passive-design reactor. Advanced boiling water reactors have been built and are currently in operation in Japan [18]. 6
• Generation III+: The Generation III+ reactors have been under development during the 1990’s and are planned to be deployed by 2010. Examples include the gas-cooled pebble-bed modular reactor (PBMR) [19] and the SWR-1000 which is a BWR incorporating a nearly complete set of passive safety features [20]. The SWR-1000 is currently being certified by the U.S. Nuclear Regulatory Commission (NRC). • Generation IV: These reactors are planned to be deployed by 2030 and are expected to be highly economical, to incorporate enhanced safety, and to produce less waste. An example is the Pb/Bi-cooled SVBR-75/100, which is a modular reactor of 100 MW power per module [21].
2.3
General safety strategies for nuclear power plants
Efforts to minimize the levels of radioactive release from nuclear power plants are expressed by the ”as low as reasonably achievable” (ALARA) principle. This means that every reasonable effort should be made to keep exposures as far as possible below the regulatory dose limits. The cost for improvements relative to benefits for public health and safety should also be taken into account [22]. The final objective of the ALARA principle is protection of the plant personnel and the public against radioactive substances coming from nuclear power plants during normal operation and most importantly during or after accidents. The three fundamental safety functions to achieve this goal are [16]: • control the nuclear fission process • cooling of the fuel • confinement of radioactive material The safety strategies to ensure the above mentioned safety functions have evolved and have been improved since the birth of nuclear reactors in the 1950’s. In Section 2.3.1-2.3.3 some of the most important safety strategies are presented, i.e. the defense in depth, redundancy, diversity, physical separation, quality assurance and maintenance.
2.3.1
Defense in depth
The central safety strategy is the defense in depth principle [16, 23]. If properly applied this principle ensures that for example equipment or human errors are fended off by the next level of defense. Figure 2.2 shows the defense in depth principle and Tab 2.3.1a presents the five levels of defenses.
7
Table 2.3.1a. Defense in depth levels [16]. Additions in brackets from author. Levels of Objective Essential means defense in depth Level 1
Preventions of abnormal operation and failures
Level 2
Control of abnormal operation and detection of failures
Level 3
Control of accidents within the design basis
Level 4a
Control of severe plant conditions, including prevention of accident progression Mitigation of the consequences of severe accidents
Level 4b
Level 5
Mitigation of radiological consequences of significant releases of radioactive materials
Conservative (inherently safe) design and high quality in construction and operation Control, limitation and protection systems and other surveillance features Engineered safety features (including passive ones) and accident procedures Complementary measures and accident management (inherent and passive features help) Complementary measures and accident management (inherent and passive features help) Off-site emergency response
The plant conditions, which correspond to a defense in depth level, are exemplified below by the standards of the German and French safety authorities. Table 2.3.1b shows which plant condition category (PCC) that corresponds to each protection level of the defense in depth strategy. The PCCs represent normal operation up to limiting accidents, whereas the risk reduction categories (RRC) prevent core damages and mitigate consequences from accidents [16], as for example passive ones.
8
Figure 2.2. The relation between physical barriers and levels of protection in defense in depth [23]. Additions in brackets by the author.
9
Table 2.3.1b. Event categories in the French-German safety approach.[16] Event categories Systems and measures to Defense cope with the events in depth Plant conditions categories PCC1
Normal operation
PCC2
Anticipated operational occurrences Incidents Limiting accidents
PCC3 PCC4
Inherently stable plant behavior operational systems Operational systems, limitation systems Safety systems Safety systems
1 2 3 3
Risk reduction categories RRC-A
Prevention of core melt
RRC-B
Prevention of large release after core melt (mitigation)
Diverse safety systems to prevent core melt Systems to mitigate consequences of core melt
4a 4b
The physical barriers comprise fuel matrix, fuel cladding, primary coolant, reactor vessel(s), and reactor containment. The strength of each barrier depends on both their design characteristics, e.g. quality and type of material, and how efficiently the barriers are protected by other safety systems [16].
2.3.2
Redundancy, diversity and physical separation
In case an operating system fails and its backup system malfunctions, the redundancy principle ensures that an additional backup system prevents accidents. Each running system has at least two backup systems and both are capable of averting the accident. For example, if the insertion of control rods fails in an LWR, borated water can instead be injected into the system. Diversity means that the equipments used to carry out a certain function are based on different design principles and are manufactured of different brands. Thus, for instance a common failure of all pumps could be avoided. Another broader example is the automatic protective systems (APS) of an LWR which consists of two separate trains of components, each using different operating principles and each capable of releasing the control rod clutches [24]. In order to avoid that a fire or an explosion of one component damages neighboring equipment they are physically separated. Practically this means that the different backup systems are located in separate compartments of the nuclear power plants. 10
2.3.3
Quality assurance and maintenance
The quality assurance within the nuclear industry is very advanced relative to the industry as whole. Each component undergoes careful inspection to avoid e.g. the use of materials or parts that are prone to fail. Only about 0.5% of all components of a nuclear power plant cause 90% of the core damages. Thus quality assurance and preventive maintenance on those components enhances safety significantly. This makes the normal operation more reliable and reduces the maintenance costs. Due to higher plant reliability the number of transients are reduced and safety challenges diminished. To determine which components are the most crucial for high safety performance the Probabilistic Safety Assessment (PSA) is often used. PSAs can also be used to find out which equipments are permitted to be out of service simultaneously for preventive maintenance. [25]
2.4
Design objectives for Generation IV reactors
The process of development of the Generation IV reactors will be open and transparent in order to gain public confidence. Its development goals are discussed and decided in a worldwide scientific forum led by the US DOE. The Generation IV reactor concepts should contribute to sustainable energy generation, very reliable in operation, extremely safe, and economically competitive with other energy sources [17]. An ADS design should also correspond to these goals. Sustainability is briefly explained in Section 2.4.1 since ADS can reduce the waste volume from conventional nuclear power and its radioactivity. The statements written in italics are quoted from the Generation IV development goals [17]. The issues of reliability and safety are covered in Section 2.4.2. Since economics is out of the scope of this thesis only the following two statements will be considered: 1. after separation (or partitioning) of the nuclear waste, transmutation of the waste from conventional nuclear power plants loaded with UOX, its transmutation in ADSs would increase the price of electricity production from nuclear power by ∼50±15% [26]. 2. burning much of the plutonium and possibly the neptunium in LWRs using thorium or inert matrix fuels could be more economical [3, 4, 5].
2.4.1
Sustainability
1: Generation IV nuclear energy systems including fuel cycles will provide sustainable energy generation that meets clean air objectives and promotes long-term availability of systems and effective fuel utilization for worldwide energy production.[17] In general most advocates of sustainable energy use do not consider nuclear power as the most attractive alternative. However, its potential role in sustainable development [27] has lately been stressed more strongly due to the growing concern 11
for the greenhouse effect. Nuclear power does not emit CO2 and consequently does not contribute to the greenhouse problem but would rather help to solve it. The nuclear power proponents also emphasize that it is a concentrated source of energy, which does not emit pollutants like CO, SO2 or NOx in contrast to fossil fuelled plants. Employing LWR technology at the present level of electricity generation, the known uranium resources will last about 60 years. With the estimated sources the use could be extended to about 220 years [28], however with drastically higher prices. A significant increase in nuclear energy production using only a oncethrough cycle is therefore questionable. It is also interesting to note that the coal resources will only last for 3-400 years at the current level of energy generation with coal [28]. Sustainability also means a much better utilization of uranium (or thorium) resources. This includes usage of reactors with better neutron economy and breeding possibilities, i.e. fast reactors which breed more new fuel (Pu or 233 U) than they burn. Therefore, to extend the life span of nuclear power, the future emphasis has to be on a massive deployment of breeder reactors. A thermal reactor generates considerably less Pu or 233 U than the 235 U or 233 U that it burns. Compared to LWRs which use only about 1% of the total energy that the fuel contains, a breeder reactor can potentially use 75% [28], i.e. the presently known resources would last some 2400 years. However, further research and development on breeder reactors is required since this technology is more difficult and demanding compared to today’s commercial LWR reactors. Thorium based fuel is another option which has the advantage to be 3 times more abundant in the earth’s crust than uranium [29]. 2: Generation IV nuclear energy systems will minimize and manage their nuclear waste and notably reduce the long term stewardship burden in the future, thereby improving protection for the public health and the environment.[17] The geological disposal of nuclear waste is a major public acceptance issue [5]. The research on Partition and Transmutation (P&T) could lead to a more acceptable nuclear waste with lower radiotoxicity, less heat generation, as well as reduced volume. For future reactors closed fuel cycles are needed in order to reduce the nuclear waste volume, so that not many more large repositories such as Yucca Mountain are necessary [5]. It is also important to generate less waste with current LWRs. This is possible by using thorium matrix fuel with enriched 235 U. Then MAs would not be generated but instead 233 U and the long-lived protactinium isotope 231 Pa. The 233 U is a good fuel whereas the 231 Pa would have to be transmuted. Reactors with a fast neutron spectrum can be used to efficiently transmute the MAs of the nuclear waste. This is achievable both in critical and subcritical reactors. However, critical reactors need fertile fuel in order to breed new fissile isotopes for maintaining the fission process. Thus the transmutation ratio for fast critical reactors is lower than for subcritical reactors using fertile-free fuel. 12
The use of fertile-free fuel in ADSs and transmutation of TRUs can reduce the effective radiotoxicity of the nuclear waste from conventional nuclear power plants by a factor of 1000 [2]. The contribution to the radiotoxic inventory is displayed relative to natural U ore in terms of fission products (FP) and TRUs, see Fig 2.3. After ∼400 years the FPs radiotoxic inventory is less than the contribution from natural uranium ore [30]. Hence, if the TRUs are transmuted, the period of time till the radiotoxicity of waste drops below that of natural uranium ore can be shortened from about 300 000 years to about 500-1000 years. If the cesium and strontium FPs are stored separately e.g. in salt formations, the remaining waste will become less toxic faster and its heat generation as well as its volume will be strongly reduced [31]
Figure 2.3. The radiotoxic inventory of fission products and transuranics of an LWR without reprocessing, Sv/g. [30]
Next generation power plants must also be designed with eventual decommissioning in mind in order to facilitate rapid disassembly, ease of decontamination, and ease of disposal [32]. 3: Generation IV nuclear energy systems and their fuel cycles will ensure that they are a very unattractive and the least desirable route for diversion or theft of weapons-usable materials.[17] The critical issue with regard to proliferation for ADSs is the reprocessing of spent fuel since the technology could be misused and altered to obtain weapons 13
utilizable material. Pyrochemical processing1 of spent fuel may be designed to be more proliferation resistant than aqueous processes2 . This is because the latter produces a pure Pu fraction, which is formed during processing, whereas the MAs still contains FPs in the pyrochemical process. On the other hand aqueous processes are more developed and would probably be operated in a large-scale plant that enhances the proliferation resistance [33]. After the reprocessing of the waste, the burning of a sizeable fraction of the Pu is possible in LWRs, and even more in FRs and ADSs. For transmutation of higher Pu isotopes or MAs only FRs and ADSs can be used.
2.4.2
Safety and reliability
An important safety objective for the next generation of reactors is the minimization of the radioactive exposure of workers, the public, and the environment. This includes a further reduction of core damage frequencies and strengthening of protection barriers of the defense in depth strategy. 1: Generation IV nuclear energy systems operations will excel in safety and reliability. [17] To avoid radioactive releases the fourth generation reactors have often fundamentally different designs in order to mitigate, to prevent, or even eliminate accident scenarios. Apart from an enhanced level of inherent and passive safety, a less complex design, economics, and reliability of the plant operation. During the 1970’s the complexity of the nuclear plants increased as more components and safety systems were employed. New components were e.g. added to increase redundancy and improve the functionality. But occasionally the complexity actually decreased the total safety level of the plant since the overview and understanding of the plants were reduced. Moreover, one component less is one component less which is susceptible to failure. Fewer components generally also require less maintenance and control [34]. A forgiving plant operation requires large margins before critical temperatures are reached, strong barriers, and long grace periods before operators need to respond. The vulnerability to operator failure is also reduced by inherent and passive safety features since they always react in accordance with the laws of nature [34]. Thus, for example a properly designed plant will shut itself down in case of elevated temperatures. After the Three Mile Island (TMI) and the Chernobyl accidents it was recognized that the human-machine interface was one of the weakest links of the nuclear 1 Several techniques can be used, e.g volatilization, liquid-liquid extraction, electro-refining in molten salt, fractional crystallization. Methods are often based on the use of low melting point salts such as chlorides or fluorides. 2 A solvent extraction process uses a nitric acid. Chemical separation of U and Pu is then performed by solvent extraction steps.
14
safety chain [35]. Too many alarms, indicators, and control devices made it difficult for the operators to have a clear overview. Thereafter large efforts have been put into simplification, computerization, and automation of the control rooms. 2: Generation IV nuclear energy systems will have a very low likelihood and degree of reactor core damage [17]. The modern safety strategies have gone beyond the measures mentioned in Section 2.3 and natural safety is achieved by exploiting the laws of physics [32]. This safety approach partially replaces traditional active safety systems by implementation of passive systems or employment of inherent characteristics in the upstream line of defense (LOD) [36]. Safe response is thus secured even if active sensing and switching equipments fail, as even if human errors would occur at the same time [37]. A commonly used example of inherent safety is a negative reactivity feedback for all foreseeable accident events, i.e. a core design that ensures reduced power generation after disturbances such as a temperature increase. Usage of low-pressure coolants like for instance liquid metals minimizes concerns about LOCAs. And a good natural circulation capability makes the plant operation safer since sufficient heat removal is assured even as forced convection is lost [36]. A core of lower power (∼600 MW) and metal coolant allows emergency decay heat removal by e.g. RVACS even though all active safety systems are lost [38]. Moreover, smaller size reactors often have economic advantages due to factory and mass production since construction time is shortened and assembly time reduced [32]. Furthermore, factory manufacturing should also improve the reliability of components and structures. However, it should be noted that passive systems loose in flexibility compared to active systems, e.g. a core cooled by natural convection cannot change flow rate quickly. 3: Generation IV nuclear energy systems will eliminate the need for offsite emergency response.[17] A complete elimination of emergency response is probably not realizable, e.g. a design that is forgiving against all types of terrorist attacks might not be achievable. Nevertheless, all measures to reach this goal should be pursued. Elimination of accident scenarios requires similar measures as mentioned above, e.g. increased level of passive safety and incorporation of inherent safety features. To strengthen barriers, in particular the containment, and the protection systems is another means to prevent radioactive release. 15
2.5
Safety issues specific for ADS
The accident scenarios for ADSs and critical reactors are often comparable, but differences exist due to the sub-critical core and the proton beam accelerator. A visible difference is the proton beam tube, which penetrates the reactor vessel as well as the containment [36]. The penetration of containment and reactor vessel has a negative impact on the defense in depth strategy since the tube provides a passage for radioactive materials through the barriers. Similar problems are encountered for LWRs too, since pipes of the intermediate heat exchangers (IHX) penetrate the vessel barrier. In BWRs, steam lines penetrate both the containment and the reactor vessel [36]. The high energy proton beam adds to the heat generation. For Ansaldo’s 80 MWth ADS design a 3 MW accelerator is used [13], of which only ∼2.25 MW has to be cooled away due to the fact that the spallation reactions are endothermic. Whereas for the 800 MWth SSC the beam power is 40 MW of which 30 MW has to be removed [39]. Both the Ansaldo and the SSC designs have separate cooling circuits for the targets. In case the target cooling malfunctions, the beam window melts soon afterwards and the vacuum pipe will be filled with Pb/Bi coolant. Thereafter the beam pipe is damaged and has to be replaced. Thus, elevated temperatures should be sensed and the proton beam interrupted quickly in order to avoid breakage of the beam window. In case the active beam shut off does not function, passive devices will interrupt the proton beam, e.g. a melt-rupture disc [6]. To avoid elevated coolant temperatures the accelerator proton beam should be interrupted after all types of severe accident conditions, e.g. LOF, LOHS and TLOP. The grace period before beam shut-off is necessary is shortest for fast beam overpower or reactivity accidents but still considerably longer than for fast reactivity accidents in fast reactors. For LOF and LOHS accidents the grace periods depend strongly on the design. A system with a high heat capacity and good natural convection is preferable in this regard. The external neutron source and the subcritical core usually result in a large core peaking factor. The power peaking increases as kef f is reduced with burn-up. Consequently, the core heat removal during normal operation has to be dimensioned for the hottest channel [36] at EOL. As mentioned in the previous section some ADSs, e.g. the SSC, are dedicated to transmute waste as efficiently as possible, and thus these cores are not loaded with fertile fuels. The consequences this has on safety for an ADS is described in the Section ”Neutronics”.
2.6
Choice of coolant
The coolant type determines many safety characteristics of a nuclear reactor. It affects for instance the operating temperature and pressure, the heat capacity, and 16
the natural convection. Furthermore, it determines the coolant freezing and boiling temperature, possible corrosion and erosion on structural materials, the neutronics, and the emergency decay heat removal capability. Since a fast neutron spectrum is necessary for an efficient transmutation of MAs [36], the candidate coolants for ADSs are limited to liquid metals or gas. Presently, Pb/Bi eutectic and He are the most probable coolants, of which a basic comparison follows from the point of view of different safety aspects. Temperature and pressure level A critical fast reactor core is very compact and of high power density. For an ADS with Pb/Bi cooling, the core can be somewhat less compact due to its low neutron leakage and low neutron absorption in heavy metal coolants. Pb/Bi coolant offers the possibility to efficiently remove heat at low coolant pressures; hence a thick pressure vessel is unnecessary. The pressure exerted on the reactor vessel pertains mainly to the coolants’ own weight. Since the liquid phase of Pb/Bi eutectic-cooled reactor range from 398 K to 1943 K the risk of freezing or boiling is limited. Instead the temperature restrictions depend on other factors, see Section on ”corrosion and erosion”. To remove sufficient amounts of heat, a gas-cooled reactor must operate at high pressures (∼40-140 bars), which means that thick pressure vessels are needed. A cooling problem could occur in case of leakage of gas from a high-pressure system [36] since the decay heat removal is difficult to accomplish at low pressures. Active systems will be necessary to resolve a depressurization accident. As gas-cooled reactors are high-temperature systems the core outlet temperature is higher than for Pb/Bi-cooled reactors. For instance the prismatic fuel modular reactor (PMR) [17] has a core outlet temperature of 1123 K compared to ∼823 K for the Pb/Bi-cooled SSC. The thermal efficiency of He-cooled reactors is thus significantly better than for Pb/Bi-cooled reactors, i.e. 45% and ∼35%, respectively [40]. Moreover, freezing and boiling issues are eliminated since the coolant is always in gas phase. In-service inspection of gas-cooled reactors is easier because of the transparency of the coolant. However, the activated materials in and around the core do not allow human access. Heat capacity and natural convection The heat capacity of the total reactor system, comprising all components and the coolant, determines the grace period until operator action is needed with regard to high temperatures after LOHS accidents. After LOF accidents the natural convection capability is, however, more important since higher coolant velocities decrease core outlet temperatures. The natural convection capability depends primarily on the thermal expansion coefficient of the coolant, the pressure resistance of the flow path, as well as the vertical distance between the thermal centers of the core and the HXs. 17
For a combined LOF and LOHS accident both the heat capacity and the natural convection capability determines which temperature maximums are reached during accident progression. The heat capacity of a Pb/Bi-cooled reactor is about ten times larger compared to a fast gas-cooled reactor of the same geometry (at 70 bars and 1000 K for the Hecooled reactor)3 If the accelerator proton beam is not shut off the total heat capacity is particularly important. Thus the period of grace available for interruption of the beam in cooling accidents is significantly lower for He-cooled than Pb/Bi-cooled systems. Sufficient heat removal during a LOF and LOHS accident for fast gascooled reactors by natural convection cooling can only be achieved for low core powers, i.e. some tens of MWs [41]. For higher core powers active safety systems are needed to remove decay heat during accidents. After depressurization of a He-cooled reactor the emergency decay heat removal requires active systems. Chemical reactions and activation Neither Pb/Bi nor He-cooled reactors have violent chemical reactions with water or air. However, for Pb/Bi-cooled reactors with water coolers, a potential problem is water ingress into the primary circuit which could result in pressurization and a large steam production. For He-cooled systems water ingress from an intercooler is imaginable during a depressurization but possibly also in full pressure conditions because of a back-pressure developing in a ruptured intercooler. A disadvantage of Pb/Bi cooling is the generation of radioactive polonium (Po), which has a half life of 138.4 days. Via neutron capture in 209 Bi the α-emitter 210 Po is formed [36]. Almost 100% of the Bi in the Pb/Bi eutectic is the 209 Bi isotope. The reaction is showed in Eq 2.1 [42]. 209
Bi(n, γ)210 Bi
β−
−−→
210
5.0d
Po
α
−−−−→ 138.8d
206
P bst
(2.1)
At operating temperatures Po is volatile and some will migrate to the cover gas where aerosols are formed. A leakage could pose contamination problems. Also with pure Pb coolant there would be Po production because 208 Pb forms 209 Bi after neutron capture. However, the rate of 209 Bi production is 1000 times lower for pure Pb coolant. The equilibrium reactivity in Pb/Bi coolant has been found to be ∼370 GBq/kg. These results come from Russian 80-reactor years experience with Pb/Bi-cooled submarines [43]. The Po problem is possible to manage during schedule maintenance [44]. Helium is not activated by neutron irradiation. However, during normal operation structural materials will be activated strongly due to the low shielding in the case of a He gas coolant. 3 A thermal He-cooled high temperature reactor (HTR) has enough heat capacity because of the large amounts of graphite present. Therefore cooling or depressurization accidents lead only to a core heat up for gas-cooled thermal reactors.
18
Corrosion and erosion Pb/Bi coolant is corrosive towards steels. Carbon steels are the most corrosion resistant, whereas low Cr steels of (<2.25% Cr) are moderately resistant and high Cr steels are subject to severe corrosion [45]. Oxygen level control of the coolant can reduce the corrosion by orders of magnitude since it creates a protective oxide layer on structural materials [46]. At too low oxygen concentration no effective oxide films of Fe3 O4 can be created, whereas at too high oxygen concentrations PbO precipitates in the molten Pb/Bi and degrades the thermal hydraulic performance significantly [45]. Therefore good mixing is important in order to have a welldistributed oxygen concentration in the coolant. In case oxygen control does not function for a long time (months) blockages can develop. The base of the protective film is Fe3 O4 [43, 47]. Equation 2.2 displays the chemical reaction in the oxide film [45], and Fig 2.4 displays how the transfer process stops as the Fe concentration reaches equilibrium near the oxide surface. F e3 O4 + 4P b 3F e + 4P bO
(2.2)
Figure 2.4. The dynamic process of lead reducing iron oxide film and iron reforming oxide constitutes the self-healing protective oxide film formation. [45]
The oxide layer is stable up to 893 K for corrosion-resistant 12Cr-Si ferritic steel and below coolant velocities of 3 m/s [48, 49]. Higher velocities and temperatures should be avoided during normal operation to avoid rupture of the oxide layer. Since helium is chemically inert it will not corrode structural materials. No particular measures have to be taken with regard to corrosion and erosion for gas coolants [36]. Neutronics The reactivity feedbacks in ADSs have small influence on the power level of the core. As long as a reactivity insertion is smaller than the sub-criticality level plus the fraction of delayed neutrons, the core power level is principally determined 19
from the power inserted by the high energy proton beam [6, 7]. Due to the large subcriticality margin a reactivity insertion that would make the core supercritical is highly unlikely. In case such a situation would occur the proton beam should be interrupted immediately and safety rods inserted. Some ADS designs use fertile free fuel concepts in order to maximize their transmutation efficiencies. The Swedish Sing-Sing Core (SSC), which is developed at the Royal Institute of Technology (KTH, Swedish abbr.) in Stockholm represents this type of reactor. The SSC additionally uses burnable absorbers in order to make up for the burnup of the MAs and to achieve a harder neutron spectrum and thereby improve the transmutation efficiency further [50]. The absence of fertile fuel reduces the Doppler feedback significantly. Also a large fraction of MAs in the fissile fuel decreases the delayed neutron fraction. However, the reduced Doppler and the smaller neutron fraction are only relevant if the subcritical system becomes critical. A larger Pb/Bi-cooled reactor can have a positive void feedback. However, core voiding is highly unlikely due the high boiling point of Pb/Bi. Whether a massive plenum fission gas release is possible and whether it can lead to core voiding remains to be investigated. In gas-cooled reactors a positive void is not possible; nevertheless, water ingress could lead to positive reactivity. In both gas-cooled and Pb/Bi-cooled systems molten cladding motion could cause positive reactivity effects. In the gas system molten fuel slumping could also cause positive effects, whereas in Pb/Bi coolant the molten fuel would disperse according to experience from a Russian submarine [51, 52]. Both Pb/Bi and He-cooled reactors have hard neutron spectra with a flux weighted average neutron energy at hundreds of keV. The gas-cooled reactor has harder spectrum because of its very low moderating capability. However, the MA transmutation effectiveness for gas and Pb/Bi-cooled cores are still comparable [36]. Ultimate emergency decay heat removal The ultimate measure to remove decay heat from a Pb/Bi-cooled ADS is by natural air circulation around the guard vessel, i.e. reactor vessel auxiliary cooling system (RVACS). Since an RVACS is a passive safety system that always functions a parasitic heat loss appears during normal operation. Figure 2.5 displays the Power Reactor Inherently Safe Module (PRISM) [53]. The RVACS of PRISM has the capacity to remove up to 2.5 MW during accident conditions [54]. Another type of RVACS has been proposed by Ansaldo [13], in which the cooling air circuit is completely separated from the guard vessel. The RVACS air coolant then flows in separate pipes outside the guard vessel. The separate cooling circuit serves as another physical barrier against transport of radioactive particles to the atmosphere in case of rupture of both vessels. Moreover, it could still cool the core if both the vessel and guard vessel ruptured. The heat removal is somewhat lower compared to a conventional RVACS. For more information see App A. 20
Figure 2.5. Schematic view of PRISM’s RVACS [55].
21
RVACSs are less favorable for gas-cooled ADS reactors, since the gas conductivity is too low to remove heat efficiently via the vessel wall [56]. The high conductivity of Pb/Bi makes temperature gradients much smaller than for He-cooled systems. For the gas-cooled system emergency HXs at the top of the vessel are proposed. These remove the emergency decay heat passively as long as the system is pressurized. If a core melt accident occurs, the Pb/Bi would still cool a degraded core well. This can be concluded from such an accident in a Russian Pb/Bi-cooled submarine, in which the melted fuel was dispersed in the coolant [51, 52]. For a core melt of a gas-cooled fast system, water cannot be used in the core region because it would lead to supercriticalities. Possibly, a Pb/Bi emergency coolant could be injected instead. For core melt scenarios a gas-cooled ADS needs a core catcher below the vessel. This could only be cooled by water below the core catcher. In the case of inert matrix type fuels, concerns about recriticalities should be investigated. In the case of Pb/Bi cooling the vessel should act as the core catcher. Whether inert matrix fuels gets dispersed enough through natural circulation to avoid recriticalities should also be investigated.
2.7
Summary and review of inherent and passive safety strategies
The fourth generation of nuclear reactors are designed to implement more inherent safety features than their predecessors. Generation IV reactors also use passive safety features when applicable as a backup system to an active safety system or to an operator error (e.g. not realizing the initiation of an accident). Moreover, large design margins and simplicity of design are desirable from the point of view of safety, reliability, and economics. The defense in depth strategy remains a fundamental strategy also for ADSs, as well as diversity and redundancy. Inherent and passive safety features are important components for all mentioned strategies. For linear accelerators, the proton beam tube weakens the defense in depth strategy since it penetrates two of the outer barriers. The subcritical core of an ADS changes the neutron dynamics since reactivity insertions will affect it insignificantly (unless they are very large). Instead the accelerator, which acts as an external neutron source, determines the power level of an ADS. One important safety parameter for the fourth generation of nuclear reactors as well as for accelerator driven transmutation systems is the type of coolant. For efficient transmutation of minor actinides a hard neutron spectrum is required which limits the choices of coolant to either gas or liquid metals. Among the liquid metals the most probable coolants today is Pb/Bi eutectic. It has good thermal properties, allows a hard neutron spectrum, and does not react with air or water. Its flaw is its corrosiveness towards steel; however, this can be resolved by creating a protective 22
oxide layer on structural materials by employing oxygen control in the coolant. Another drawback of Pb/Bi is the Po production in the coolant. A He-cooled reactor also has a hard neutron spectrum. Helium is not corrosive, does not create violent chemical reactions with air or water, and will not become activated. However, a high pressure level is required to efficiently remove heat. An active system is thus needed to resolve depressurization accidents for gas-cooled fast reactor cores. Due to the low moderation of neutrons, structures inside the vessel and the vessel itself will be activated. The heat capacity is lower than for Pb/Bi-cooled reactors, which make grace periods for beam shut-off shorter. The thermal efficiency of gas-cooled reactors is considerably higher.
23
24
Chapter 3
Validation of the STAR-CD for natural convection flows The commercial fluid dynamics code STAR-CD [57] was validated for large scale natural air convection and thermal radiation heat removal in a benchmark against experimental results from the PASsive COntainment cooling (PASCO) facility at ForschungsZentrum Karlsruhe (FZK), Germany. Section 3.1 describes the PASCO experiment and Sec 3.2-3.3 presents results from the STAR-CD calculations.
3.1
The PASCO experiment
The objective of the PASCO experiment was to validate heat removal for containment cooling at high Rayleigh numbers, large channel geometries, and with strong interaction between convection and radiation heat transfer [58]. Additionally, numerical simulations were carried out with the one-dimensional code PASCO and the three-dimensional code FLUTAN [59]. The experimental channel’s width was 0.5 m, the depth varied from 0.5 to 1.0 m, and the height was 8.0 m, see Fig 3.1. Experiments were executed at different wall temperatures and at two surface emissivities (0.4 and 0.9). Temperatures and velocities were measured at several locations along the channel.
25
Figure 3.1. The PASsive COntainment cooling (PASCO) test channel [58].
26
Figure 3.2 depicts a steep temperature reduction with greater distance from the heated wall. The minimum temperature appears in the central region, whereas temperatures close to side walls and the back wall are elevated due to thermal radiation heat transfer. Local maximums appear in the channel corners.
Figure 3.2. Measured temperature distribution at the PASCO channel outlet. X is the width and Y is the depth of the channel [58].
The velocity distribution, see Fig 3.3, has a similar profile, except that very close to the wall surface the velocity is reduced due to friction [58].
Figure 3.3. Measured velocity distribution at the PASCO channel outlet. X is the width and Y is the depth of the channel [58].
A reduction of the surface emissivity leads to lower temperatures near the back wall, e.g. a change of wall emissivity from 0.9 to 0.4 decreases the air mass flow and the heat flux by about 20 to 30%, respectively [58]. If instead the distance between 27
the heated and the back wall is reduced a larger fraction of the emitted radiation reach the back wall, i.e. the back wall has a larger view factor and thus temperatures and velocities close to the back wall increase. For instance, reduction of channel depth from 1.0 m to 0.5 m decreases the airflow rate by 40%, whereas the heat flow rate remains nearly unchanged. An increase of the heated wall temperature will augment the airflow rate and the heat flux to the air. For a temperature increase from 373 to 473 K, the heat removal rate increases more than 200%, whereas the air mass flow rate increases about 50% [58].
3.2
STAR-CD calculations using the Two-Layer model
Some validation efforts incorporated STAR-CD’s detailed Two-Layer model. This employs a low Reynolds number formulation, which should be used on fine grids within the boundary layers. In the free-stream a high Reynolds number model is employed on a coarser mesh [57]. In order to resolve the temperature profile accurately, at least 15 cells are needed next to the wall surface. Consequently, the mesh size increases and calculations become time consuming. The governing momentum, turbulence, temperature and density was discretised with a MARS scheme, see App B.7.2. In the PASCO facility, the air passes an inlet bend of 90 degrees angle relative the test channel. Vanes are used to guide the flow smoothly into the channel. The STAR-CD simulation takes this into account as an additional pressure loss added to the channel inlet. The pressure loss was calculated analytically according to Eq 3.1 [60]. Le V 2 ρ (3.1) D 2 For a heated wall temperature of 423 K the PASCO experiment showed an average flow velocity of 1.25 m/s. For a 90-deg elbow of medium radius, the bend radius to pipe diameter L/D is chosen to be 26, and the friction factor, f , for a smooth surface is 0.026 thus the pressure drop will be 2.01 Pa. With the pressure loss included, the STAR-CD calculation predicts a mass flow of 0.37 kg/s and a heat removal rate of 6700 W. This is to be compared with the PASCO experiment, which showed a heat removal rate of 6500 W and a mass flow of 0.34 kg/s. The calculated temperature profile is shown in Fig 3.4, and the velocity profile in Fig 3.5. ∆p = 4f
28
423 404.4 385.9 367.3 348.7 330.1 311.6 293
Figure 3.4. Temperature distribution at the PASCO channel at outlet calculated with STAR-CD using the Two-Layer model.
2.633 2.257 1.881 1.505 1.129 0.7524 0.3762 0.0
Figure 3.5. Velocity distribution at the PASCO channel at outlet calculated with STAR-CD using the Two-Layer model.
3.3
Heat transfer correlations
Examinations employing wall functions in the boundary layers were conducted too. Such models usually give less accurate estimates, but are attractive since they are less expensive regarding computational time. Several heat transfer correlations for natural convection were investigated. The correlation that showed the best agreement with experimental results is displayed below, see Eq 3.2 [55]. It was developed at Argonne National Laboratory (ANL) within the American liquid metal reactors’ (ALMR) program for RVACS calculations.
29
N u = 1.22 · Re0.456 P r0.4
(3.2)
where N u, Re, and P r are Nusselt, Reynolds and Prandtl numbers, respectively. This correlation predicted a mass flow rate of 0.355 kg/s and a heat removal rate of 6600 W. This should be compared with the standard heat transfer correlation of STAR-CD, which calculated a mass flow of 0.361 kg/s, and a heat removal rate of 6994 W. The temperature and velocity profile can be seen in Fig 3.6 and 3.7, respectively. As already mentioned in Section 3.2, the PASCO experiment showed a heat removal rate of 6500 W and a mass flow of 0.34 kg/s.
423 404.4 385.9 367.3 348.7 330.1 310.6 293
Figure 3.6. Temperature distribution at the PASCO channel at outlet calculated with STAR-CD using the ANL heat transfer correlation.
2.181 1.870 1.558 1.246 0.938 0.623 0.3116 0.0
Figure 3.7. Velocity distribution at the PASCO channel at outlet calculated with STAR-CD using the ANL heat transfer correlation.
30
3.4
Convergence on a finer mesh
The convergence was examined with a four times finer mesh on the 80 MWth Ansaldo ADS design [12]. The temperature was shown to fluctuate more on the coarse mesh just after the accident initiation. This was probably due to the fact that the flow is not finely modelled in the inner parts of the vessel on the coarse mesh. However, these fluctuations are local temperature variations and the average temperature of the coolant is stable. The temperature differences diverge somewhat for both the wall and core outlet, and 40 hours after accident initiation the differences are 5 K and 8 K, respectively. The coarse mesh calculation estimates a heat removal rate, which is ∼4% lower throughout the whole calculation. This can most likely be explained by the different heat transfer models used next to the reactor vessel wall on the Pb/Bi side, since for the fine mesh the wall function is no longer appropriate and the detailed Two-Layer model had to be used [57]. The velocity evolution is in good agreement for the calculations on fine and coarse mesh.
3.5
Hand calculation
To validate the heat removal by the RVACS from the ADS-HXR200 vessel, hand calculations were performed and compared with the STAR-CD calculations (see Chapter 4 for more information about the ADS-HXR200 design). PN The total thermal resistance, p=1 Rp , from the Pb/Bi coolant to the inflowing air of the P RVACS can be defined as the ratio between the overall temperature N difference p=1 ∆Tp to the heat flux q, see Eq 3.3 [65]. N X p=1
Rp =
PN
p=1
∆Tp
q
(3.3)
where ∆Tp = Tp − Tp+1 is the temperature difference between two facing walls or between two surfaces of the same wall. The resistance in the walls is calculated from Eq 3.4. l (3.4) k where l is the wall thickness and k is the wall conductivity. Between walls where the heat transfer is by thermal radiation and convection the thermal resistance can be calculated as Eq 3.5. R=
R∗ =
{N u kl
+
σ 2 2− (θp
1 2 )(θ + θ + θp+1 p p+1 )}
(3.5)
where N u = hL κ , surface emissivity, σ Stephan-Boltzmann constant, θ temperature in K [65]. 31
For natural convection flows between vertical plates the Nusselt number is calculated from Eq 3.6 [65]. N u = 0.59(GrP r)1/4 3
(3.6)
c µ
∞ )L where Gr = gβ(Ts −T and P r = pk . ν2 A comparison between the STAR-CD results and the hand calculations is presented in Tab 3.5. They differ less than 10% for both temperature levels.
Table 3.5. Hand calculation to verify the STAR-CD predictions. Temperature Hand calculation, kW STAR-CD calculation, kW coolant, K 573 291 316 873 1079 1197
3.6
Conclusions from the validation calculations
Two important conclusions can be drawn from the comparison between the STARCD calculations and the results from the PASCO experiments. First, the STAR-CD code can properly simulate the PASCO experiment with the detailed Two-Layer model with regard to mass and heat transfer in the hydrodynamic and heat transfer boundary layers. However, in long transient calculations with CFD codes such as STAR-CD, it is of key importance to use coarse meshes together with appropriate heat transfer correlations. The ANL correlation showed very good agreement with the PASCO experiment, which is not surprising since it was developed for natural air circulation at high temperatures, within the ALMR program. The heat flux was overestimated by 1.5% compared to the PASCO experiment and the mass flow by 4%. The convergence behavior for a finer mesh and the hand calculations are also satisfactory. However, the proper treatment of the Pb/Bi thermal-hydraulic behavior by the STAR-CD code could not be fully verified yet. The use of the detailed Two-Layer model on surfaces could only be used for smaller meshes. However, to use such fine meshes would be too time consuming for regular calculations. A comparison of the finer mesh using the Two-Layer approach and the coarser mesh using the Reynolds analogy showed a 4% difference of heat removal rate. Fortunately the coarse mesh calculations were conservative, i.e. showed a decreased emergency cooling of the ADS under investigation. Moreover, the treatment of the two-phase flow of the Ansaldo gas-lift approach was simplified by assigning an equivalent density effect to the Pb/Bi. In the meantime an assessment of the state of the art CFD tools is made on the basis of existing and planned experiments. This is done in the assessment of computational fluid dynamics codes for heavy liquid metal coolants (ASCHLIM) in a concerted action of the EU 5th Framework program [62].
32
Chapter 4
Investigations of a 200 and 800 MWth ADS-HXR In accordance with design objectives for the fourth generation of nuclear power plants (see Section 2.4) possible inherent safety features were investigated for an ADS. The study concerned the location the heat-exchangers in the riser instead of the downcomer of an ADS (ADS-HXR). This would be a particular safety improvement, for the case of water coolant in the secondary circuit, after a rupture of the HXs since steam bubbles would not be dragged into the core for this design. Section 4.1 describes the 200 MWth ADS-HXR design and physical properties. Section 4.2 presents the results from the investigations on the 200 MWth ADS-HXR, whereas Sec 4.4 describes the 800 MWth ADS-HXR.
4.1
Design and physical properties of the 200 MWth ADS-HXR
The Pb/Bi-cooled ADS-HXR has a pool-type vessel of 9 m height and 6 m diameter. Figure 4.1 illustrates the ADS-HXR schematically. The reactor vessel steel is AISI 316, i.e. the same type of steel as for PRISM-S and the Ansaldo design. Since Pb/Bi-coolant is a liquid metal, the heat removal is efficient also at low pressures. Nevertheless, a guard vessel encloses the reactor vessel to maintain a high coolant level in case the reactor vessel would break. The maximum tolerable temperatures for structural materials and oxide layers are shown in Tab 4.1.
33
Figure 4.1. Drawing of the ADS-HXR from the side and from the top.
34
Table 4.1. Critical temperature limitations for structural materials and protective oxide layers. Characteristic problem1 Temperature when problem occurs Corrosion of structural material ASME level C2 ASME level D3 The limiting temperature to avoid creep failure in several hours under the given pressure conditions4 Melting point AISI316
893K [49] 922K [63] 977K [63] 1173K [64]
1670K [65]
The normal operating power of ADS-HXR is 200 MWth . The core is located in the lower region of the vessel. Its geometry and pressure resistance is assumed to be the same as in the Ansaldo 80 MWth XADS demonstrator [13]. An accelerator beam of maximum 7.5 MW is used to maintain the power level at 200 MWth at EOL. This beam impacts into a Pb/Bi target in which spallation reactions release neutrons that diffuse out from the target to the core. The tall vessel and large vertical distance between the thermal centers between the core and the HXs increases the natural convection capability. This is due to greater static pressure difference between the risers and the downcomers, see Eq 4.1. ∆p = g · h · ∆ρ
(4.1)
Furthermore, the low pressure resistance of the core improves natural circulation. Bypass routes are used to lower the pressure resistance after LOF accidents, see Section 4.1.2. To increase the Pb/Bi flow rate during normal operation gas injection into the risers will probably be used, which is preferred because of its simplicity and robustness. However, an uncertainty remains as to how the heat removal from the HXs is affected as the argon lift gas passes this region. In case the gas injection impedes the heat removal severely, electromagnetic pumps could be employed instead. To increase safety margins a higher heat capacity relative to the core power is desirable. This would prolong the grace period before a beam-stop is required after LOHS accidents.
4.1.1
Heat-Exchangers located in the riser of the vessel
The HXs are located in the risers of the reactor vessel in order to decrease the risk for transport of steam bubbles to the core region in case of a HX tube rupture. Hence, the steam bubbles collect in the upper plenum, and thus, will not reach the core region. A burst disc would be installed that breaks at elevated pressures in order to open a passage for the steam/gas mixture to an emergency condenser [66]. 35
Flaps will be used to block backward flow below the HXs in case of major tube break. For a Pb/Bi reactor the probability for tube cracking is unknown. However, test series performed in IPPE in Obninsk, Russia, for 5000-10000 hours at 873 K and 1.8 m/s demonstrated no corrosion damages [49]. For a PWR the probability for water leakage from the secondary circuit is 1.1x10−2 /reactor year (RY) and the probability for a major steam leakage is estimated to 8.4x10−4 [67]. In 1982 the NRC published a list on unresolved safety issues on LWRs. One of the fifteen items described was steam generator (SG) tube integrity. At that time the main contributor to SG ruptures originated from chemical impurities of the coolant water, which caused corrosion. The water chemistry control has since been improved since and the corrosion problem is almost eliminated [68]. Thereafter degrading of SG tubes in LWRs was mainly by denting, which is caused from corrosion of the carbon steel support plates and the buildup of corrosion products in the crevices between tubes and the tube support plates. Countermeasures to control denting have been taken, but other phenomena still continue to cause tube cracking. The tube cracking events that have occurred in the US are presented in Tab 4.1.1. Table 4.1.1. Steam generator tube rupture events [68]. Date Plant Leakage rate 2/75 9/76 10/79 1/82 5/84 7/87 3/89 3/93 2/00
Point Beach 1 Surry 2 Prairie Island 1 Ginna Fort Calhoun North Anna 1 McGuire 1 Palo Verde 2 Indian Point 2
470 liters per minute 1250 lpm 1480 lpm 2880 lpm 420 lpm 2080-2460 lpm 1890-2270 lpm 910 lpm 570 lpm
A negative consequence from locating the HXs in the risers is that the distance between the thermal centers of the core and HXs is reduced in case of a LOF accident. Thus, the HXs can only be situated in the risers of low-power ADS, for which a lower natural circulation capability can be tolerated, see Sec 4.3.
4.1.2
Bypass Routes
Bypass routes (BPR) are employed to reduce the pressure losses and thereby increase the Pb/Bi coolant velocities after LOF accidents. Due to the lower pressure resistance of the BPRs, the coolant in this region will have higher velocities compared to the HX region. It has to be taken into consideration that the larger BPR, 36
although more efficient after LOF accidents, leads to higher core outlet temperatures during normal operation. For the ADS-HXR design the BPRs cover 5% of the azimuthal distance of the vessel. The four BPRs are built in parallel with the HXs, see Fig 3.7. Backward flow through the BPRs during normal operation is prevented by argon gas injection into this region too. But the injection rate is lower than in the HXs region, since during normal operation the coolant flow rate should be high in the HX-region and as low as possible in the BPR-region.
4.1.3
Emergency Decay Heat Removal
The ADS-HXR uses an RVACS as the ultimate emergency decay heat removal system. The RVACS heat removal is based on thermal expansion of heated air around the reactor vessel. The buoyancy forces coming from the heated air’s lower density compared to the surrounding air makes it rise. In the channel between the guard vessel and the collector wall, the air velocity is typically 2-3 m/s after accidents. At 650 K the heat removal rate is ∼1.2 MW for the ADS-HXR design. A parameter study for RVACS was performed as to the surface emissivity, surface roughness, fin pitch, the surface temperature, and the gap width between the guard vessel and collector wall [69], see Paper 5. The most important parameters were found to be the surface emissivity and the vessel wall temperature. The heat transfer rate due to the latter is proportional to Ta4 − Tb4 where the subscripts a and b refers to the interchanging surfaces. It was also examined how the filling of the gap between the guard and reactor vessel with Pb/Bi affects the heat removal, see Paper 1. Due to higher conductivity between the vessels the thermal resistance decreases and hence the guard vessel temperature increases. To boost the heat removal further, water droplets were sprayed on the guard vessel surface since evaporation of water consumes much more energy than heating of air. Finally, the combined effect of filling the gap between the vessels with Pb/Bi and using water spray cooling was examined. Once the gap is filled and the water spray cooling functions at full capacity, the heat removal will exceed the decay heat generation of a 200 MWth nuclear plant almost immediately after accident initiation. However, this strongly enhanced cooling of the vessel outside cannot remove the nominal power in a LOHS accident with beam on.
4.1.4
Accelerator Beam-stop Devices
Interruption of the accelerator proton beam has the same effect for a sub-critical reactor as scram of the control rods for a critical reactor. For the great majority of accident conditions of an ADS, the proton beam is simply interrupted by shutting off the accelerator current. However, situations can occur where the normal accelerator beam interruption malfunctions. For these conditions passive shut-off devices are 37
needed to secure interruption of the accelerator. For example, a melt-rupture disc [6] can be included in the wall of the vacuum beam pipe, see Fig 4.2 and Paper 3.
Figure 4.2. Melt-rupture disc included in the accelerator beam pipe [6].
This disc would melt as the coolant temperature exceeds a certain design limit. After melting, Pb/Bi would flood the accelerator beam pipe and consequently the impact point of the proton beam would be relocated from the core to the upper region of the reactor vessel. Thereafter, the core produces decay heat power, whereas the accelerator generates about 5.6 MW (7*0.75) in the upper region. A drawback of the melt-rupture disc design is that the vacuum pipe has to be replaced after filling it with Pb/Bi. Moreover, heat deposition from the accelerator beam is still large in the upper region of the vessel and will rather certainly lead to a severing of the beam pipe unless a special design for accommodating the beam power is developed. The melt-rupture disc may only be suitable for rather low power systems since the temperature at which the disc should melt have to be higher than the temperature peak after a LOF accident. Otherwise, the melt-rupture disc would melt too frequently which also leads to unnecessary replacements of the vacuum pipe. A passive beam-stop device that interrupts the accelerator beam without filling the vacuum pipe and interrupts the proton beam current would be desirable. Such a device could be triggered at a lower temperature than the melt-rupture disc. Hartmut Wider and the author are presently working on such a device.
4.2
Investigations of design basis accidents for the 200 MWth ADS-HXR
A design requirement for ADS-HXR is to avoid critical temperatures after LOF and TLOP accidents. Also, grace periods after LOHS accidents should be extended in 38
order to provide enough extra time for operators to counteract.
4.2.1
Total-loss-of-power accident
A total-loss-of-power (TLOP) accident can for example be initiated by a station blackout. Thus active heat removal systems do not work any longer, argon gas injection stops (or the electromagnetical pumps stop), and the accelerator proton beam is interrupted. Nevertheless, the core continues to generate decay heat of ∼6.2% of nominal power right after accident initiation and ∼1% of normal power after one hour [70]. After ∼1 hour the flow reverses and runs backward through the core. The flow reversal is due to a high heat transfer rate through the separating walls between the HXs and the BPRs on the downcomers’ side. The temperature difference after accident initiation is 175 K, which gives a heat transfer rate of ∼2.8 MW. After one hour the heat transfer through the walls exceeds the core decay heat generation and make the flow reverse. Figure 4.3 shows the flow field of ADS-HXR one hour after accident initiation.
Figure 4.3. Flow field of the ADS-HXR one hour after accident initiation.
The core outlet (which is the inlet during normal operation) has a temperature peak at 664 K after almost 10 hours, see Fig 4.4.
39
0.4
1200 Reactor vessel temp. Core outlet temp. Core inlet temp. Core outlet vel.
0.3 0.2
800
0.1 0.0
600
Velocity, m/s
Temperature, K
1000
-0.1 400
0
5
10
15
Time, hours
Figure 4.4. Temperature and velocity evolution at core outlet and in the reactor vessel wall after a TLOP accident for an ADS-HXR.
Delayed accelerator beam-stop A very unlikely accident scenario is a malfunction of all cooling systems but not the accelerator proton beam. This appears only conceivable in case the accelerator has a separate power supply. Within ∼4 minutes the accelerator proton beam should be interrupted in order to avoid exceeding the ASME level D. Then the core outlet temperature peaks at 922 K after 8 minutes, see Fig 4.5. About 2 hours after accident initiation the coolant changes flow direction through the core. The reversal is later than for an immediate beam-stop since the temperature differences between the coolant in the HXs and the BPRs have diminished during the initial 4 minutes. Thus, the heat transfer through the walls separating the downcomers from the HXs and the BPRs is decreased.
40
Reactor vessel temp. Core outlet temp. Core inlet temp. Core outlet vel.
1000
0.3 0.2 0.1
800 0.0 600
Velocity, m/s
Temperature, K
1200
-0.1 0
5
Time, hours
Figure 4.5. Temperature and velocity evolution at core outlet after a TLOP accident with a 4 minutes delayed beam-stop for ADS-HXR.
A delayed beam-stop of eight minutes is possible without exceeding the fast vessel creep limit of 1173 K.
4.2.2
Loss-of-flow accident
During a loss-of-flow (LOF) accident only the argon gas injection malfunctions (alternatively the electromagnetical pumps stop). Hence the enhanced circulation gradually diminishes. Thereafter the coolant circulates by natural convection, i.e. only driven by temperature differences in the coolant. For this accident scenario, the accelerator continues to operate and the core produces normal power whereas the flow rate diminishes. One minute after accident initiation the temperature at core outlet peaks at 1085 K. Thereafter the buoyancy forces are strong enough to accelerate the flow; after 3 minutes the temperature maximum has decreased to 1030 K. At this point the natural circulation becomes stable. The reactor vessel reaches its temperature peak after 11 minutes at 932 K. Figure 4.6 depicts the temperature evolution after a LOF accident.
41
0.4
1200
Temperature, K
0.2 800
Reactor vessel temp. Core outlet temp. Core inlet temp. Core outlet vel.
600
400
0.1
Velocity, m/s
0.3
1000
0.0
0
5
10
15
20
25
-0.1
Time, minutes
Figure 4.6. Temperature and velocity evolution at core outlet and in the reactor vessel wall after a LOF accident for ADS-HXR.
To avoid elevated temperatures and damages on structural materials during longer periods of time, the accelerator proton beam should be shutdown soon after accident initiation.
4.2.3
Loss-of-heat-sink accident
In case the normal heat removal system stops to function, whereas the core still operates at nominal power, the coolant temperature would increase by ∼0.6 K per second on average. Since the flow rate remains elevated also after accident initiation, the temperature differences in the coolant are smaller than for LOF accidents. Hence, the core outlet temperature increases more slowly for LOHS relative to LOF accidents. The temperature evolution at core inlet and outlet, as well as the maximum reactor vessel wall temperature are displayed in Fig 4.7. After LOHS accidents the accelerator beam should be shut down within a few minutes in order to avoid elevated temperatures in supporting structures and fuel elements.
42
0.4
2500
0.3 0.2 1500 0.1 Reactor vessel temp. Core outlet temp. Core inlet temp. Core outlet vel.
1000
500 -5
0
5
10 15 20 25 30 35 40 45
Velocity, m/s
Temperature, K
2000
0.0 -0.1
Time, minutes
Figure 4.7. Temperature and velocity evolution at core outlet and in the reactor vessel wall after LOHS accident for ADS-HXR.
4.3
Design and physical properties of the 800 MWth ADS-HXR
The Sing-Sing Core (SSC) that is developed at the Royal Institute of Technology in Stockholm, Sweden, was used for the investigations on the 800 MWth design. Since the SSC at this stage only comprises the core without surrounding structures, the Ansaldo core was used as reference design to estimate the total pressure drop of the SSC, see Paper 2. The active core of the ADS-HXR is 78 cm in diameter and 100 cm tall [50]. The height of the reactor vessel was increased to 17 m in order to improve natural circulation, whereas the vessel diameter remained at 6 m.
4.4 4.4.1
Investigations of design basis accidents for the 800 MWth ADS-HXR Total-loss-of-power accident
The TLOP accident with immediate beam-stop shows a favorable temperature evolution. Shortly after accident initiation the coolant flow direction through the core reverses. After 15 hours the core outlet temperature peaks at 758 K, see Fig 43
4.8. Hence a TLOP accident will not damage the core or structural materials. The coolant temperature of the ADS-HXD peaks at 847 K after 18 hours.
850
Temperature, K
800 750 700 650
ADS-HXR ADS-HXD
600 550 0
5
10
15
Time, hours
Figure 4.8. Comparison of the temperature at core outlet after a TLOP accident for an 800 MWth ADS-HXR and 800 MWth ADS-HXD.
4.4.2
Loss-of-flow accident
The LOF accident leads to elevated temperatures at the core outlet for the 800 MWth ADS-HXR. As the gas injection stops core outlet temperatures rapidly rise to ∼1700 K, which is far above acceptable temperatures limits. Figure 4.9 depicts a comparison between ADS-HXR and ADS-HXD with regard to the temperature evolution at core outlet. For ADS-HXD the temperature stabilizes at 999 K.
44
Temperature, K
2000
1500 ADS-HXR ADS-HXD
1000
500
0
5
Time, minutes
Figure 4.9. Comparison of the temperature at core outlet after a LOF accident for an 800 MWth ADS-HXR and 800 MWth ADS-HXD.
After an LOF accident the accelerator proton beam should be shut off promptly for the 800 MWth ADS-HXR.
4.4.3
Loss-of-heat-sink accident
The grace period during which the proton beam should be shut off for a LOHS accidents is 6 minutes for the 800 MWth ADS-HXR. Otherwise the ASME level D of 977 K is exceeded. Another 6 minutes thereafter the fast vessel creep limit of 1173 K will be exceeded. The temperature evolution for ADS-HXR and ADS-HXD are comparable for this accident scenario, see Fig 4.10.
45
Temperature, K
1400
1200
1000 ADS-HXR ADS-HXD
800
600
0
5
10
15
Time, minutes
Figure 4.10. Comparison of temperature at core outlet after a LOHS accident for 800 MWth ADS-HXR and ADS-HXD.
After LOHS accidents the proton beam should be shut off within a couple of minutes in order to avoid damages on the core and structural materials.
4.5
Computational set-up and assumptions
All walls are treated as hydrodynamically smooth, i.e. the roughness is such that boundary layers next to surfaces are assumed to have an undisturbed profile. The core is modelled as a porous medium. Experimental data for pressure drops of the Ansaldo core are used as reference values, i.e. at 0.42 m/s velocity the pressure drop over the core is 20 kPa [13]. Then the pressure losses are proportional to the coolant velocity squared. The power insertion in the core is divided into three regions, which reflects the power profile of Ansaldo’s 80 MW demo design [13]. The heat transfer correlation employed on the air side was developed for RVACS calculations within the American Liquid Metal Reactor program [55], see Eq 3.2. A correlation for liquid metals is used to calculate the heat transfer coefficient on the Pb/Bi coolant side, see Eq 4.2. The latter correlation is derived from the theory of the relationship between temperature and velocity profiles in liquid metals [71]. N u = 0.565Re0.5 P r0.5
(4.2)
where N u, Re and P r are the Nusselt, Reynolds, and Prandtl numbers, respectively. The emissivity is set to 0.7, which is a rather conservative value for steels with heavily oxidized surface. 46
A first order upwind differencing scheme is used for modeling momentum, energy, and density equations for ADS-HXR, see App B.7.2. This scheme was chosen because of its numerical stability. A centered differencing scheme is utilized for temperature calculations. The k − standard model is used for turbulence estimations, see App B.4.1. The decay heat generated in the core after beam-stop is based on experimental data from existing LWRs. This is because ADSs are loaded with the Pu and MAs from LWR fuel that has a similar fuel composition, and thus the decay heat generation will be comparable. Within about 10 seconds after accident initiation, the core power has dropped to ∼6.2% of normal operation power, i.e. 12.4 MW for the 200 MWth ADS-HXR. From then on, Eq 4.3 is used to model the core power. Pd (t, T ) = 0.0622 · P0 (t−0.2 − (T + t)−0.2 )
(4.3)
where P0 is normal operating power, t is time after the reactor was shut down, and T is the time during which the core has operated at normal power.
4.6
Conclusions on the 200 MWth and 800 MWth ADS-HXR
To locate heat-exchangers in the risers would be an advantage in case of a tube rupture in the heat-exchangers. Thus, it could be avoided that steam bubbles are dragged into the core since they would collect in the upper plenum where a burst disc is installed to guide the steam/gas mixture to an emergency condenser. However, the distance between the thermal centers of the HXs and the core is smaller after a LOF accident which impedes the natural circulation. The emergency decay heat removal for both systems is performed by an RVACS. The heat removal capacity of an RVACS depends primarily on parameters like surface emissivity and temperature. At an emissivity of 0.7 the heat removal for a guard vessel temperature of 650 K is ∼1.2 MW for the ADS-HXR design. After a TLOP accident the RVACS heat removal is better for ADS-HXR than for ADS-HXD, because after ∼1 hour after accident initiation the coolant flow reverses and flows at higher rate through the core. The flow reversal is due to the fact that the bypass routes are constructed in parallel to the heat-exchangers, which creates buoyancy forces opposite to the normal flow direction. The core outlet temperature peaks at 664 K after almost 10 hours for the 200 MWth ADS-HXR. The 200 MWth ADS-HXR performs satisfactory after both LOF and LOHS accidents. The LOF accident results in a temperature peak at core outlet of 1085 K, which decreases to 1030 K after about two minutes. The grace period in LOHS accidents for beam shut-off is 4 minutes in order to remain below the ASME level D (977 K). To avoid fast vessel creep the grace period would be 8 minutes (1173 K). 47
The 800 MWth ADS-HXR performs well after TLOP and LOHS accidents. However, the LOF accident leads to very high core outlet temperatures within one minute after accident initiation, i.e. ∼1700 K. Hence, this type of approach is not safe enough for an ADS. The natural convection of the 800 MWth ADS-HXD is sufficient to provide adequate cooling in all cooling accidents. This is rather certainly the only approach to cool a larger ADS with the beam on during these accidents. However, the use of water as a secondary coolant is probably not possible in this design. Besides the use of oil, which can only be used for temperatures lower than about 700 K, one could use Pb/Bi in the secondary circuit to avoid any problems with steam or oil ingress into the ADS core.
48
Chapter 5
Abstracts of appended papers 5.1
Paper 1
Emergency Decay Heat Removal from an Accelerator-Driven System Nuclear Technology, 140, No. 1, pp. 28-40, (2002) The passive emergency decay heat removal during severe cooling accidents in Pb/Bi-cooled 80 MWth and 250 MWth Accelerator-Driven System designs was investigated with the CFD code STAR-CD. For the 80MWt design, the calculations show that no structural problems occur as long as the accelerator proton beam is switched off immediately after accident initiation. A highly unlikely delay of beam stop by 30 minutes after a combined Loss-Of-Heat-Sink and Loss-Of-Flow accident would lead to increased reactor vessel temperatures, which will not cause creep failure. By using a melt-rupture disc on the vacuum pipe of the accelerator proton beam to interrupt the beam at elevated temperatures in a passive manner the grace period before beam stop is necessary is increased from 30 minutes to 6 hours. An emergency decay heat removal design which would prevent radioactive release to the atmosphere even more reliably than the PRISM design, was also investigated. For an ADS of 250 MWth power with the same vessel as the 80 MWth ADS examined, the maximum wall temperature reaches 745 K after an immediate beam stop. This does not cause any structural problems either. The grace period until a beam stop becomes necessary for the 250 MWth system was found to be about 12 minutes. To reduce elevated vessel temperatures more rapidly after a beam stop, alternative cooling methods were investigated, for example filling the gap between the reactor and the guard vessel with liquid metal and the simultaneous use of water spray cooling on the outside of the guard vessel. This decreases the coolant temperatures already within minutes after switching off the proton beam. The use of chimneys 49
on the Reactor Vessel Auxiliary Cooling System, which increase the air flow rate, lowers the maximum reactor vessel wall temperature only by about 20K. It can be concluded that the critical parameter for the emergency cooling of an ADS is the time delay in switching off the accelerator after accident initiation.
5.2
Paper 2
Comparison of Safety Performance of Pb/Bi-cooled Accelerator-Driven Systems for different heat-exchanger locations and power levels Submitted to Nuclear Technology A safety investigation on the location of heat-exchangers in the risers or the downcomers of a Pb/Bi-cooled accelerator-driven system of 200 and 800 MW(thermal) powers was performed. In a pool type design with a simple flow path the use of heat-exchangers in the risers will have advantages in case of heat-exchanger tube failures. This is particularly true if water is used as the secondary fluid, since it can be avoided that steam bubbles are dragged into the core region by the Pb/Bicoolant. The safety implications with regard to the temperature evolution during loss-of-flow, loss-of-heat-sink, and total-loss-of-power accidents were compared with designs where the heat-exchangers are located in the downcomers. During a loss-of-flow accident for the 200 MW(thermal) system with heat-exchangers in the risers, the core outlet temperature increases to about 1020 K. For the acceleratordriven system with heat-exchangers in the downcomers the temperature maximum is nearly 150 K lower. After a loss-of-heat-sink accident the grace period before the proton beam has to be shut off was found to be ∼5 minutes for both the 200 MW(thermal) designs. During a total-loss-of-power accident for the case with heatexchangers positioned in the riser, the core outlet temperature peaks after 10 hours at about 750 K, which is 100 K lower than in the case with heat-exchangers in the downcomer. The investigations on the 800 MW(thermal) system were performed in a taller vessel of 17 m height. The design with the heat-exchangers in the riser showed elevated temperatures of more than 2000 K during a loss-of-flow accident, which would severely damage the core. The case with heat-exchangers in the downcomers showed acceptable results for all accident types investigated. Therefore, an 800 MW(thermal) accelerator-driven system of pool design with simplified flow path must have the heat-exchangers in the downcomer. The computational fluid dynamics code STAR-CD was used in all calculations.
5.3
Paper 3
Passive Safety Approaches in Lead/Bismuth-Cooled Accelerator-Driven Systems 50
Annual Meeting on Nuclear Technology, Bonn, Germany Bonn, 2000 For the safety of Accelerator-Driven Systems (ADS) it is important that the proton beam is shut off soon after the initiation of an accident and that the decay heat can still be removed if the regular heat removal fails. This paper presents a new device for passively blocking the proton beam when a Pb/Bi-cooled core heats up during an accident. An investigation was made of the passive removal of the decay heat and also the heat generated by a proton beam that impinges on the Pb/Bi coolant surface (due to the blocking of the beam pipe with Pb/Bi) during a Loss-of-Heat-Sink accident. A Reactor Vessel Auxiliary Cooling System (RVACS) similar to the one used in PRISM is considered. The results show that a small ADS with a blocked-off beam can be safely cooled for a few days. For larger ADSs vessel creep may start after several hours if the beam is not switched off completely. RVACS with novel features can cool larger ADSs with blocked beam for many days.
5.4
Paper 4
Safety Aspects of Heavy Metal-Cooled Accelerator-Driven Waste Burners Journal de Physique IV, 9, pp. 127-135, 1999 Accelerator-driven, subcritical lead / bismuth cooled systems have several safety advantages. The critical accident initiators in such systems lead only to a relatively slow coolant heat-up that should be noticed by the reactor operators who will initiate a shutting down of the accelerator. This decreases the reactor power to decay heat levels. If the coolant temperature increase should go unnoticed, passive systems will lead to an automatic shutdown of the accelerator or a blocking of the proton beam. Emergency decay heat removal by natural air circulation cooling of the vessel outside is an attractive option for such a system. If no active or passive beam shut-off took place during a coolant overheating due to a significant Loss of Heat Sink (LOHS) accident, a core melt could eventually occur. Oxide fuel would probably mix with the heavy metal coolant with its high boiling point and circulate in the primary system in a coolable fashion. This type of scenario seems to have happened in a core melt accident in a Russian Alpha submarine with its lead / bismuth cooled critical reactor.
5.5
Paper 5
Decay Heat Removal by Natural Convection and Thermal Radiation from the Reactor Vessel Proc. of Accelerator Driven Transmutation Technologies and Applications ’99, 51
Prague, Czech Republic, (June 1999) Analytical investigations of the natural air circulation cooling at the external vessel of a heavy metal cooled Accelerator Driven System (ADS) are on the way. The preliminary demo design of Ansaldo and the air cooling of the earlier US advanced reactor program are considered. The fluid dynamics code Star-CD is used for the analysis. For the testing of the Star-CD code and the heat transfer correlations for coarse mesh calculations, the FZK Pasco experiments were analysed. Then a search for an optimal decay heat removal rate was made and parameters like the surface roughness, surface emissivity, and the gap width between the guard vessel and the collector wall have been investigated on a simplified geometry only containing the air channel and the vessel wall at constant temperature. Calculations on the complete geometry have been performed at two decay heat generation rates in order to find the equilibrium temperature in the reactor vessel (1% and 5% of normal operation power). Even for the higher rate the temperature in the reactor vessel remains below 1273 K.
5.6
Paper 6
Application of Burnable Absorbers in an Accelerator-Driven System Nuclear Science and Engineering, 137, No. 1, pp. 96-106, (Jan 2002) The application of burnable absorbers (BA) for minimization of power peaking, reactivity loss and capture to fission probabilities in an accelerator driven waste transmutation system (ADS,ATW) has been investigated. 10 B enriched B4 C absorber rods were introduced into a lead/bismuth cooled core fuelled with TRU discharges from light water reactors in order to achieve smallest possible power peakings at a BOL sub-criticality level of 0.97. Detailed Monte Carlo simulations show that a radial power peaking equal to 1.2 at BOL is attainable using a four zone differentiation in BA content. Using a newly written Monte Carlo Burnup code (MCB), reactivity losses were calculated to be 640 pcm per percent transuranium burnup, for unrecycled TRU discharges. Comparing to corresponding values in BA free cores, BA introduction diminishes reactivity losses in TRU fuelled sub-critical cores by about 20%. Radial power peaking after 300 days of operation at 1200 MWth power was less than 1.75 at a sub-criticality level of ∼ 0.92, which appears to be acceptable, with respect to limitations in cladding and fuel temperatures. In addition, the use of BA yields significantly higher fission to capture probabilities in even neutron number nuclides. Fission to absorption probability ratio for 241 Am equal to 0.33 was achieved in the configuration here studied. Hence, production of the strong α-emitter 242 Cm is reduced, leading to smaller fuel swelling rates and pin pressurization. Disadvantages following BA introduction, such as increase of void worth and decrease of Doppler feedback in conjunction with small values of βef f , need to be addressed by detailed studies of sub-critical core dynamics. 52
Chapter 6
Conclusions The defense in depth strategy remains a fundamental strategy also for ADSs, as well as diversity and redundancy. For the ADS considered in this thesis inherent safety features such as a low-pressure coolant excludes accident initiators. A very high boiling point, high density, and good natural circulation capability improves control and mitigation of severe accidents. Passive safety features such as natural convection cooling for emergency decay heat removal and a melt rupture disc were examined in order to increase grace periods during accidents. Both inherent and passive safety features increase diversity and redundancy of an ADS. The subcriticality of ADSs changes the core neutron dynamics since reactivity insertions have little effect unless they are very large. Of main importance for the dynamics, however, is that the accelerator determines the power level of an ADS. Due to the subcriticality and the Pb/Bi cooling, uncontrolled large power excursions cannot develop. Nevertheless, for larger inadvertent reactivity insertions the accelerator beam should be shut off shortly after accident initiation in order to avoid elevated temperatures. Normally this is performed actively but in case the interruption malfunctions, passive shut-off system can be used, e.g. a meltrupture disc. The subcriticality of an ADS is thus only a partially inherent safety mechanism. In unprotected loss-of-coolant and loss-of-heat-sink accidents a well designed critical systems behaves better. This is because its negative feedbacks shut these reactors down in such cooling accidents. The Ansaldo design, which is one of the topics of this thesis, is an 80 MWth Pb/Bi-cooled Accelerator-Driven System which has a vessel height of 8 m and a diameter of 6 m. The good natural circulation of this design leads to a temperature increase of only 85 K after a loss-of-flow accident. After a loss-of-power accident where also the accelerator is switched off the emergency decay heat removal can be easily achieved with both a PRISM-type and the new Ansaldo-type RVACS. The high heat capacity in relation to the power generated of this ADS makes a beam stop only necessary 40 minutes after a loss-of-heat-sink accident. If a melt-rupture disc is included and designed to break at 150 K above normal core outlet temperature 53
or ∼5 minutes after initiation of the loss-of-heat-sink accident, the grace period for beam shut-off is increased from 40 minutes to 6 hours. For the same vessel geometry, but an operating power of 250 MWth the structural materials can be kept below the temperature of 1173 K (no fast creep problems yet) in case the proton beam is shut off immediately. If beam shut-off is delayed, additional cooling methods are needed to improve the heat removal rate. For instance filling the gap between the guard and the reactor vessel with liquid metal coolant, in combination with water spray cooling on the guard vessel surface, helps significantly. The 200 MWth ADS-HXR has the heat-exchangers located in the risers of the reactor vessel. This is a safety advantage especially in the case when water is used as coolant in the secondary circuit. Hence the probability that steam bubbles are dragged into the core region is much lower in case of a heat-exchanger tube rupture. Also, if oil is used in the secondary circuit it could be a safety improvement, but this has to be investigated further. The emergency decay heat removal for both ADS-HXR and ADS-HXD is performed by an RVACS. The heat removal capacity at an average guard vessel temperature of 650 K is about 1.2 MW. After a totalloss-of-power accident the heat removal functions better for an ADS-HXR than for ADS-HXD (which has the heat-exchangers in the downcomers) because the coolant reverses and reaches a higher speed compared to the ADS-HXD after ∼1 hour. The flow reversal is due to the usage of bypass routes in parallel to the heat-exchangers. The core outlet temperature peaks at 664 K after almost 10 hours. A loss-of-heatsink accident requires a beam-shut off within 4 minutes to avoid exceeding the ASME level D (977 K). The grace period before fast vessel creep is 8 minutes for the same accident event. The 800 MWth ADS-HXR performs well after total-loss-of-power and loss-ofheat-sink accidents. However, for a loss-of-flow accident in which the accelerator beam is not switched off, the core outlet temperature (∼1700 K) becomes very high within less than one minute after accident initiation. Hence, this approach is not safe enough to be used for an 800 MWth ADS. Nevertheless, a critical reactor with the heat-exchangers in the riser would probably respond better since it would shut itself down because of the negative reactivity feedback as fuel temperatures rise. The excellent natural convection of an 800 MWth ADS with heat-exchangers in the downcomers is sufficient to provide adequate cooling during all unprotected cooling accidents. The location of heat-exchangers in the downcomers is a better solution for cooling larger ADSs during accidents where the accelerator beam is not switched off. However, the use of water as a secondary coolant is probably not possible for this design. Besides the use of oil, which can only be used for temperatures lower than about 700 K, one could use Pb/Bi in the secondary circuit to avoid any problems with steam or oil ingress into the ADS core.
54
Appendix A
The Ansaldo design The integral investigations of this paper are based on the design of an 80 MWth ADS demonstration facility by the Italian company Ansaldo [13]. Figure A.1 and Tab A present a schematic view and the principal aspects of this design.
Figure A.1. Schematic view of the Ansaldo design. The numbers in the figure represent: 1. Core 2. Reactor Vessel 3. Rotating Plug 4. Above Core Structure (ACS) 5. Target Unit 6. Sub-Assembly (SA) Transfer Machine 7. Intermediate HX 8. SA Handling Channel 9. SA Basket 10. Cover Gas Cooler.
55
Table A. Main characteristics of the Ansaldo ADS Demo design [13] Reference Solution Plant Area Power
Accelerator Target unit Fuel Coolant and moderator Steel, reactor vessel
80 MWth , kef f 0.97 at BOL kef f 0.94 at EOL Two-stage cyclotron scheme, max 3MW Pb/Bi eutectic, window type undecided U and Pu MOX Pb/Bi eutectic AISI 316L
This design is of pool type, with the reactor vessel including the heat exchangers. The vessel is 8 m tall and has an inner diameter of 6 m. A guard vessel surrounds the reactor vessel. A cyclotron delivers a 3 mA/600 MeV proton beam current to a Pb/Bi target where spallation reactions occur. The Pb/Bi coolant flow is mainly driven by natural convection, and the flow rate is increased and made better controllable by the injection of Argon bubbles above the core. The pressure drop over the core is 20 kPa at a mass flow rate of ∼5300kg/s, corresponding to a coolant velocity of 0.42m/s in the average core channel. The pressure drop is rather small due to the large pitch to diameter ratio of 1.58 [13]. The remaining flow path has a pressure drop of 9 kPa at a flow rate of ∼5800kg/s. The difference in mass flow rate is due to the ∼500kg/s bypass flow through the dummy zone. As mentioned in the Section ”Ultimate emergency decay heat removal” the ultimate backup system to remove decay heat the Ansaldo design is by natural air circulation around the guard vessel, i.e. a reactor vessel auxiliary cooling system (RVACS). A special feature of the Ansaldo RVACS is that it is physically separated from the guard vessel [13], see Fig.B.1. Eighty U-pipes, in which air flows by natural convection, are positioned concentrically around the guard vessel. The heat transfer between the guard vessel and the U-pipes is by thermal radiation and natural air convection.
56
Figure A.2. Schematic view of Ansaldo’s RVACS [13].
57
58
Appendix B
Theory and calculation methods B.1 B.1.1
Fluid dynamics models Mass and momentum conservation
In STAR-CD, the Navier-Stokes type equations are used for the mass and the momentum transfer [57]. If constant gravity is assumed, and a non-moving mesh is used, Eq B.1 and B.2 can be derived. ∂ ∂ (ρ) + (ρ) = sm ∂t ∂xj
(B.1)
∂ ∂p ∂ (ρui ) + (ρui − τij ) = − + si ∂t ∂xj ∂xi
(B.2)
k where τij = 2µsij − 23 µ ∂u ∂xk δij
∂ui and where δij = 12 ( ∂x + j
∂uj ∂xj )
The term si = gi (ρ − ρ0 ) when the buoyancy force is included in the momentum conservation equation.
B.1.2
Buoyancy driven flows
The density of liquid metals is nearly insensitive to pressure. Consequently, the Pb/Bi coolant is assumed to be isobaric and the Boussinesq approximation is used 59
in all calculations, i.e. the density depends linearly on the temperature and is independent of pressure, see Eq B.3. ρ=
ρ0 1 + βT (T − T0 )
(B.3)
On the contrary, the density of air is very sensitive to pressure, see Eq B.4 [60]. However, for thermal convection with open pressure boundaries where the density change depends on the temperature the Boussinesq approximation can still be used. ρ=
B.2 B.2.1
p RT
(B.4)
Boundary layer models The law of the wall representation
When a high Reynolds number formulation is used on a coarse mesh, wall functions or the law of the wall representations are used in the boundary layer in which the velocity, temperature, and turbulence are modelled. The wall function, u+ , is + modelled as linear up to a point, ym , from where a logarithmic relationship will follow to half the cell width in the layer next to the wall [57], see Eq B.5. + + y , y + ≤ ym + (B.5) u = 1 + + + κ ln(Ey ) , y > ym + The u+ is the wall function and the ym is calculated to satisfy Eq B.6.
1 + ln(Eym )=0 (B.6) κ The thickness of the cell layer next to the wall should be chosen so that the computational boundary or the center of the cell falls within the range 30< y + <100 [72]. + ym −
B.2.2
The Two-Layer model
In order to avoid fine grids over the whole computational mesh, several thin cell layers is sometimes constructed adjacent to the wall. The objective is to save computational time and to model the boundary layer more correctly. The TwoLayer approximation is based on the idea that the viscous and turbulence effects are only significant close to solid surfaces. Consequently, only dissipative terms in the direction normal to the surface need to be retained. In order to resolve the velocity profile within the Two-Layer approximation, it is necessary to place the grid point nearest the at a location y + < 5, i.e. in the laminar sublayer[73]. Similarly, the proper resolution of the temperature profile in the normal direction may require the nearest grid point to the wall to be located such that y + < 1. 60
B.3
Heat transfer models
The heat transfer is implemented through the enthalpy conservation equation. If no chemical reactions appear, Eq B.7 is applied for a general fluid mixture. The energy equations for the fluid and solid are solved simultaneously and continuity of energy flux is enforced at the fluid/solid interfaces. ∂ ∂ ∂ ∂p ∂ui (ρht ) + (ρht − Fht ,j ) = (p) + + τij + sh ∂t ∂xj ∂t ∂xj ∂xj
(B.7)
On a coarse mesh, wall functions are employed to model the velocity and temperature profiles in the boundary layer. STAR-CD’s standard heat transfer coefficient is calculated as Eq B.8. Ht =
c¯p ρuc Fφ,w = φ − φw σφ,t (u+ + P )
(B.8)
To take into account the different Prandtl numbers in the boundary layer and in the bulk of the flow, a sublayer resistance factor, P, is added when the heat transfer coefficient is calculated. It is an additional resistance that takes into account the reduction of heat transfer in the laminar sublayer [74], see Eq B.9. P ≡ 9.24[(
σφ 3/4 −0.007σφ ) − 1][1 + exp( )] σφ,t σφ,t
(B.9)
For Two-Layer models, the heat transfer between a solid-fluid interface is calculated with the energy equation, where the velocity terms for the solid are set to zero [57]. An interpolation practice takes into account that there is a change of material properties, and that the heat transfer is continuous over the interface.
B.4
Turbulence models
For turbulent flow the molecular diffusion of mass and heat are modelled as Eq B.10. Fht ,j ≡ k
X ∂mm ∂T − ρ¯u0j¯h0t + hm,t ρDm ∂xj ∂xj m
(B.10)
where the middle terms contains thermal enthalpy fluctuations. The molecular diffusion equation feeds into the equation conservation of static enthalpy, see Eq B.7.
B.4.1
Standard k − model equations (linear)
The complexity of the governing equations motivates the use of a general and cost efficient turbulence model, for example a type of k − formulation. The linear k − 61
models in STAR-CD assumes that the Reynolds stresses and scalar fluxes are linked similarly as their laminar counterparts, see Eq B.11-B.13 [57]. 2 ∂uk −¯ ρu0i¯u0j = 2µt sij − (µt + ρk)δij 3 ∂xk
(B.11)
µt ∂h ρ¯u0j¯h0 = − σh,t ∂xj
(B.12)
where k≡
u0i¯u0j 2
(B.13)
k is the turbulent kinetic energy. The turbulent viscosity is then linked to k and via Eq B.14. µt = fµ
Cµ ρk 2
(B.14)
The turbulent Prandtl number also contains empirical quantities, which usually are constants. In the boundary layer a ’law of the wall’ representation is used, see B.2.1. The transport equations can be seen the two following sections. Turbulence energy With the relations in Section B.4.1, the thermal enthalpy equations can be modelled as Eq B.15. ∂ µef f ∂k 2 ∂ui ∂ui ∂ (ρk) + (ρ − ) = µt (P + PB ) − ρ − (µt + ρk) ∂t ∂xj σk ∂xj 3 ∂xi ∂xi
(B.15)
where µef f = µ + µt ∂ui ∂xj
(B.17)
gi 1 ∂ρ σh,t ρ ∂xi
(B.18)
P ≡ 2sij PB ≡ −
(B.16)
σk is an empirical coefficient. The first term on the right hand side of Eq B.15 represents the turbulent generation by shear and normal stresses, the second viscous dissipation, and the third amplification due to compressibility effects. 62
Turbulence dissipation rate
∂ µef f ∂ ∂ (ρ) + (ρ − = ∂t ∂xj σ ∂xj =
2 ∂ui ∂ui 2 ∂ui [(C1 P + C3 PB ) − (µt + ρk) ] − C2 ρ + C4 ρ k 3 ∂xi ∂xi k ∂xi
(B.19)
where σ , C1 , C2 , C3 and C4 are empirical coefficients, see Tab B.4.1. Table B.4.1. Coefficients employed in the standard k − model. σk σ C1 C2 C3 C4 Cµ 0.09 1.0 1.22 1.44 1.92 0.0∗ -0.33 ∗
κ
E
0.42
9.0∗∗
C3 = 1.44 for PB > 0 and zero other wise. For smooth walls.
∗∗
B.4.2
The Two-Layer model
In the Two-Layer model, the kinetic dissipation k is calculated in the transport equation as for the k − approach, whereas is obtained from an algebraic function. The solution is matched to the k− equations at the ’edge’ of the viscosity-influenced region. STAR-CD contains several models; the Norris & Reynolds model [75] is used in these investigations see Eq B.20-B.23. Wolfstein’s [76] and Hassid & Poreh’s [77] models were used too, but no significant difference of results was found in these investigations. =
C k 3/2 (1 + ) l Rey
fµ = 1 − exp(−
B.5 B.5.1
1 Rey ) Aµ
(B.20) (B.21)
√ ky Rey = ν
(B.22)
l = κCµ−0.75 y
(B.23)
Droplet model Direct liquid-solid contact heat transfer
The heat transfer per unit area between a hot surface and a liquid is calculated as Eq B.24. [78] 63
Ts,0 − Ti 00 qs,i (t) = ks √ παs t
(B.24)
The total heat transfer rate is then proportional to the droplet’s spreading area according to Eq B.25. 00 q(t) = qs,i (t)As (t)
B.5.2
(B.25)
Heat transfer by dispersed spray cooling
When droplets impact on a surface without interference, the amount of heat removed by the droplets is proportional to the number of droplets. If interaction between the droplets occurs, this has to be considered. The interactions can be categorized in two groups; first spreading interference, when droplets impact close to each other, and impact interference, when droplets impacts on top of other droplets. Spreading interference affects the heat transfer since the spreading diameter of the droplet is reduced, whereas impacting interference has negligible effect on heat transfer according to experiments [79]. At a mass flux of 0.025 g/cm2 s, the droplet interference starts affecting the heat transfer noticeably. Figure B.1 displays the heat transfer effectiveness as a function of mass flux. Note that at 0.025 g/cm2 s the effectiveness starts to level off.
Figure B.1. The heat transfer efficiency as a function of water spray mass flux [78].
The effectiveness of the heat transfer is defined as Eq B.26. =
Q md (hf g + cp ∆Tsub )N
(B.26)
Several parameters affect the heat transfer effectiveness, see Eq B.23. For example the droplet sub-cooling and the droplet size. A comparison where the mass 64
flux is fixed shows that the heat transfer increases with droplet diameter [79]. This is due to spreading dynamics of the droplet. The droplets’ velocities is another important factor. The relationship between the heat transfer effectiveness and the Weber number is shown in Fig B.2. The higher the velocity of the impinging droplets, the more they spread at impact and consequently the heat transfer efficiency is augmented. In terms of efficiency, it is of interest to reach a We number of about 80, which corresponds to a velocity of 1.7 m/s.
Figure B.2. The heat transfer efficiency as a function of the Weber number [78].
B.6
Thermal radiation
STAR-CD uses so called ’patches’ that are defined on boundaries of the calculation domain to model thermal radiation. From each patch center a chosen number of beams are emitted uniformly in all directions. Then, each beam is traced until it reaches an opposing patch, thus defining a pair, which may undergo radiation exchange. The amount of energy transported is then calculated from the radiation transport equation and the boundary conditions.
B.6.1
View factor calculation
The view factor Fij , is the fraction of the total radiation leaving patch i, which is received at patch j, see Eq B.27. NL,i
Fij =
X
k=1
65
αk fij
(B.27)
Where f is the view factor of a single beam, α is set to 1 if the beam strikes patch j and zero otherwise [57]. The sum of F over all j for a value of i is always one, since the quantity of the radiant energy should be conserved.
B.6.2
Radiant fluxes
The total radiation leaving patch i is the sum of the emitted energy and the reflected incoming energy as is given by Eq B.28 [57], see Fig B.3. Ji = r EB,i + ρr Ii
(B.28)
and the net radiant heat flux at the wall, is given by the difference between the absorbed incoming flux and the emitted flux, see Eq B.29 [65]. qr00 = Ii − Ji = αr I − r EB,i
(B.29)
Figure B.3. Radiative heat transfer at patch i. [65].
B.6.3
Combined radiation and convection heat transfer
The total heat flux from the wall is then the sum of the radiation and the convection heat transfer, see Eq B.30. [65] 00 qw = qr00 + qc00
(B.30)
qc00 = −h(Tw − Tf )
(B.31)
00 qw = αr Ii − r σTw4 − h(Tw − Tf )
(B.32)
where the convection term is
Eqs B.30 and B.31 combined gives
66
B.7
Computation methods
In general, the governing equations, auxiliary conditions (initial and boundary) are well posed mathematically if three conditions are met, namely: the solution exists, the solution is unique, and the solution depends continuously on the auxiliary data. Similarly, a computation is well-posed if three conditions are met; if the computational solution exists, the computational solution is unique, and the computational solution depends continuously on the approximate auxiliary data [73].
B.7.1
Boundary and Initial Conditions
The auxiliary data are the starting point to obtain the interior solution, particularly for transient problems [73]. The auxiliary conditions can be specified in three ways. 1. Dirichlet condition, e.g. u = f on ∂R 2. Neumann (derivative) condition, e.g. ∂u/∂n = f or ∂u/∂s = g on ∂R 3. Mixed or Robin condition, e.g. ∂u/∂n + ku = f , k > 0 on ∂R. In auxiliary conditions 2 and 3, ∂/∂n denotes the outward derivative.
B.7.2
Discretisation schemes
To be able to solve the governing equations numerically they have to be discretised. This is a conversion of the continuous partial differential equations into a discrete system of algebraic equations. The Finite Difference Method (FDM) is the most direct way of discretisation in which the derivatives are replaced with FDM expressions. For example the transient head conduction equation would be discretised as Eq B.33 and B.34. Tjn+1 − Tjn ∂T = ∂t ∆t
(B.33)
and α
n n α(Tj−1 − 2Tjn + Tj+1 ) ∂ 2 T¯ = ∂x2 ∆x2
(B.34)
If the solution method is explicit, the equations are then manipulated and the unknown value is represented in an algorithm. Again, the transient heat conduction equation would be as Eq B.35. 67
Tjn+1 = Tjn +
α∆t n n (T − 2Tjn + Tj+1 ) ∆x2 j−1
(B.35)
Apart from the FDM, the Finite Element Method (FEM), Finite Volume Method (FVM) and spectral methods are commonly employed. Usually the time derivatives are discretised using the FDM, whereas spatial derivatives are discretised either by FDM, FEM, FVM, or spectral methods [73]. In STAR-CD, the differential equations that govern for example the conservation of mass, momentum, and energy within the fluid are discretised by the FVM. This means that they are first integrated over the individual computational cells and thereafter approximated in terms of the cell-centered nodal values of the dependent variables. The advantage of the FVM is that it preserves the conservation properties of the parent differential equations. For the FV discretisation it is convenient to work with a general coordinate free form of the conservation equations [57], see Eq B.40. 1 ∂ √ →ρΦ − Γ gradΦ) = s ( gρΦ) + div(− u √ r Φ Φ g ∂t
(B.36)
where Φ stands for any of the dependant variables, i.e. ui , k, m1 etc. An exact form of Eq B.40 valid for an arbitrary time-varying volume V bounded by a moving closed surface S can be written as Eq B.37. d dt
Z
ρΦdV +
V
Z
S
→ →Φ − Γ gradΦ)d− (ρ− u S = r Φ
Z
sΦ dV
(B.37)
V
From here, the approximations are introduced. The first term on the left hand side is discretised as Eq B.41. T1 ≈
(ρΦdV )nP − (ρΦV )0P δt
(B.38)
where the superscripts 0 and n refer to ’old’ and ’new’ time levels, respectively. The second term in Eq B.37 left hand side is split into Cj and Dj due to convection and diffusion, respectively, each is expressed in terms of average values over cell faces. T2 ≈
X j
X → → − − → →Φ− (ρ− u (ΓΦ Φ )d S = r S )j − j
Z
sΦ dV
(B.39)
V
The third term in B.37 may in general contain components representing sources or sinks of the transported property, and additional flux terms, see Eq B.43. T3 ≈ s1 − s2 ΦP 68
(B.40)
The monotone advection and reconstruction scheme (MARS) The spatial flux discretisation can be performed with both lower-order and higher order schemes. Lower-order schemes yield equations that are easy to solve. Their solutions follow the boundary conditions, but lead sometimes to numerical diffusion [73]. This effect is reduced with a refined mesh. A higher order scheme results in equations that are more difficult to solve but can resolve steep gradients better [73]. Near these gradients can numerical dispersion or non-physical oscillations occur. Generally, for correct solutions, second or higher order schemes ought to be used for the discretisation of momentum and energy equations. The upwind scheme can be used for the turbulent equations to increase the numerical stability. The calculations included in the licentiate thesis have primarily used a MARS scheme [57], whereas an upwind scheme was used for the calculations for the doctoral thesis due to that numerical instabilities were encountered for higher order schemes. The MARS, is a multidimensional second order differencing scheme, which functions in two steps, namely reconstruction and advection [57]. The reconstruction step consists of calculating a set of monotone gradients using a Total Variation Diminishing scheme. Generally, monotone schemes can only be first-order accurate in space. Therefore, such schemes are diffusive, and shock profiles are smeared and accurate solutions cannot be obtained without reasonable grid refinement. However, an improvement in accuracy, without loosing the strong theoretical foundation, can be achieved by replacing the monotonicity preserving requirement with the Total Variation Diminishing (TVD) requirement. The total variation of the numerical solution is defined by Eq B.41. T V (un ) =
∞ X
j=−∞
|unj+1 − unj |
(B.41)
Consequently a numerical scheme is TVD if Eq B.42 is valid. T V (un+1 ) ≤ T V (un )
(B.42)
TVD schemes do not generate spurious oscillations and can achieve second order accuracy where the solution varies smoothly. Together with the cell flow properties this completely defines a second-order accurate spatial discretisation. In the advection step, the new cell-face flow properties are used to calculate the face fluxes for all advocated properties using a monotone and bounded advection scheme [57]. This incorporates a variable compression level, which controls a second order upwind scheme affecting the order of accuracy. Of all schemes that are supported by STAR-CD, the MARS scheme is the least sensitive to mesh structure and skewness. The upwind scheme The upwind scheme is of first-order for which the information is collected upstream, whereas downstream information is ignored. This type of interpolation preserves 69
the correct physical bounds under all conditions, but leads to more numerical diffusion than higher order schemes. Since upwind schemes add numerical viscosity or diffusion to the equations, thus the viscosity solution is also smeared. The upwind scheme is usually applied for problems where there is a directional bias in the propagation speed.
B.7.3
Convergence and consistence
The difference between the exact solution of the partial differential equation and the exact solution of the system of algebraic equations is called the solution error, it can be denoted by enj , see Eq B.43. enj = T¯(xj , tn ) − Tjn
(B.43)
For the equations that govern fluid flow the convergence is very difficult to demonstrate theoretically [73]. But, problems for which an exact solution exists, like the diffusion equation, numerical solutions can be obtained on a successively refined grid for which solution errors are computed. This is a rather expensive process since usually very fine grids are necessary. The system of algebraic equations generated by the discretisation process is said to be consistent with the original partial differential equation if, in the limit that the grid spacing tends to zero, the system of algebraic equations is equivalent to the partial differential equation at each grid point.
B.7.4
Stability
Growth, or decay, of errors introduced at any stage of the computation is a measure of stability. The errors can for example arise due the computer not being able to give answers to an infinite number of decimal places. The error can be estimated using standard methods. However, the numerical solutions are usually more accurate than these estimates indicate, because the stability analyze often assumes the worst possible combination of individual errors [73]. The higher order schemes have often more severe stability restrictions then those based on lower order discretisations.
70
Bibliography [1] Delpech, M., ”The Am and Cm transmutation - physics and feasibility”, Proc. Int. Conf. on Future Nuclear Systems, GLOBAL 99, (1999) [2] US Department of Energy, ”A Roadmap for Developing Accelerator Transmutation of Waste (ATW) Technology”, DOE/RW-0519, (Oct. 1999) [3] Galperin, A. and Raizes G., ”A pressurized Water Reactor Design for Plutonium Incineration: Fuel Cycle Options”, Nuclear Technology, 117, pp. 125132, (Feb. 1997) [4] Kaesemeyer J.J. et al, ”Comparison of Pressurized Water Reactor Core Characteristics for 100 % Plutonium-Containing Loadings, Nuclear Technology, 122, (April 1998) [5] US Department of Energy, ”Report to Congress on Advanced Fuel Cycle Initiative: The Future Path for Advanced Spent Fuel Treatment and Transmutation Research”, (Jan. 2003) [6] Wider, H.U., Karlsson, J., Safety Aspects of Heavy Metal-Cooled AcceleratorDriven Waste Burners, Journal de Physique IV, 9, (Apr. 1999) [7] Eriksson, M., and J. Cahalan, ”Inherent shutdown capabilities in acceleratordriven systems”, Annals of Nuclear Energy, 29/14, pp. 1689-1706, (May 2002) [8] The European Technical Working Group on ADS, ”A European Roadmap for Developping Accelerator Driven Systems (ADS) for Nuclear Waste Incineration”, ISBN 88-8286-008-6, (2001) [9] Van Tuyle, G.J., Slovik, G.C., Chan, B.C., Kennett, R.J., Cheng, H.S., and Kroeger, P.G., Summary of Advanced LMR Evaluations - PRISM and SAFR, Brookhaven National Laboratory, pp. 87-91, (Oct. 1989) [10] Van Tuyle, G.J., Slovik, G.C., Chan, B.C., Aronson, A.L., Kennett, R.J., Evaluations of 1990 PRISM Design Revisions, Brookhaven National Laboratory, pp. 4-28, (March 1992) 71
[11] Baumeister, E.B., et al., Inherent Safety Features and Licensing Plan of the SAFR plant, Proc. Int. Conf. on Fast Breeder Systems, (Sept. 1987) [12] Carlsson, J., ”Decay Heat Removal from the Guard Vessel by Thermal Radiation and Natural Convection”, Licentiate Disseration, TRITA-Fys 3072, Stockholm, (2000) [13] Cinotti, L., and Corsini, G., A proposal for enhancing the primary coolant circulation in an ADS, unpublished [14] Spencer, B.W., Hill, R.N., Wade, D.C., Hill, D.J., Sienicki, J.J., Khalil, J.E., Cahalan, J.E., Farmer, M.T., Maroni, V.A., Leibowitz, L., ”An Advanced Modular HLMC Reactor Concept Featuring Economy, Safety, and Proliferation Resistance”, Proc. of the 8th Int. Conf. on Nucl. Eng., Baltimore, USA, (April 2-6 2000) [15] Juhn, P.E., Kupitz, J., Cleveland, J., Cho, B., Lyon, R.B., ”IAEA activities on passive safety systems and overview of international development”, Nuclear Engineering and Design, 201, pp. 41-59, (2000) [16] Frish, W., Gros, G., ”Improving the safety of future nuclear fission power plants”, Fusion Engineering and Design, 56-57, pp. 83-93, (2001) [17] prepared by Nuclear Energy Reseach Avisory Committee and Generation IV Int. Forum (GIF), ”Final Generation IV Goals”, http://gen-iv.ne.doe.gov [18] Gail, H.M., ”Considering the next generation of nuclear power plants”, Progress in Nuclear Energy, 37, No. 1-4, pp. 5-10, (2000) [19] The Pebble Bed Modular Reactor home page, http://www.pbmr.com/ [20] Frameatome ANP GmbH, ”SWR 1000”, anp.com/anp/e/foa/anp/products/s11 2.htm
http://www.de.framatome-
[21] Zrodnikov, A.V., Chitaykin, V.I., Gromov, B.F., Grigoriev, O.G., Dedoul, A.V., Toshinsky, G.I., Dragunov, Yu.G., Stepanov, V.S., ”Multipurposed Reactor Module SVBR-75/100”, Proc. of 8th Int. Conf. on Nuclear Engineering, Baltimore, USA, (April 2000) [22] ICRP/90/G-01, (1990) [23] International Nuclear Safety Advisory Group (INSAG), ”Basic Safety Principles of Nuclear Power Plants”, IAEA Safety Series No. 75-INSAG-3, Vienna, (1988) [24] Bennet, D.J., Thomson, J.R., ”The Elements of Nuclear Power”, Longman Scientific & Technical, ISBN: 0-582-02224-X, 1991 72
[25] Wall, I.B., Haugh, J.J., Worlege, D.H., ”Recent applications of PSA for managing nuclear power plant safety”, Progress in Nuclear Energy, 39, No. 3-4, pp. 367-425, (2001) [26] Westln, D., Gudowski, W., Wallenius, J., Tuˇcek, K., ”A Cost Benefit Analysis of an Accelerator Driven Transmutation System”, Nuclear Applications in the New Millennium (AccAPP-ADTTA’01), Reno, Nevada, USA, (Nov. 2001) [27] Stevens, G., ”Nuclear Energy and Sustainability”, Sustainable Development - OECD Policy Approaches for the 21st Century, Cp. 10. Organisation for Economic Co-operation and Development, Paris, France, (1997) [28] International Energy Agency, Nuclear Power: Sustainability, Climate Change and Competition, ISBN: 92-64-16954-7, (1998) [29] Barthel, F.H., Dahlkamp, F.J., ”Thorium deposits and their availability”, IAEA TECDOC-650, Vienna, pp. 104-115, (1992) [30] Wallenius, J., Lecture in transmutation physics - Nuclear Waste and Radiotoxicity, http://www.neutron.kth.se/courses/transmutation/NuclearWaste.pdf [31] Forsberg, C.W., ”Rethinking High-Level Waste Disposal: Separate Disposal of High-Heat Radionuclides (90 Sr and 137 Cs)”, Nuclear Technology, 131, pp. 252-268, (Aug. 2000) [32] Kadak, A.C., Ballinger, R.G., Alvey, T., Kang, C.W., Owen, P., Smith, A., Wright, M., Yao, X., ”Nuclear Power Plant Design Project - A Response to the Environmental and Economic Challange of Global Warming”, Massachusetts Institute of Technology, Independent Activities Period, (Jan. 1998) [33] Koch, L., ”Criteria to Assess the Proliferation Resistance of Nuclear Fuel Cycles”, Int. Topical Workshop on Proliferation-Resistance in Innovative Reactors and Fuel Cycles, Como, Italy, (July 2001) [34] DeVine, J.C. Jr., ”Conceptual benefits of passive nuclear power plants and their effect on component design”, Nuclear Engineering and Design, 165, pp. 299-305, (1996) [35] Chang, S.H., Choi, S.S., Park, J.K., Heo, G., Kim, H.G., ”Development of an advanced human-machine interface for next generation nuclear power plants”, Reliability Engineering and System Safety, 64, pp 109-126, (1999) [36] Nuclear Energy Agency, ”Accelerator-Driven Systems (ADS) and Fast Reactors (FR) in Advanced Nuclear Fuel Cycles - A Comparative Study”, ISBN: 92-64-18482-1, Available on the Web at: http://www.nea.fr/html/ndd/reports/2002/nea3109.html 73
[37] Ion, Sue, E., Bruschi, Howard, J., Matzie, Regis, A., ”Meeting the energy challange through innovations in passive nuclear power technology”, World Energy Council, 18th Congress, Buenos Aries, (Oct. 2001) [38] Birkhofer, A., ”Advanced Power Reactors with Improved Safety Characteristics”, Appl. Radiat. Isot., 46, No. 6/7, pp. 701-706, (1995) [39] Kadi, Y., Buono, S., ”The Energy Amplifier Demonstration Facility: neutronic analysis of high-energy beam - target interaction”, Conf. on Heavy Liquid Metal Coolants in Nuclear Technology, Obninsk, Russia, (1998) [40] The Royal Academy of Engineering and The Royal Society, ”Nuclear energy the future climate”, ISBN: 0-85403-526-5, (June 1999) [41] Carluec, B., Fiorini, G-L, Rodriguez, C.R., ”Preliminary safety analysis of the gas-cooled ADS concept”, IAEA workshop on gas-cooled reactors, (July 2000) [42] Adamov, E.O., Orlov, V.V., ”Naturally Safe Lead-Cooled Fast Reactor for Large-Scale Nuclear Power”, Moscow, (2001) [43] Gromov, B.F., Toshinsky, G.I., Stepanov, V.S., et al.,”Use of Lead-Bismuth Coolant in Nuclear Reactors and Accelerator-Driven Systems”, Nuclear Engineering and Design, 173, pp. 207-217, (1997) [44] Zrodnikov, A.V., Chitaykin, V.I., Toshinsky, G.I., Grigoriev, O.G., Dragunov, U.G., Stepanov, V.S., Klimov, N.N., Kopytov, I.I., Krushelnitsky, V.N., Grudakov, A.A., ”Lead-Bismuth Cooled Fast Neutron Reactors for Using in Nuclear Power and as One of the Trends of Realizing the President’s Initiative”, Ministry of the Russian Federation, (2002) [45] Li, N., ”Active control of oxygen in molten lead-bismuth systems to prevent steel corrosion and coolant contamination”, J. Nuclear Materials, 300, pp. 73-81, (2002) [46] He, X.Y., Li, N., Mineev, M., J. Nuclear Materials, 297, pp. 214, (2001) [47] Adamov, E.O., Orlov, V.V., “Nuclear development on the basis of new concepts of nuclear reactors and fuel cycle”, Proc. Heavy Liquid Metal Coolants in Nuclear Technology HLMC-98, Vol. 1, SSC RF-IPPE, Obninsk, pp. 24, (1999) [48] Novikova, N., Pashkin, Y., Chekunov, V., ”Some features of subcritical blankets cooled with lead-bismuth”, Proc. 3rd Int. Conf. Accelerator-Driven Transmutation Technologies and Applications, ADTTA 99, (1999) [49] Rousanov, A.E., et al., ”Design and study of cladding steels for fuel elements of NPP using heavy coolant”, Proc. Heavy Liquid Metal Coolants in Nuclear Technology, HLMC 98. IPPE, (1998) 74
[50] Wallenius, J., Tucek, K., Carlsson, J., Gudowski, W., ”Application of burnable absorbers in an accelerator driven system”, Nuclear Science and Eng., 137, No. 1, pp. 96-106, (Jan. 2002) [51] Gromov, B.F., Grigoriev, O.G., Dedoul, A.V., Toshinsky, G.I., Stepanov, V.S., Nikitin, L.B., ”The Analysis of Operating Experience of Reactor Installations Using Lead-Bismuth Coolant and Accidents Happened”, Orlov, Heavy Liquid Metal Coolants Conference in Nuclear Technology, Vol.1, Obninsk, Russia, (Oct. 1998) [52] L. Likhanskij et. al.: ”Modeling the Dynamics of Fuel Fragmentation in Liquid Lead in Analyzing Core Disintegration During A Fast-Reactor Accident”, Atomic Energy, 78, No.1 (1995) [53] Van Tuyle, G.J., Slovik, G.C., Chan, B.C., Kennett, R.J., Cheng, H.S., and Kroeger, P.G., ”Summary of Advanced LMR Evaluations - PRISM and SAFR, Brookhaven National Laboratory”, pp. 87-91, (Oct. 1989) [54] Landry, R.R., King, T.L., Wilson, J.N., ”Draft Preapplication - Safety Evaltuation Report for Power Reactor Inherently Safe Module Liquid Metal Reactor”, NUREG-1368, (1989) [55] Hunsbedt, A., Magee, P.M., ”Design and performance of the PRISM natural convection decay heat removal system”, Proc. Int. Topical Meeting on Safety of Next Generation Power Reactors, pp. 844-851, (1988) [56] Wider, H., Wilkening, H., Maschek, W., ”Safety Advantages of Heavy Metal Versus Gas-Cooled Accelerator-Driven Systems”, Proc. 3rd Int. Conf. Accelerator-Driven Transmutation Technologies and Applications, ADTTA 99, (1999) [57] Computational Dynamics Ltd., 1999, Methodology Volume 3.10 [58] Cheng. X., Mller, U., ”Turbulent natural convection coupled with thermal radiation in large vertical channels with asymmetric heating”, Int. J. Heat Mass Transfer, 41, No. 12, pp. 1681-1692, (1998) [59] Grtzbach, G., Cheng, X., ”Development progress of the FLUTAN code for modeling heat transfer in LWR-systems” (in German), Proc. Annual Meeting of Nuclear Technology ’96, Mannheim, Germany, pp. 168-171, (May 1996) [60] Fox, W.F., McDonald, A.T., Introduction to Fluid Mechanics, John Wiley and Sons, New York, pp. 63, ISBN: 0-471-01909-7, (1978) [61] Carlsson, J., Wider, H., ”Emergency Decay Heat Removal from an AcceleratorDriven System”, Nuclear Technology, 140, No. 1, pp. 28-40, (Oct. 2002) 75
[62] Assessment of computational fluid dynamics codes for heavy liquid metal coolants, FIKW-CT-2001-80121, http://www.cordis.lu/ [63] King, T.L., Landry, R.R., Throm, E.D., Wilson, J.N., ”Preapplication safety evaluation report for the sodium advanced fast reactor (SAFR) liquid metal reactor”, Office of Nuclear Regulation, pp. 4-14, (Sept. 1991) [64] Shibli I A, European Creep Collaboration Committee (ECCC), ECCC coordination, ECCC Data Sheets, (1999) [65] Incropera, F.P., DeWitt, D.P., ”Fundamentals of Heat and Mass Transfer”, Johan Wiley and Sons, New York, ISBN: 0-471-30460-3, pp. 657, (1996) [66] Toshinsky, G.I., Grigoriev, O.G., Efimov, E.I., Leonchuk, M.P., Novikova, N.N., Pankratov, D.V., Skorikov, D.E., Klimov, N.N., Stepanov, V.S., ”Safety Aspects of SVBR-75/100 Reactor”, Workshop Proc. of Advanced Nuclear Reactor Safety Issues and Research Needs, Paris, France, (Feb. 2002) [67] Jeong, J.H., Kweon, Y.C., ”The effect of tube rupture location on the consequences of multiple steam generator tube rupture event”, Annals of Nuclear Energy, 29, pp. 1809-1826, (2002) [68] US Nuclear Regulatory Commission, ”Fact Sheet on Steam Generator Tube Issues”, http://www.nrc.gov/reading-rm/doc-collections/factsheets/steam-gen.html [69] Karlsson, J., ”Decay heat removal by natural convection and thermal radiation from the reactor vessel”, Proc. of ADTTA ’99, Prague, Czech Rep., (1997) [70] Pershagen, B., ”Light water reactor safety”, Pergamon Press, Oxford, England, ISBN 0-08-035915-9, pp.131, (1989) [71] Eckert, E.R.G., Drake, P.M., Heat and Mass Transfer, McGraw-Hill Book Compnany, New York, pp. 299, (1959) [72] Launder, B.E., and Spalding, D.B., ”The numerical computation of turbulent flows”, Computer Methods in Applied Mechanics and Engineering, 3, pp. 269289, (1974) [73] Fletcher, C.A.J., Computational Techniques for Fluid Dynamics, Vol. 1., Springer Verlag, Berlin, ISBN: 3-540-533058-4, (1997) [74] Jayatilleka, C.L., ”The influence of Prandtl number and surface roughness on the resistance of the laminar sub-layer to momentum and heat transfer”, NASA TM-105993, (1969) [75] Norris, L.H., Reynolds, W.C., ”Turbulent channel flow with a moving wavy boundary”, Report No.FM-10, Dep. of Mech. Eng., Stanford University, USA, (1975) 76
[76] Wolfstein, M., ”The velocity and temperature distribution in one-dimensional flow with turbulence augmentation and pressure gradient”, Int. J. Heat Mass Transfer, 12, pp. 301-318, (1969) [77] Hassid, S., Poreh, M., ”A turbulent energy dissipation model for flows with drag reduction”, J. Fluids Eng., 100, pp. 107-112, (1975) [78] Delcorio, B., Choi, K-J., ”Analysis of Direct Liquid-Solid Contact Heat Transfer in Monodispersed Spray Cooling”, J. of Thermophysics, 5, No. 4, (Oct.-Dec. 1991) [79] Yao, S.C., Choi, K.J., ”Heat Transfer Experiments of Mono-Dispersed Vertically Impacted Sprays”, Int. J. Multiphase Flow, 13, No. 5, pp. 639-648, (1978)
77
78
1
1
Emergency Decay Heat Removal by Reactor Vessel Auxiliary Cooling System from an Accelerator-Driven System Johan Carlsson
∗
Royal Institute of Technology, Department of Nuclear and Reactor Physics, S-10691 Stockholm, Sweden
Hartmut Wider Joint Research Center, Institute for Energy, NL-1755 LE Petten, Netherlands (4 May, 2002)
Abstract The passive emergency decay heat removal during severe cooling accidents in Pb/Bi-cooled 80MWt and 250MWt Accelerator-Driven System designs was investigated with the CFD code STAR-CD. For the 80MWt design, the calculations show that no structural problems occur as long as the accelerator proton beam is switched off immediately after accident initiation. A delay of beam stop by 30 minutes after a combined Loss-Of-Heat-Sink and Loss-Of-Flow accident would lead to reactor vessel temperatures in the secondary creep domain, which do not cause severe structural problems. By using a melt-rupture disc on the vacuum pipe of the accelerator proton beam to interrupt the beam at elevated temperatures in a passive manner the grace time before beam stop is necessary is increased from 30 minutes to 6 hours. An emergency decay heat removal design which would prevent radioactive release to the atmosphere even more reliably than the PRISM design was also investigated. For an ADS of 250MWt power with the same vessel as the 80MWt ADS examined, the maximum wall temperature reaches 745K after an immediate beam stop. This does not cause any structural problems either. The grace time until a beam stop becomes necessary for the 250MWt system was found to be about 12 minutes. To reduce elevated vessel temperatures more rapidly after a beam stop alternative cooling methods were investigated, for example filling of the gap between the reactor and the guard vessel with liquid metal and the simultaneous use of water spray cooling on the outside of the guard vessel. This decreases the coolant temperatures already within minutes after switching off the proton beam. The use of chimneys on the Reactor Vessel Auxiliary Cooling System, which increase the air flow rate lowers the maximum reactor vessel wall temperature only by about 20K. It can be concluded that the critical parameter for the emergency cooling of an ADS is the time delay in switching off the accelerator after accident initiation.
∗
Email:
[email protected]
1
II. NATURAL AIR CONVECTION COOLING OF AN 80MWT ADS USING A PRISM TYPE RVACS
I. INTRODUCTION
This paper investigates the emergency decay heat removal by a Reactor Vessel Auxiliary Cooling System (RVACS) of a Pb/Bi-cooled Accelerator-Driven System (ADS) [1]. The RVACS will be particularly needed in station blackout accidents (Loss-Of-Power, LOP) where the normal heat removal by the secondary system is unavailable (Loss-Of-Heat-Sink, LOHS) and the primary flow is lost (Loss-Of-Flow, LOF). In this accident the accelerator is normally switched off either by active or passive means. However, it may also be possible that the primary pumps stop running and the accelerator continues to operate either because this accident condition was not noticed or because the passive beam shut-off system did not work. An even less likely event is a combined LOF and LOHS without beam shut-off. The combination of LOF and LOHS occurs in a station blackout/Loss-of-Power accident, but without off-site electricity, the accelerator will also be shut-off unless it has autonomous power supply. The basic principle of a RVACS is that of air circulation around the guard vessel by natural convection. This is a passive system, which is governed only by physical laws, hence operational failures are extremely unlikely. RVACS or similar systems were planned for the Power Reactor Inherently Safe Module (PRISM) [2] and the Sodium Advanced Fast Reactor (SAFR) [3]. As the name implies, an ADS is a type of nuclear power plant, which is controlled by the neutrons generated by an accelerator proton beam impacting on a heavy metal target. The core is sub-critical with a kef f of about 0.93-0.97. Possible coolants under consideration include, Pb/Bi eutectic, gas (helium) and sodium. The objective of the ADS is the transmutation of waste from conventional nuclear power plants. Following transmutation the volume of the waste requiring storage could be reduced by a factor of about 100 compared to direct storage [4,5]. The effective radioactive dose can similarly be reduced by a factor of 1000. Section II contains the 80MWt reactor and design basis and severe accident analyses. It includes also calculations of a new separate RVACS and the use of a meltrupture disc to block passively the proton beam and shut down the system. Excerpts of a parametric study are presented in Sect. III. In Sect. IV a comparison between the computational tool, STAR-CD, and experimental results are presented. In Sect. V the effect of different power generation to vessel surface area ratio is evaluated, i.e. the calculations were performed with a greater power level of 250MWt in the same reactor vessel. On the 250MW reactor the safety margins are reduced because of its larger power compared the 80MW reactor. To mitigate this problem complementary systems were investigated to increase the safety margins, e.g. water spray cooling. Finally, the conclusions are presented in Sect. VI.
A. The Ansaldo Design
The integral investigations of this paper are based on the design of an 80MWt ADS demonstration facility by the Italian company Ansaldo [6]. Figure 1 and Tab.I present a schematic view and the principal aspects of this design.
FIG. 1. Schematic view of the Ansaldo design. The numbers in the figure represent: 1. Core 2. Reactor Vessel 3. Rotating Plug 4. Above Core Structure (ACS) 5. Target Unit 6. Sub-Assembly (SA) Transfer Machine 7. Intermediate HX 8. SA Handling Channel 9. SA Basket 10. Cover Gas Cooler
This design is of pool type, with the reactor vessel including the heat exchangers. The vessel is 8m tall and has an inner diameter of 6m. A guard vessel surrounds the reactor vessel. A cyclotron delivers a 3mA/600MeV proton beam current to a Pb/Bi target where spallation reactions occur. The neutrons released in these reactions sustain the fission process in the subcritical core. Table I. Main characteristics of the Ansaldo ADS Demo design [7]
2
Plant Area Power Accelerator Target unit Fuel Coolant and moderator Steel, reactor vessel
Reference Solution 80MWth, kef f 0.97 at BOL kef f 0.94 at EOL Two-stage cyclotron scheme, max 3MW Pb-Bi eutectic, window type undecided U and Pu MOX Pb-Bi eutectic AISI 316L
B. Allowable Temperatures for Different Accident Conditions
The structural material in the reactor vessel is steel AISI 316L. Maximum acceptable temperature limits for different accident conditions are shown in Tab.II. Table II. Critical temperature limitations for structural materials and protective oxide layers. Characteristic problem Corrosion of structural material Secondary creep domain of the reactor vessel Melting point AISI316 ASME3 level C ASME level D
The Pb/Bi coolant flow is mainly driven by natural convection, and the flow rate is increased and made better controllable by the injection of Argon bubbles above the core. This lowers the average density of the coolant in the riser, and thus, increases the static pressure difference between the riser and the downcomer which in turn increases the flow rate. The pressure drop over the core is only 20kPa at a mass flow rate of ∼5300kg/s, corresponding to a coolant velocity of 0.42m/s in the core. The latter is rather small due to the large pitch to diameter ratio of 1.58 [6]. The remaining flow path has a pressure drop of only 9kPa at a flow rate of ∼5800kg/s. The difference in mass flow rate is due to the ∼500kg/s bypass flow through the dummy zone. Since the Demo-ADS and Light Water Reactors (LWR) have similar short lived fission products their decay heat generation curves are very much alike [8]. The following relation, which was developed for a Boiling Water Reactor (BWR), is used to model the decay heat generation in the present analysis, see Eq.1 [9]. Pd (t, T ) = 0.062P0 (t−0.2 − (T + t)−0.2 )
2
1670K [14] 922K [15] 4 977K [15] 5
This scenario investigated a simultaneous Loss-ofHeat-Sink (LOHS) and Loss-of-Flow (LOF) accident in which the beam is switched off instantaneously after accident initiation, i.e. comparable to a classical station blackout or LOP accident. In Fig.2, the reactor operates at normal conditions until the LOHS and LOF accidents, which are assumed to begin after 3000 seconds or almost 1 hour. Within seconds after the accelerator is switched off, the decay heat power generated in the core is reduced to about 6.2% of nominal power. One hour thereafter the decay heat is around 1% [16]. The reactivity feedbacks in this accident scenario has negligible influence on the power generation due to the large sub-criticality [17]. After the gas injection in the riser stops the flow rate declines. Thus, the temperature difference over the core grows and buoyancy forces will build up. Twenty minutes after shutdown, the flow in the riser reaches an equilibrium velocity of 1.6cm/s. Fourteen hours after accident initiation the coolant temperature above the core peaks at a maximum of 615K, which is actually lower than the core outlet temperature at nominal power.
(1)
(2)
Equation 3 [11] which was developed by the Argonne National Laboratory is used to calculate the heat transfer coefficient on the air-side. The choice of heat transfer correlation is described in Chapter IV. N u = 1.22Re0.456 P r0.4
1173K [13]
C. LOHS plus LOF Accident with Immediate Beam Stop
where P0 is the normal operating power, t the time after shutdown, and T the period of time that the core has operated at normal power. The heat transfer mode from the Pb/Bi coolant to the reactor vessel is by convection, whereas the heat transfer between the reactor and guard vessel is by convection and thermal radiation. Also the heat transfer between the guard vessel and the collector wall is by convection and thermal radiation. On the liquid metal coolant side, the heat transfer correlation used is Eq.2 [10]. N u = 0.565Re0.5 P r0.5
Temperature when problem occurs 893K [12] 1
(3)
3
650
Compared to conventional RVACS the Ansaldo design is improved in the sense that in the unlikely case of a rupture of both the guard and reactor vessel radioactive particles still cannot be transported to the atmosphere. The heat removal capability is slightly impaired in comparison to a conventional RVACS. The temperature evolution in the reactor vessel wall for the case of an Ansaldo type RVACS can be seen in Fig.4.
0.5
640 0.4 Wall temperature Core velocity
620
0.3
610 0.2
600
Velocity, m/s
Temperature, K
630
590 0.1 580 570
0
5
10
15
20
25
30
35
40
45
650
0.0
640
Time, hours
Regular RVACS Ansaldo RVACS
630
Temperature, K
FIG. 2. Temperature and velocity evolution at core outlet in an 80-MW(thermal) reactor after an LOF and LOHS accident.
The highest reactor vessel wall temperatures appear above the heat exchanger (HX) where the hot coolant coming from the riser first hits the wall. The coolant then turns downwards and passes the HX. The highest downward velocity below the HX is about 5cm/s at a distance of 1cm from the vessel wall, which is due to the increased density of the well-cooled Pb/Bi near the wall.
620 610 600 590 580 570
0
5
10
15
20
25
30
35
40
45
Time, hours
FIG. 4. Temperature evolution after a LOHS and LOF accident in the reactor vessel wall above the HXs in a 80-MW(thermal) PRISM reactor with Ansaldo-type RVACS.
D. LOHS plus LOF Accident with Immediate Beam Stop using a New Ansaldo Type RVACS
E. LOHS plus LOF Accident with 30 Minutes Delayed Beam Stop
The air circuit of the Ansaldo RVACS design is physically separated from the guard vessel [18], see Fig.3. The cooling air flows in 80 U-pipes of 15.2cm diameter positioned around the perimeter of the guard vessel.
This calculation assumes that the proton beam continues to operate for 30 minutes after accident initiation, and the core power generation is assumed to remain at its normal operating level of 80MW during this period. The target cooling system is supposed to function as normal, i.e. the power deposited in the target is removed. A loss of target circulation during a delayed beam stop scenario would eventually melt the target structures and flood the beam pipe. This would lead to a strong reduction of the ADS power which is discussed in paragraph I. Regarding the delayed beam stop discussed in the current section, it should be pointed out that it is a highly unlikely scenario since in the case of a Loss-of-Power (LOP) accident the accelerator would usually lose its power supply too. During these 30 minutes the average coolant temperature increases rapidly at about 0.25K per second. The wall temperature will increase to 1065K, see Fig.5. Vessel creep will be very limited, but the ASME level D is exceeded for about 4 hours. The 30 minutes delayed beam stop will create a hot cloud in the upper part of the reactor vessel, which impedes coolant flow over the separating cylinder between the riser and the downcomer. Thus, the decay heat removal relies on conduction instead of convection until the hot cloud dissolves about 2 hours after beam stop. Because of the blockage of the normal flow path, eddies develop in the remaining regions.
FIG. 3. Schematic view of the Ansaldo RVACS.
4
garding elevated temperatures of structural materials. This is of course due to the fact that a large amount of energy is transported to the coolant and much less is removed. Compared to the combined LOHS plus LOF accident discussed earlier the reactor vessel wall temperature increases somewhat more slowly and decreases more rapidly because the heat is distributed more evenly in the coolant, i.e. a hot cloud is not established in the upper part of the vessel. In Fig. 7 the temperature evolution is displayed for a LOHS accident where the beam-stop is delayed 40 minutes. The vessel will remain in the secondary creep domain [19].
1300 1200 Temperature wall Temperature core outlet
Temperature, K
1100 1000 900 800 700 600
0
5
10
15
20
25
30
35
40
45
Time, hours
FIG. 5. Temperature evolution after an LOF and LOHS accident when the beam stop is delayed 30 minutes in an 80-MW(thermal) ADS.
0.5
1200 Temperature wall Coolant velocity
1100
0.4
Temperature, K
1000
F. LOF Accident with No Beam Stop
A pure Loss-of-Flow accident was also examined for the Ansaldo reactor. As mentioned earlier, a LOF accident for a design without mechanical pumps, means that the bubble injection above the core is interrupted. Thus the coolant is driven by natural convection only. The calculations showed that the Ansaldo design copes well with a LOF accident. The coolant flow rate remains more than half the steady state velocity due its to good natural circulation capability, e.g. the pressure drop through the core is very low. The temperature at the core outlet will increase only by 80K, which should cause no problems for the structural materials, see Fig.6.
690
0.3
660 0.2
650
0
10
20
30
40
0
10
20
30
40
0.0
Time, hours
FIG. 7. Temperature evolution in the reactor vessel after an LOHS accident with 40-min delayed beam stop.
H. Use of a Passive Melt-Rupture Disc to Interrupt the Proton Beam in the Case of Delayed Beam Stop
A melt rupture disc can be constructed in the side of the vacuum pipe in order to switch off the accelerator proton beam once the coolant temperature exceeds a predesigned temperature limit. Hence, the beam pipe would be flooded with Pb/Bi and the impact point of the proton beam relocated from the core to the upper region of the vessel, only the three MW from the proton beam plus the decay heat generation from the core will be deposited in the coolant instead of 80MW. Thus, the melt rupture disc increases the grace time from 30 minutes to more than 6 hours before a total shut down of the accelerator is required, see Fig.8. It should also be mentioned here that the beam impacting high up in the pipe will melt the beam pipe quickly. This will make it difficult to withdraw this pipe later. Some investigations about improving the cooling of the pipe in this condition might be needed [20]. A LOHS plus LOF accident with a 3MW heat source in the upper part of the primary pool together with core decay heat in the lower section leads to a difficult natural circulation problem. Again, a hot cloud of Pb/Bi coolant is created in the upper region of the vessel, which disturbs the normal coolant flow path until the cloud is dispersed.
0.1
640 630
500
Velocity, m/s
Temperature, K
670
0.2
700 0.1
0.4 Wall temperature Coolant velocity
800
600
0.5
680
0.3
900
0.0
Time, hours
FIG. 6. Temperature evolution in the reactor vessel above the HX and the coolant velocity evolution above the core after an LOF accident with no beam stop.
G. LOHS Accident with 40 Minutes Delayed Beam Stop
Relative to a LOF accident, a Loss-of-Heat-Sink (LOHS) accident causes much more severe problems re5
1300 1200
Temperature, K
1100 Immediate beam-stop No beam-stop Melt rupture disc
1000 900 800 700 600 0
10
20
30
40
Time, hours
FIG. 8. Temperature evolution in the reactor vessel wall after an LOF and LOHS accident with a melt-rupture disk used on the vacuum pipe.
I. Loss of Target Circulation or Beam Misalignment
Most designs rely on forced target circulation in order to remove the heat deposited in the target by the accelerator proton beam. If the circulation is lost due to a seizure of the pump or loss of power (the latter would usually also switch off the accelerator), the window would probably melt first and lead to flooding of the beam pipe. A windowless design would behave similarly. A beam misalignment could lead to a melting of the vacuum beam pipe as well. As mentioned in the previous paragraph a filling of the beam pipe would not switch off the proton beam, but would relocate the impact point to the upper region of the vessel. In both accidents described above, the accelerator has to be switched off in order to avoid melting of the vacuum beam pipe.
FIG. 9. Schematic view of PRISM’s RVACS.
III. PARAMETRIC STUDY OF PRISM TYPE RVACS
Parametrical studies were conducted using STARCD on a simplified RVACS based on the PRISM [21] and SAFR [22] designs, see Fig.9. These examinations were performed in order to evaluate the parameters’ significance with regard to the heat removal rate.
6
wide and a variable depth from 0.5-1 meter. One wall was heated to a temperature of 423K.
The parameters studied were the surface roughness, the fin pitch, the surface emissivity, the surface temperature, and the gap width between the guard vessel and collector wall [23]. All calculations except the surface roughness study used wall functions together with the ANL heat transfer correlation, see Eq.2. The surface temperature and the surface emissivity were found to have the greatest impact on the heat removal rate from the guard vessel [24]. Figure 10 illustrates that the heat removal rate more than triples when the wall temperature rises from 573K to 973K. For the same temperature range, the mass flow increases moderately from 9.5kg/s to 12kg/s. Note the relatively greater contribution from thermal radiation at higher temperatures. This is due to the fact that the heat transfer rate from thermal radiation is dependent on the difference between the wall temperature to the fourth power and the surrounding temperature to the fourth power, i.e. Q ∝ Ta4 − Tb4 .
3.0 With radiation Without radiation
Heat removal, MW
2.5 2.0 1.5 1.0
FIG. 11. Schematic view of the PASCO test channel. 0.5 500
600
700
800
900
1000
The tests of STAR-CD considered both a detailed Two-Layer model with about 15 cells in the boundary layer as well as wall functions applied on a coarser mesh in the cell next to the wall. This means that a logarithmic function is used to model the velocity, and temperature profiles etc in the boundary layer. Different heat transfer correlations were investigated on the air-side in the wall function calculations; the correlation which lead to the most accurate results was developed at Argonne National Laboratory (ANL) for RVACS calculations on PRISM, see Tab.III. Table III. Results from employing different heat transfer correlations on the PASCO channel.
Temperature, K
FIG. 10. Heat removal rate with and without radiation as a function of wall temperature.
Investigations with vertical fins on the guard vessel bridging nearly the entire gap between the guard vessel and the collector wall, show that the optimal heat removal rate occurs at a fin pitch between 5 and 10cm [23]. This extensive use of fins leads to a doubling of the heat removal rate relative to the case without fins. IV. COMPARISON OF STAR-CD CALCULATIONS TO EXPERIMENTAL RESULTS FROM A NATURAL CONVECTION FLOW PROBLEM
Heat Transfer Correlation Experimental results N u = 1.22Re0.456 P r0.4 N u = 0.13(GrP r)1/3 N u = 0.11(GrP r1.69 )1/3 N u = 6.036Gz 0.314 Two-Layer model
The commercial Computational Fluid Dynamics (CFD) code STAR-CD [25] was tested against experimental data for buoyancy driven flows in the PASsive COntainment cooling (PASCO) facility at Forschungszentrum Karlsruhe (FZK), Germany [26]. Figure 11 shows the PASCO facility. The geometrical data of the PASCO channel are 8 meters tall, 1 meter
Heat Removed, W 6500 6600 5400 4400 7300 6700
To determine the fraction of thermal radiation emitted from one cell surface that is received by another cell surface STAR-CD performs a view factor calculation. 7
where ∆Tp = Tp −Tp+1 is the temperature difference between two facing walls or between two surfaces of the same wall. The resistance in the walls is calculated from Eq.6.
The total radiation emitted from a cell surface is the sum of the emitted energy plus the reflected incoming energy, see Eq.4. qr” = αr I − r E
(4) R=
where αr the reflective constant, I the incoming radiation heat transfer, the emissivity, and E the energy emitted from a black body. The total heat flux from the cell surface is thus the sum of the radiation and the convection heat transfer components. For the momentum, energy, and density equations a second order differencing scheme is employed. Turbulence is modelled with a linear k − turbulence equation. The experimental results from the PASCO facility gave a heat flux of 6500W and a mass flow of 0.34kg/s. The STAR-CD calculations using the Two-Layer model predicted a heat removal rate of 6700W and a mass flow of 0.37kg/s. Wall functions employed with the heat transfer correlation according to Eq.2 predicts a heat removal rate of 6600W and a mass flow rate of 0.355kg/s. The error regarding the heat transfer rate for the Two-Layer model and wall function was about 3% and 1.5%, respectively. Thus the heat removal is modelled satisfactorily.
l k
where l is the wall thickness and k is the wall conductivity. The heat transfer between parallel walls is a combination of radiation and convection. Consequently the thermal resistance between two parallel walls is calculated according to Eq.7. R∗ =
{N u kl
+
σ 2 2− (θp
1 2 )(θ + θ + θp+1 p p+1 )}
N u = 0.59(GrP r)1/4
1. Four Times Finer Mesh
To verify the results, calculations on a mesh four times finer than normal were carried out. The temperature evolution diverges to some extent both at the wall and the core outlet. Forty hours after accident initiation the differences are 5K and 8K, respectively. Apparently, the coarse mesh calculation underestimates the heat removal rate compared to the fine mesh by about 4% throughout the calculation. This can be partly explained by the different heat transfer models used next to the reactor vessel wall on the Pb/Bi side. Because the wall function approximation is no longer accurate on the finer mesh, the more detailed Two-Layer model had to be used. The velocity prediction agrees very well in the two calculations.
Temperature coolant, [K] 600 900
p=1
q
∆Tp
Hand calculation, [MW] 0.464 1.781
STAR-CD calculation, [MW] 0.534 1.484
A. LOHS plus LOF Accident with Immediate Beam Stop
Investigations on the heat removal from a higher power ADS reactor were also performed. This is of interest since future ADS plants may have a higher power generation relative to the vessel surface area than the 80MWt Ansaldo design. According to the designers from Ansaldo, the power can be upgraded from 80MW to 250MW in the present vessel [29]. The accelerator proton beam is interrupted immediately after the LOF and LOHS accident initiation. Relative to the 80MW reactor the peak appears 18 hours later and at a 127K higher temperature due to the greater heat generation to surface area ratio. In other words it takes
As a check the heat removal rate was also estimated by a hand calculation. The total thermal resistance, PN p=1 Rp can be defined as the ratio between the overall PN temperature difference p=1 ∆Tp to the heat flux used q, see Eq.5 [27]. p=1
c µ
V. ENHANCED NATURAL AIR CONVECTION COOLING AND OTHER HEAT REMOVAL IMPROVING MEANS FOR A 250MWT ADS REACTOR USING A PRISM TYPE RVACS
2. Hand Calculation
Rp =
(8)
∞ )L where Gr = gβ(Ts −T and P r = pk . ν2 The hand calculation predicts a heat removal rate, which differs by 20% from the STAR-CD results, see Tab.IV. Table IV. Comparison between hand calculation and STAR-CD predictions.
A. Verification of the 80MWt Calculation
PN
(7)
where N u = hL κ , surface emissivity, σ StephanBoltzmann constant, θ temperature in K [28]. For natural convection between vertical plates the Nusselt number is calculated from Eq.8 [28].
3
N X
(6)
(5)
8
longer before the RVACS heat removal rate will exceed the decay heat generation. The average coolant velocities above the core will stabilize at slightly more than 3cm/s. Figure 12 displays the reactor vessel wall temperature evolution.
700
0.5
Temperature wall Coolant velocity
0.3
500
0.2
20
30
0
10
20
30
40
FIG. 13. Temperature evolution in the reactor vessel wall above the HXs in a 250-MW(thermal) reactor. Calculations were performed with and without chimneys.
0.0 10
No chimneys Chimneys with step bend Straight chimenys
Time, hours
0.1
0
600
550
600
500
650
0.4
Velocity, m/s
700
Temperature, K
750
Temperature, K
800
800
40
Time, hours
C. LOHS plus LOF Accident with Immediate Beam Stop and a Higher Emissivity of the Vessel Walls
FIG. 12. Temperature and velocity evolution at core outlet of a 250-MW(thermal) reactor.
The radiation heat transfer between the reactor and guard vessel is significantly improved if the emissivity is increased on their opposing wall surfaces. This is because a large fraction of the heat transfer between the vessels is by radiation. In the 250MWt case, the maximum temperature of the reactor vessel wall is lowered by 34K as the emissivity is increased from 0.7 to 0.9, see Fig.14.
B. LOHS plus LOF Accident with Immediate Beam Stop and 10m Tall Chimneys
A taller RVACS loop increases the static pressure difference between the riser and the downcomer due to the fact that gravity will act on a taller column of air. However, the increased length and the bends also increase the pressure drop, which has a negative effect on the heat removal. Chimneys will improve the heat removal only if the increase in static pressure difference due to their use is greater than the pressure drop. Ten meter tall chimneys are thus only advantageous when the reactor vessel exceeds a temperature of about 700K. For an immediate beam stop in an 80MWt design the chimneys will have a negative effect on the heat removal rate, since the vessel temperature is relatively low. However, for the 250MWt reactor chimneys have a positive effect on the heat removal rate because of the higher coolant temperatures. Straight 10m chimneys lower the maximum reactor vessel wall temperature by 18 degrees, whereas chimneys with step bends reduce the temperature by only 6 degrees, see Fig.13.
800
750
Temperature, K
700
650 Emissivity 0.7 Emissivity 0.9
600
550
500
0
10
20
30
40
Time, hours
FIG. 14. The temperature evolution in the reactor vessel wall above the HXs. Calculations with 0.7 and 0.9 emissivity on the reactor and guard vessel walls.
Equation 9 describes the radiation heat transfer between infinite concentric cylinders [30], which gives an idea how the emissivity influences the radiation heat transfer between the vessels. A change of emissivity from 0.7 to 0.9 at T1 = 600K and T2 = 400K shows an increase of radiation heat transfer by a factor 1.5. q12 =
9
σA1 (T14 − T24 ) 1−ε2 r1 2 1 ε1 + ε2 ( r2 )
(9)
where suffix 1 represent the inner vessel and suffix 2 the outer vessel.
800
750
D. LOHS plus LOF Accident with Immediate Beam Stop using Water Spray Cooling and Filling of the Gap Between the Vessels
Temperature, K
700
To decrease the coolant temperatures rapidly in LOHS and LOF accident events, water spray cooling can be used on the exterior of the guard vessel [31]. Since the process of evaporation of water droplets consumes large amounts of energy, spray cooling can be employed to boost the operation of conventional RVACS. This could be favorably applied in larger power systems to decrease the temperatures of structural materials more rapidly. In this examination, the heat removal from evaporation of water droplets is calculated analytically. The droplet size was chosen as 2mm diameter and the spray flow rate as 100kg/s. The heat transfer efficiency for these parameters is about 0.1 above 500K [32]. At 500K the heat removal from the guard vessel is around 30MW. The heat transfer efficiency is basically defined as the ratio between the fraction of the droplet that is actually evaporated to the total energy needed to heat and evaporate an entire droplet, see Eq.10 [32]. As a conservative approach, the heat transfer efficiency is assumed to decay linearly to zero from 500K to 373K. This is consequently a conservative approach to investigate the potential of water spray cooling. ε=
Q md (hf g + cp ∆Tsub )N
Regular RVACS Spray cooling + filled gap
650
600
550
500
0
5
10
15
20
25
30
35
Time, hours
FIG. 15. The temperature evolution in the reactor vessel wall above the HXs. Water spray cooling was used on the guard vessel surface, and the gap between the vessels was filled with Pb/Bi.
E. LOHS plus LOF Accident with Immediate Beam Stop in a Sodium Cooled Reactor
For reasons of comparison to traditional liquid metal coolants a calculation was performed with sodium (Na) instead of Pb/Bi. The density and specific heat of these two coolants are very different. For Pb/Bi the density, ρ, is 10234kg/m3 and the specific heat, cp , is 0.147kJ/(kg·K) at 644K, whereas for Na these are 860.2kg/m3 and 1.30kJ/(kg·K), respectively, at the same temperature. The product of the density and the specific heat determines the heat capacity. The heat capacity per unit volume turns out to be of the same order of magnitude, hence the temperature evolution after a LOHS+LOF accident is rather similar for these two coolants, see Fig.16.
(10)
where Q is the energy transferred to the spray, md the mass of droplet, hf g the latent heat of vaporization, cp the specific heat, ∆Tsub is subcooling temperature, and N the number of droplets in a square plane. To increase the heat transfer between the reactor and the guard vessel, the gap is filled with Pb/Bi coolant. Hence, good heat conduction between the reactor and guard vessel is accomplished, concurrently with effective heat removal through evaporation of droplets. As can be seen in Fig.15, the decay heat removal exceeds the decay heat generation within a few minutes after the LOHS and LOF initiation. The ASME level C is not exceeded for the reactor vessel wall.
800
Temperature, K
700
600 Sodium - wall PbBi - wall 500
400
0
10
20
30
40
50
Time, hours
FIG. 16. Temperature evolution in the reactor vessel wall above the HXs in 250-MW(thermal) sodium- and Pb/Bi-cooled reactors.
10
VII. FUTURE WORK
F. LOHS plus LOF Accident with Delayed Beam Stop
The investigation of full scale ADSs with larger vessels is important. Of interest is also the investigation of a different type of emergency cooling using an air-cooled water tank around the guard vessel. This is proposed in the Russian SVBR-75/100 design [33]. A confirmation of the natural and forced circulation of Pb/Bi eutectic with STAR-CD code. More detailed thermal hydraulic data from ongoing experiments on Pb/Bi are currently becoming available.
The grace time until accelerator beam stop is needed is naturally shorter for a 250MW reactor compared to an 80MW design. A beam stop delay of more than 6 minutes exceeds the ASME level C, whereas 12 minutes is the limit to avoid fast vessel creep (1173K). Figure 17 shows the temperature evolution in the reactor vessel at delays before beam stop of 6, 12 and 18 minutes.
VIII. ACKNOWLEDGEMENTS
1500 Immediate beam-stop 6min delayed beam-stop 12min delayed beam-stop 18min delayed beam-stop
1400
Temperature, K
1300
The European Commission’s Joint Research Centre funded this project. Prof W Gudowski from KTH Stockholm and K-F Nilsson from JRC Petten who made valuable comments.
1200 1100 1000 900 800 700 600
0
10
20
30
40
Time, hours
FIG. 17. The temperature evolution in the reactor vessel wall above the HXs for different delays of beam shut-off.
VI. CONCLUSIONS
The calculations of worst case cooling accidents in the 80MWt Ansaldo ADS show that an RVACS with aircooling is a very attractive passive approach for lower power systems. For the 80MWt design a 30 minutes delay before beam stop can be tolerated without approaching creep failure. A melt-rupture disc can be used on the vacuum pipe which will relocate the impact point of the proton beam above the core if elevated coolant temperatures are reached. This increases the grace time from 30 minutes to more than 6 hours when the accelerator should be definitely switched off. Other measures for improving the RVACS performance were also evaluated for a 250MWt system in the same vessel. An increase in the emissivity between the guard and the reactor vessel will reduce the maximum reactor vessel wall temperature during the accident by 34K. The use of chimneys with and without bends lowers the maximum vessel temperature by 6 and 24K, respectively. Water spray cooling on the guard vessel outside together with filling of the gap between the reactor and the guard vessel with liquid Pb/Bi was investigated. It leads to rapid cool down of the ADS once the beam is switched off. A general conclusion of this research is that the delay before beam shut-off after the initiation of cooling accident is the most critical parameter. 11
Footnotes 1. Rupture of the protective oxide layer on structural components becomes severe. 2. Fuel failures occur, but the reactor vessel can withstand this temperature for several hours. 3. American Society of Mechanical Engineers 4. Investment might be jeopardised, but fuel damage or radioactive release is unlikely. 5. Endanger the plant from an investment standpoint, although significant fuel failures and radioactive releases are unlikely.
[13]
[14]
[15]
[16]
[17] [1] ENEA, ”A European Roadmap for Developing Accelerator Driven Systems (ADS) for Nuclear Waste Incineration”, ISBN 88-8286-008-6 [2] , Van Tuyle, G.J., Slovik, G.C., Chan, B.C., Kennett, R.J., Cheng, H.S., Kroeger, P.G., ”Summary of Advanced LMR Evaluations - PRISM and SAFR, Brookhaven National Laboratory”, pp 87-91, (Oct. 1989) [3] Baumeister, E.B., et al., ”Inherent Safety Features and Licensing Plan of the SAFR plant”, Proc. Int. Conf. on Fast Breeder Systems, Washington, USA (Sept. 1987) [4] Delpech, M., ”The Am and Cm transmutation - physics and feasibility”, Proc. Int. Conf. on Future Nuclear Systems, GLOBAL99, Jackson Hole, USA (1999) [5] Rubbia, C., Buono, S., Kadi, Y., Rubio, J.A., ”Fast Neutron Incineration in the Energy Amplifier as Alternative to Geologic Storage: The Case of Spain”, CERN/LHC/97-01 (EET) [6] Cinotti, L., Corsini, G., ”A proposal for enhancing the primary coolant circulation in an ADS”, Ansaldo Nucleare, unpublished (1999) [7] Cinotti, L., Corsini, G., ”XADS Pb-Bi Cooled Experimental Accelerator Driven System - Reference Configuration”, Ansaldo Nucleare, unpublished (2001) [8] , Buono, S., Rubbia, C., 1996, ”Simulation of a Total Loss of Power accident in the Energy Amplifier”, CERN/ET Internal note 96-01, (1996) [9] Pershagen, B., Light Water Reactor Safety, Pergamon Press, Oxford, England, pp.50 (1989), ISBN 0-08-0359159 [10] Eckert, E.R.G., Drake, R.M., Heat and Mass Transfer, McGraw-Hill Book Company Inc., New York, USA, pp. 299, (1959) [11] Hunsbedt, A., Magee, P.M., ”Design and performance of the PRISM natural convection decay heat removal system”, Proc. Int. Topical Meeting on Safety of Next Generation Power Reactors, Seattle, Washington, USA, pp 844-851, (1988) [12] Rousanov, A.E., et al., ”Design and study of cladding steels for fuel elements of NPP using heavy coolant”, Proc. Heavy Liquid Metal Coolants in Nuclear
12
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25] [26]
[27]
[28]
[29] [30]
Technology, Obninsk, Russia (1998) Krieg, R., ”Reactor Pressure Vessel under Severe Accident Loading (RPSVA) Final Report of Eu-Project Conctract FI4S-CT95-002”, FZK 6358, (Dec. 1999) Incropera, F.P., DeWitt, D.P., Fundamentals of Heat and Mass Transfer, John Wiley & Sons, New York, USA, pp.492-493 (1996) King, T.L., Landry, R.R., Throm, E.D., Wilson, J.N., ”Preapplication, Safety Evaluation Report for the Sodium Advanced Fast Reactor (SAFR) Liquid-Metal Reactor”, U.S. Nucelar Regulatory Commission, 15-1 (1991) Pershagen, B., Light Water Reactor Safety, Pergamon Press, Oxford, England, pp.50 (1989), ISBN 0-08-0359159 Eriksson, M., and J. Cahalan, ”Inherent shutdown capabilities in accelerator-driven systems”, Annals of Nuclear Energy, vol 29/14 pp 1689-1706, (May 2002) Cinotti, L., Corsini, G., ”A proposal for enhancing the primary coolant circulation in an ADS”, ANSALDO Nucleare, unpublished (1999) Krieg, R., ”Reactor Pressure Vessel under Severe Accident Loading (RPSVA) Final Report of Eu-Project Conctract FI4S-CT95-002”, FZK 6358, (Dec. 1999) Rubbia, C. et al., ”Conceptual Design of a Fast Neutron Operated High Power Energy Amplifier”, CERN/AT/9544(ET), (Sept. 1995) , Van Tuyle, G.J., Slovik, G.C., Chan, B.C., Kennett, R.J., Cheng, H.S., Kroeger, P.G., ”Summary of Advanced LMR Evaluations - PRISM and SAFR, Brookhaven National Laboratory”, pp 87-91, (Oct. 1989) Baumeister, E.B., et al., ”Inherent Safety Features and Licensing Plan of the SAFR plant”, Proc. Int. Conf. on Fast Breeder Systems, Washington, USA (Sept. 1987) Karlsson J., ”Decay Heat Removal by Natural Convection and Thermal Radiation from the Reactor Vessel”, 3rd International Conference on Accelerator Driven Transmutation Technologies and Applications, Prague, Czech Rep. (1999) Hung, T-C., ”The model development of a passive system and its application in residual heat removal”, Proc. of the 1994 Int. Mech. Eng. Congress & Exposition, pp 41-51, Chicago, USA (Nov. 1994) Computational Dynamics Ltd., Methodology Volume 3.10a (2000) Cheng, X., M¨ uller, U., ”Turbulent natural convection coupled with thermal radiation in large vertical channels with asymmetric heating”, Int. J. Heat Mass Transfer 41,pp 1681-1692 (1998) Incropera, F.P., DeWitt, D.P., Fundamentals of Heat and Mass Transfer, John Wiley & Sons, New York, USA, pp.739 (1996) Incropera, F.P., DeWitt, D.P., Fundamentals of Heat and Mass Transfer, John Wiley & Sons, New York, USA, pp.76-86 (1996) Cinotti, L., personal communication Incropera, F.P., DeWitt, D.P., Fundamentals of Heat and Mass Transfer, John Wiley &
Sons, New York, USA, pp.76-86 (1996) [31] Carluec B., Framatome, personal communication to H.Wider (1999) [32] Delcorio, B., Choi, K-J., ”Analysis of Direct Liquid-Solid Contact Heat Transfer in Monodispersed Spray Cooling”, J. of Thermophysics, (1991) [33] Toshinsky, G.I., Grigoriev, O.G., Efimov, E.I., Leonchuk, M.P., Novikova, N.N., Pankratov, D.V., Skorikov, D.E., ”Safety Aspects of SVBR-75/100 Reactor”, OECD Workshop on Advanced Nuclear Reactor Safety Issues and Research Needs, Paris, (Feb 2002)
13
2
Comparison of Safety Performance of Pb/Bi-cooled Accelerator-Driven Systems for different heat-exchanger locations and power levels Johan Carlsson
∗
Royal Institute of Technology, Department of Nuclear and Reactor Physics, S-10691 Stockholm, Sweden
Hartmut Wider Joint Research Center, Institute for Energy, NL-1755 LE Petten, Netherlands (21 March 2003)
Send invoice to: Prof. Waclaw Gudowski Department of Nuclear and Reactor Physics Royal Institute of Technology Stockholm Center for Physics, Astronomy and Biotechnology Roslagstullsbacken 21 S-106 91 Stockholm Sweden Tel. +46 8 5537 82 00 Fax. +46 8 5537 84 65 E-mail.
[email protected]
Total number of pages in the preprint edition are 42.
∗ Email:
[email protected]
1
Contents
I
Introduction
6
II
Methodology
7
III
Comparison of temperature evolutions during LOF, LOHS, and TLOP accidents for ADS-HXR200, ADS-HXR200 without BPRs, and ADSHXD200 designs of 200 MW(thermal) power
9
A
Main characteristics of ADS-HXR200 and ADS-HXD200 . . . . . . . . . .
9
1 Specifics for an ADS with heat-exchangers located in the risers . . . .
12
B
Total-loss-of-power accident . . . . . . . . . . . . . . . . . . . . . . . . . .
14
C
Loss-of-flow accident . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
D
Loss-of-heat-sink accident . . . . . . . . . . . . . . . . . . . . . . . . . . .
16
IV
Verification of the 200 MW(thermal) calculation
V
Comparison of temperature evolutions during LOF, LOHS, and
17
TLOP accidents for ADS-HXR800, and ADS-HXD800 designs of 800 MW(thermal) powers
18
A
Main characteristics of ADS-HXR800 and ADS-HXD800 . . . . . . . . . .
18
B
Total-loss-of-power accident . . . . . . . . . . . . . . . . . . . . . . . . . .
19
C
Loss-of-flow accident . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19
D
Loss-of-heat-sink accident . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
VI
Conclusions
20
VII
Future Work
22
VIII Acknowledgements
22
2
List of Figures
1
Schematic view of the ADS-HXD200. . . . . . . . . . . . . . . . . . . . . . .
31
2
Schematic view of the RVACS of ADS-HXR200 and ADS-HXD200. . . . . .
32
3
Schematic view of the ADS-HXR200. . . . . . . . . . . . . . . . . . . . . . .
33
4
Schematic view of displacement of thermal centers after an LOF accident for ADS-HXR200 and ADS-HXD200. . . . . . . . . . . . . . . . . . . . . . . . .
34
5
Coolant flow field of ADS-HXR200 one hour after TLOP accident initiation.
35
6
Temperature evolution at core outlet during a TLOP accident for ADSHXR200, ADS-HXR200 without BPRs, and ADS-HXD200. . . . . . . . . . .
7
Temperature evolution at core outlet during an LOF accident for ADSHXR200, ADS-HXR200 without BPRs, and ADS-HXD200. . . . . . . . . . .
8
40
Temperature evolution at core outlet during an LOF accident for ADSHXR800 and ADS-HXD800. . . . . . . . . . . . . . . . . . . . . . . . . . . .
12
39
Temperature evolution at core outlet during a TLOP accident for ADSHXR800 and ADS-HXD800. . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
38
Temperature evolution at core outlet during an LOHS accident for ADSHXR200, ADS-HXR200 without BPRs, and ADS-HXD200. . . . . . . . . . .
10
37
Temperature evolution in the reactor vessel wall during an LOF accident for ADS-HXR200, ADS-HXR200 without BPRs, and ADS-HXD200. . . . . . . .
9
36
41
Temperature evolution at core outlet during an LOHS accident for ADSHXR800 and ADS-HXD800. . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
42
List of Tables
I
Critical temperatures for structural materials and protective oxide layers
II
Main characteristics of the ADS-HXR200
III
Comparison between hand calculation and STAR-CD predictions
4
Abstract - A safety investigation on the location of heat-exchangers in the risers or the downcomers of a Pb/Bi-cooled accelerator-driven system of 200 and 800 MW(thermal) powers was performed. In a pool type design with a simple flow path the use of heat-exchangers in the risers will have advantages in case of heat-exchanger tube failures. This is particularly true if water is used as the secondary fluid, since it can be avoided that steam bubbles are dragged into the core region by the Pb/Bi-coolant. The safety implications with regard to the temperature evolution during loss-of-flow, loss-of-heat-sink, and total-loss-of-power accidents were compared with designs where the heat-exchangers are located in the downcomers. During a loss-of-flow accident for the 200 MW(thermal) system with heat-exchangers in the risers, the core outlet temperature increases to about 1020 K. For the accelerator-driven system with heat-exchangers in the downcomers the temperature maximum is nearly 150 K lower. After a loss-of-heat-sink accident the grace period before the proton beam has to be shut off was found to be ∼5 minutes for both the 200 MW(thermal) designs. During a total-loss-of-power accident for the case with heat-exchangers positioned in the riser, the core outlet temperature peaks after 10 hours at about 750 K, which is 100 K lower than in the case with heat-exchangers in the downcomer. The investigations on the 800 MW(thermal) system were performed in a taller vessel of 17 m height. The design with the heat-exchangers in the riser showed elevated temperatures of more than 2000 K during a lossof-flow accident, which would severely damage the core. The case with heat-exchangers in the downcomers showed acceptable results for all accident types investigated. Therefore, an 800 MW(thermal) accelerator-driven system of pool design with simplified flow path must have the heat-exchangers in the downcomer. The computational fluid dynamics code STAR-CD was used in all calculations.
5
I. INTRODUCTION
This paper investigates the effects from locating the heat-exchangers (HX) in the risers of a Pb/Bi-cooled accelerator-driven system (ADS) of 200 and 800 MW(thermal) powers, which is compared with a nearly identical design that has the heat-exchangers in the downcomers, however. A pool type ADS with a simplified flow path and gas lift pumps as in the Ansaldo design [1] is considered. The incentive behind locating the HXs in the risers is to increase the inherent safety of the reactor. This would be particularly relevant if water was used as the secondary coolant. If the HXs are located in the downcomer, a HX-tube rupture may lead to steam bubbles being dragged into the core region, which would result in a positive reactivity insertion. If instead the HXs are located in the risers the steam bubbles would rise with the Pb/Bi coolant and exit at the free surface. The consequences of a major water leakage is discussed later in the paper. Also for silicon oil as secondary coolant it could be of a certain advantage to locate the HXs in the risers. Although the oil pressure is lower than the Pb/Bi pressure, a back-pressure may develop in the secondary circuit that could lead to the insertion of oil droplets into the primary coolant. Enhancement of the inherent and passive safety is in accordance with the safety trends for the next generation of nuclear power plants, as proposed by the American Generation IV initiative [2]. Implementation of inherent safety for a reactor design can exclude accident scenarios or at least mitigate accidents. Examples are negative reactivity coefficients or a low pressure coolant. Passive safety systems are very reliable since they depend on physical laws, and do not need to be activated by mechanical or electrical devices [3]. An example is a reactor vessel auxiliary cooling system (RVACS). Inherent and passive safety measures also contribute to simpler reactor designs in which less components and less active safety systems are needed. The capital investment can be lower for reactor designs which includes inherent and passive safety systems [4]. An ADS [5] is a sub-critical nuclear reactor, which is intended for transmuting radioactive waste from conventional nuclear power plants. ADSs may be cooled by gas or liquid 6
metals in order to achieve a hard neutron spectrum. High energy protons from an accelerator are directed onto a heavy metal target in which spallation reactions occur that release high-energy neutrons. The neutrons then diffuse into the sub-critical core. The hard neutron spectrum and the sub-critical core make the ADS suitable for transmuting the transuranics (TRU: Pu, Np, Cm and Am). The water soluble and radiotoxic long-lived fission products (LLFP:
99
Tc,
129
I) will also be transmuted to more stable isotopes. After additional par-
titioning of uranium and the short-lived isotopes like Cs and Sr, the volume of high-level nuclear waste can be reduced by a factor of ∼100, and the effective radioactive dose from a repository by a factor of ∼1000 [5–7]. The reduction of the Pu inventory, decreases the risk for spreading of nuclear weapons material. Section II describes the methodology used for this article. In Section III the conceptual accelerator-driven system of 200 MW(thermal) with the heat-exchangers in the risers (ADSHXR200) and the downcomers (ADS-HXD200) are described. A comparison between the two reactors is then performed for loss-of-flow, loss-of-heat-sink, and total-loss-of-power accidents. The calculations are verified with hand calculations in Sect IV. And finally, corresponding investigations is done in Sect V on an ADS design of 800 MW(thermal).
II. METHODOLOGY
The computational fluid dynamics (CFD) code STAR-CD was used for all calculations presented in this paper [8]. A first-order upwind scheme has been used to model the fluids’ velocity, density, and viscosity, whereas the temperature is solved by a first-order centerdifference scheme. Time stepping was performed with a first-order implicit scheme. The calculations incorporating the bypass routes (BPR) were three dimensional, whereas the remaining examinations were two dimensional. The radiation transport model between the reactor, guard, and collector walls is based on view factors for each individual cell and is also calculated by STAR-CD. The heat transfer correlation used on the air side was developed at the Argonne Na7
tional Laboratory (ANL) within the American Liquid Metal Reactors’ (ALMR) program for RVACS calculations [9], see Eq 1. N u = 1.22 · Re0.456 P r0.4
(1)
The Nusselt number is defined as N u = hL/k where h is the heat transfer coefficient, L the characteristic length, and k the conductivity. The Reynold number is Re = V L/v where V is the velocity in the bulk, and v the dynamic viscosity. The Prandtl number is P r = v/α where α is the thermal diffusivity. A correlation for liquid metals is used to calculate the heat transfer coefficients next to the walls in the Pb/Bi coolant [10], see Eq 2. N u = 0.565Re0.5 P r0.5
(2)
The core and HXs are modelled as porous regions, in which the pressure drop is proportional to the coolant velocity squared. The core properties of the Ansaldo design are used for the 200 MW(thermal) calculations. Hence the total pressure resistance of the core is 20 kPa at a coolant velocity of 0.42 m/s in a channel of the core [1]. The core power profile is modelled as a core divided into three regions of different power density in accordance to the Ansaldo design. The power level is reduced to decay power levels in case of the TLOP accident. In the LOF and LOHS calculations the power was left constant, since it is known that reactivity feedbacks do not change the power strongly [11]. For the 800 MW(thermal) calculations the Sing-Sing Core developed at the Royal Institute of Technology, Stockholm, Sweden is used. Since the surrounding core structures of the SSC concept remain to be designed, only the pressure drop of a core coolant channel can be calculated analytically, see Eq 3-5. To estimate the total pressure drop, data extrapolated from an existing design is used, i.e. the Pb-cooled critical reactor BREST [12]. ∆p = M · fsmooth ·
L ρ·V2 · Dei 2
(3)
where L is length of channel, f is friction factor, V is velocity, and Dei is hydraulic diameter, which is defined as 8
Dei =
4A Pwi
(4)
where Pwi is the wetted perimeter, and A the channel cross sectional area. The pressure losses due to entrance and exit effects are estimated by Eq 5. ∆pe = where K is calculated as Ke1 = 0.5(1 −
KρV 2 2 Ai ) A
at the entrance and as Ke2 = (1 −
(5) Ai 2 ) A
at
the exit. The additional pressure drop from supporting structures was determined by calculating the ratio between pressure drops from experiments, and the analytical pressure drop from channel plus entrance and exit effects for the BREST reactor. The ratio was then used to estimate the total pressure drop for the core and its structures of ADS-HXR800 and ADS-HXD800. The pressure ratio for BREST was estimated to be 1.51.
III. COMPARISON OF TEMPERATURE EVOLUTIONS DURING LOF, LOHS, AND TLOP ACCIDENTS FOR ADS-HXR200, ADS-HXR200 WITHOUT BPRS, AND ADS-HXD200 DESIGNS OF 200 MW(THERMAL) POWER
Comparisons of the temperature and velocity evolution for ADS-HXR200, ADS-HXR200 without bypass routes (BPR), and ADS-HXD200 were performed for LOF, LOHS, and TLOP accidents. In all investigations the reactor vessels sizes and the core designs were the same for ADS-HXR200 and ADS-HXD200. Moreover, cross-sections, heights of the HXs, and their pressure resistance at a given velocity were identical in all analyses.
A. Main characteristics of ADS-HXR200 and ADS-HXD200
The pool-type reactor vessel is 9 m tall and has a diameter of 6 m. A guard vessel encloses the reactor vessel in order to reduce the probability for a loss-of-coolant accident 9
(LOCA) and a release of radioactivity. Both vessels, as well as the walls that separate the riser and the downcomer, are made of steel of type AISI 316L [13]. Since Pb/Bi is a liquid metal, high pressures are not needed to remove heat efficiently. Thus, the vessel can be thinner than that of an LWR. Figure 1 illustrates a schematic view of ADS-HXD200. Figure 1. Schematic view of the ADS-HXD200. Pb/Bi eutectic is the coolant and moderator of the primary circuit. The heavy metal coolant provides good shielding and low neutron leakage, which improves the neutron economy and diminishes radiation damages on the surrounding structures. The Pb/Bi provides a hard neutron spectrum because only a small fraction of the neutron energy is lost in each collision with the heavy Pb/Bi atoms [14], and thus minor actinides (MA) will be efficiently transmuted. The Pb/Bi coolant has high conductivity, good natural convection capability, and a high heat capacity. Moreover, the Pb/Bi does not interact violently with water or air, although large amounts of steam can be produced in case of water ingress. A negative characteristic of the Pb/Bi coolant is its production of polonium (Po) due to neutron absorption in Bi [15], which becomes an α-emitter of 138.4 days half life. Furthermore, the Pb/Bi is relatively corrosive towards steel [16]. However, the corrosion can be reduced by orders of magnitude by maintaining a specific oxygen level in the coolant [17]. This creates a self-sustaining protective oxide layer on the surfaces of structural materials, which separates the metal from the Pb/Bi coolant. At too low oxygen concentration no effective oxide films of Fe3 O4 can be created [18,19], thus the Fe dissolves more rapidly. Eq 6 displays the chemical reaction of the oxide film [16]. F e3 O4 + 4P b * ) 3F e + 4P bO
(6)
At too high oxygen concentrations PbO precipitates in the molten Pb/Bi and degrades the thermal hydraulic performance [16]. In order to maintain the oxide layers the coolant velocities and temperatures should not exceed 3 m/s and 873 K [20,21] for long term operation. The temperature limitations for structural materials and oxide layers are presented in Tab I. 10
Table I. Critical temperature limitations for structural materials and protective oxide layers. Argon gas injection [1] will probably be used for the 200 MW(thermal) design instead of conventional pumps to enhance the flow rate during normal operation. The volume percentage of gas at the point of injection is about 5%. As the argon gas rises it will expand as the hydrostatic pressure decreases. Near the free surface the percentage of argon gas is 19%. Thus, from the gas injection point upwards the coolant density decreases, and thus, the static pressure difference increases between the riser and downcomer. The gas then exits through the Pb/Bi free surface to a gas plenum. Here it is sucked up by a compressor and pumped back to the insertion point above the core [1]. For the 800 MW(thermal) design, gas injection is probably not possible since a rather high gas content is necessary to reach sufficient coolant flow rates during normal operation. This would probably raise the free surface too much. However, one could possible reduce the flow rate for a design with good natural convection. The accelerator power needed to drive the 200 MW(thermal) core is ∼7.5 MW at end of life (EOL). Since the spallation reactions in the target are endothermic only ∼75% has to be removed, i.e. ∼5.6 MW at EOL [22]. A separate cooling circuit accomplishes the heat removal from the target. In case of an emergency the ion source of the accelerator can be switched off quickly or the proton beam can be redirected into a beam dump. However, it is conceivable that the active interruption malfunctions or that the accident initiation is overlooked. For such conditions passive shut-off mechanisms are needed to ensure interruption of the spallation process in the ADS, e.g. a melt-rupture disc inserted into the wall of the vacuum beam pipe [11] can be used. The melt-rupture disc melts as the designed temperature limit is exceeded and the accelerator vacuum pipe is flooded by Pb/Bi coolant. Thus, the impact point of the accelerator beam is relocated from the core region to the upper section of the vessel. Instead of the normal operating power of 200 MW(thermal), the heat generation decreases to decay heat level and 5.6 MW is deposited by the accelerator proton beam. The former is initially 6.2% (12.4 MW) and after one hour 1% (2 MW) [23]. 11
In case that the normal heat removal and safety systems fail, the ultimate emergency decay heat removal system is a reactor vessel auxiliary cooling system (RVACS). The RVACS is based on the principle of natural air convection and thermal radiation from the reactor vessel, see Fig 2. Figure 2. Schematic view of the RVACS of ADS-HXR200 and ADS-HXD200.
1. Specifics for an ADS with heat-exchangers located in the risers
Figure 3 shows a schematic drawing of the 200 MW(thermal) accelerator-driven system with heat-exchangers in the risers (ADS-HXR200). The ADS-HXR200 main design parameters can be found in Tab II. Figure 3. Schematic view of the ADS-HXR200. Table II. Main characteristics of the ADS-HXR200. As mentioned above argon gas injection is used to enhance the Pb/Bi coolant flow rate. It provides robust operation and allows easy maintenance due to the fact that the gas compressors are located outside the reactor vessel. However an uncertainty in this design is how the argon gas effects the heat removal in the HXs for the ADS-HXR200. Therefore this paper uses a conservative volume to power ratio of the HXs, i.e. three times larger than that of the the Pb-cooled BREST design. For the case when water is used as coolant of the secondary circuit, a rupture may lead to steam bubbles that are dragged by the Pb/Bi coolant through the core and cause a power increase by voiding. For PWRs the probability for steam leakage from the secondary circuit is 1.1x10−2 /reactor year (RY) and the probability of a major steam leakage is estimated to 8.4x10−4 /RY [24]. For Pb/Bi-cooled designs the probability for a steam leakage is unknown due to limited civilian experience with this coolant. The problems in the above mentioned accident scenario could be eliminated by placing the HXs in the risers instead of the downcomers. Thus, a leakage from the HXs leads to an upward sweepout of the steam bubbles along the Pb/Bi flow direction. In case of a massive 12
rupture, passively closing flaps could be included at the bottom of the HXs in order to avoid that the steam can reach the core. However, these flaps should not close for limited negative flow rates that can occur in cooling accident conditions. A pressure release pipe could be used to deal with overpressure in the primary circuit. This penetrates the vessel head and includes a diaphragm that ruptures as the plenum pressure increases. The steam/gas mixture is then led into a large water condenser. This approach of removing steam from a gas plenum is proposed for the SVBR75/100 Pb/Bicooled reactor [25]. The natural convection was found to be impeded somewhat when the HXs are positioned in the risers. This is particularly evident during an LOF accident, since the distance between the thermal centers of the core and HXs becomes about one meter shorter for the HXs located in the risers. On the contrary the distance between the thermal centers becomes about one meter longer during an LOF accident if the HXs are located in the downcomers. Consequently the static pressure difference between the risers and downcomers decreases the natural convection capability of the ADS-HXR200, whereas the natural circulation increases for the ADS-HXD200. Figure 4 shows the movement of the thermal centers during an LOF accident for ADS-HXR200 and ADS-HXD200. Figure 4. Schematic view of displacement of thermal centers during an LOF accident for ADS-HXR200 and ADS-HXD200. There are low pressure resistance bypass routes (BPR) constructed in parallel to the HXs in order to increase the coolant flow rate during LOF accidents, see Fig 2. The BPRs cover the entire circumference to the reactor vessel walls and occupy a 10 cm wide annular passage. Argon gas is also injected into the BPR region in order to avoid backward flow during normal operation. The injection rate is lower than that into the HX region, because ideally all coolant should pass through the HXs and the flow in the BPR should be stagnant. However, this ideal solution is impossible to achieve in reality. Hence, the intention is use a low upward velocity through the BPRs during normal operation. 13
A drawback of the BPRs is that during an LOF accident the coolant temperature will increase from 800 K to 950 K in 7 minutes, which can cause thermal stresses in the reactor vessel wall. For a conventional design with HXs in the downcomers an LOHS accident would, however, result in an even faster temperature transient in the reactor vessel wall below the HX [26]. Since the hot coolant is always next to the reactor vessel wall for the ADS-HXR200, a larger thermal parasitic loss appears during normal operation compared to the ADSHXD200, ∼1.3 MW vs 0.6 MW. On the other hand, the RVACS heat removal during accidents is more efficient for the ADS-HXR200. The parasitic loss during normal operation can be reduced for both ADS-HXR200 and ADS-HXD200 by using a liner next to the reactor vessel wall. Coolant overflow of the liner would then take place after a certain thermal expansion of the coolant. Before liner overflow, the heat transfer would be first by conduction and afterwards by convection.
B. Total-loss-of-power accident
A total-loss-of-power (TLOP) accident could be initiated, for example, by a station blackout. In such an event the normal heat removal systems malfunction, the argon gas injection stops, and the accelerator proton beam is interrupted. Nevertheless, the core will generate decay heat of ∼6.2% normal power right after accident initiation and ∼1% after one hour [23]. The reactivity feedbacks have negligible influence on the power generation [11,27]. Instead the proton beam decides the power level of the core. To ensure the integrity of the reactor components, the decay heat removal has to be assured. The flow field of the Pb/Bi-coolant during normal operation for ADS-HXR200 is shown in Fig 3. After ∼1 hour the flow reverses and moves backwards through the core. This is due to the large temperature differences between the downcomers from the HXs and BPRs, which leads to a high heat transfer rate through the separating walls. After accident initiation the temperature difference is 175 K and the heat transfer rate through the walls 14
is about 2.8 MW. The coolant in the downcomers from the BPRs is cooled both through the reactor vessel wall and the separating walls of the HXs and BPRs. Correspondingly, the downcomers of the HXs are heated through the separating walls which creates buoyancy forces in this region. The flow field one hour after accident initiation for ADS-HXR200 is shown in Fig 5. The flow directions in the ADS-HXD200 design during a TLOP accident remain the same as during normal operation. Figure 5. Coolant flow field the ADS-HXR200 one hour after TLOP accident initiation. A comparison of the temperature evolution reveals that the core outlet temperature is 74 K higher for ADS-HXR200 than for ADS-HXD200 after 10 minutes. This is due to the better natural convection of the ADS-HXD200 until one hour after accident initiation. Thereafter the reversing coolant flow rate through the core is higher for the ADS-HXR200 design. The BPRs of ADS-HXR200 cause a decrease of the core outlet temperature of ∼90 K after one hour. The temperature evolution at core outlet during the TLOP accident is displayed in Fig. 6. Figure 6. Temperature evolution at core outlet during an TLOP accident for ADSHXR200, ADS-HXR200 without BPRs, and ADS-HXD200.
C. Loss-of-flow accident
During loss-of-flow (LOF) accidents the argon gas injection malfunctions (or the mechanical pumps stops), hence the enhanced circulation gradually diminishes until the argon gas has disappeared through the Pb/Bi free surface. About twenty seconds after accident initiation the effect on coolant density from the gas injection has disappeared and the flow is only driven by natural convection. The HXs continue to remove heat. A spike of 1080 K appears during an LOF accident for ADS-HXR200, whereas ADSHXD200 peaks at 930 K. The ADS-HXR200 without BPRs peaks at 1140 K, which is 60 K higher than for ADS-HXR200. The higher peak for ADS-HXR200 compared to ADSHXD200 is due to faster coolant deceleration, see Section ’Specifics for an ADS with heat15
exchangers located in the risers’ for more information. After ∼5 minutes the temperature at core outlet is ∼145 K higher for ADS-HXR200 than ADS-HXD200. Figure 7 displays the temperature evolution for ADS-HXR200, ADS-HXR200 without BPRs, and ADS-HXD200. Figure 7.
Temperature evolution at core outlet during an LOF accident for ADS-
HXR200, ADS-HXR200 without BPRs, and ADS-HXD200. The temperature evolution in the reactor vessel wall behaves differently. The increase of the vessel wall temperature is most rapid for ADS-HXR200 due to that hot coolant hits the wall directly via the BPRs after an LOF accident. Regarding the vessel temperatures, the ADS-HXR200 without BPRs performs much better than ADS-HXR200 since the Pb/Bi is cooled by the HXs before it reaches the reactor vessel wall. The temperature evolution in the reactor vessel wall for ADS-HXR200, ADS-HXR200 without BPRs, and ADS-HXD200 is displayed in Fig 8. Figure 8. Temperature in the reactor vessel wall during an LOF accident for ADSHXR200, ADS-HXR200 without BPRs, and ADS-HXD200.
D. Loss-of-heat-sink accident
After loss-of-heat-sink (LOHS) accidents the normal heat removal systems stops whereas the power generation of the core remains at 200 MW(thermal). The heat removal rate through the vessel wall is small compared to the core power. Due to the enhanced convection the temperature transient is less rapid than during an LOF accident. However, the coolant temperature increases steadily by ∼0.6 K/sec. Within ∼5 minutes the proton beam should be interrupted for both ADS-HXR200 and ADS-HXD200 in order to avoid exceeding the ASME level D of 977 K. Fast vessel creep in the vessel would only become a problem after more than 10 minutes. The increase of velocity immediately after accident initiation for ADS-HXR200 is due to a temporarily taller column of hot coolant after an LOHS since heat is not removed from the HXs anylonger. Thus the distance between the thermal centers increases and the static 16
pressure difference between the riser and downcomer increases until the hot coolant reaches the downcomer. The temperature rise of ADS-HXR200 is about 30 K above that of ADS-HXD200. After five additional minutes the temperature difference has increased to 60 K. The BPRs lessen the core outlet temperature by about 50 K. Figure 9 depicts the temperature evolution at core outlet for ADS-HXR200, ADS-HXR200 without BPRs, and ADS-HXD200. Figure 9. Temperature at core outlet during an LOHS for ADS-HXR200, ADS-HXR200 without BPRs, and ADS-HXD200.
IV. VERIFICATION OF THE 200 MW(THERMAL) CALCULATION
Hand calculations on a simplified geometry were performed to validate the heat removal by the RVACS from the ADS-HXR200 vessel, see below. Additionally, the STAR-CD code has earlier proved to predict natural air convection flows accurately [28]. The total thermal resistance,
PN
p=1
Rp , for heat flowing from the Pb/Bi coolant to the
inflowing air of the RVACS can be defined as the ratio between the overall temperature difference
PN
p=1
∆Tp to the heat flux used q, see Eq 7 [13]. N X
Rp =
PN
p=1
∆Tp
q
p=1
(7)
where ∆Tp = Tp −Tp+1 is the temperature difference between two facing walls or between two surfaces of the same wall. The resistance in the walls is calculated from Eq.8. R=
l k
(8)
where l is the wall thickness and k is the wall conductivity. Between the reactor and guard vessel walls the heat transfer is by to thermal radiation and convection. Hence, the thermal resistance between two parallel walls is calculated as Eq.9. R∗ =
{N u kl
+
σ (θ2 2− p
1 2 + θp+1 )(θp + θp+1 )}
17
(9)
where N u =
hL , κ
surface emissivity, σ Stephan-Boltzmann constant, θ temperature in
K [13]. For natural convection between vertical plates the Nusselt number is calculated from Eq.10 [13]. N u = 0.59(GrP r)1/4 where Gr =
gβ(Ts −T∞ )L3 ν2
and P r =
(10)
cp µ . k
A comparison between the STAR-CD results and the hand calculations is presented in Tab III. They differ less than 10% for both temperature levels. Table III. Hand calculation to verify the STAR-CD predictions.
V. COMPARISON OF TEMPERATURE EVOLUTIONS DURING LOF, LOHS, AND TLOP ACCIDENTS FOR ADS-HXR800, AND ADS-HXD800 DESIGNS OF 800 MW(THERMAL) POWERS
An examination on temperature maximums reached during LOF, LOHS and TLOP accidents for 800 MW(thermal) power ADS with heat-exchangers located either in the risers or the downcomers.
A. Main characteristics of ADS-HXR800 and ADS-HXD800
In order to improve natural circulation and increase the heat capacity the vessel height was increased to 17 m. The vessel diameter remained at 6 m as for the 200 MW(thermal) designs. The 800 MW(thermal) ’Sing-Sing Core (SSC)’, which is developed at the Royal Institute of Technology, Sweden, is used for these examinations [29]. The pressure drops used are described in the ’Methodology’ section.
18
B. Total-loss-of-power accident
For the 800 MW(thermal) design the core produces decay heat of 49.6 MW (6.2%) ten seconds after accident initiation and ∼8 MW (1%) after one hour. During TLOP accidents the core outlet temperature peaks after 19 hours at 848 K for ADS-HXD800, whereas for ADS-HXR800 it peaks at 769 K after ∼17 hours. Figure 10 displays the temperature evolution during a TLOP accident. Figure 10. Temperature evolution at core outlet during a TLOP accident for ADSHXR800 and ADS-HXD800.
C. Loss-of-flow accident
For ADS-HXR800 the core outlet temperature peaks at 2000 K and then slowly increases. It can thus be concluded that LOF accidents reach unacceptable temperatures due to insufficient natural convection for cooling the core. The ADS-HXD800 peaks only at 1100 K and 2 minutes later the temperature levels off at ∼1000 K. The ASME level D of 977 K is thus slightly exceeded. The reason for the much better performance of ADS-HXD800 compared to ADS-HXD800 is due to better natural circulation. The reason for this is described in Section ’Specifics for an ADS with heatexchangers located in the risers’. Figure 11 illustrates the temperature evolution during an LOF accident. Figure 11. Temperature evolution at core outlet during an LOF accident for ADSHXR800 and ADS-HXD800. As mentioned earlier the power removal versus the volume ratio of the HXs is conservatively chosen, since it is uncertain how the passage of argon gas through the HXs affects their heat removal. When the HXs are located in the downcomers this conservative measure is unnecessary since the argon gas will not pass through there. Thus, an investigation was performed where the ratio was the same as for the BREST reactor, i.e. three times shorter
19
than for the previous investigation. For this case the temperature stabilizes at 992 K during a LOF accident in the modified ADS-HXD800.
D. Loss-of-heat-sink accident
The average temperature increase of the coolant is 1.1 K/sec after the heat sink is lost. The grace period before beam stop is necessary is ∼3 minutes in order to avoid exceeding the ASME level D for both the ADS-HXR800 and ADS-HXD800, see Figure 12. Figure 12. Temperature evolution at core outlet during an LOHS accident for ADSHXR800 and ADS-HXD800.
VI. CONCLUSIONS
To locate the heat-exchangers in the risers is a potentially inherent safety enhancement for smaller type reactors with a simplified flow path. This is particularly important for the case when water is used as coolant in the secondary circuit. This design inherently avoids that a leakage of steam from the heat-exchangers sweeps steam bubbles into the core region, as could be the case if the with heat-exchangers in the downcomers. If the secondary coolant is oil, the static pressure of the Pb/Bi-coolant is higher than the pressure of the oil and therefore the Pb/Bi would primarily go into the oil tubes. A back pressure that could develop may lead to some oil in the Pb/Bi flow. Whether the leaking oil bubbles would rise or go with the Pb/Bi flow has to be proven experimentally. For a 200 MW(thermal) reactor with heat-exchangers located in the riser there is a satisfactory behavior for loss-of-flow, loss-of-heat-sink, and total-loss-of-power accidents. For loss-of-flow accidents with the accelerator still operating, the core outlet temperatures stabilizes at 1020 K, which is slightly above the ASME level of 977 K. For the 200 MW(thermal) design with heat-exchangers in the downcomers, the natural convection is considerably better during an loss-of-flow accident and the temperature stabilizes at 882 K.
20
After a loss-of-heat-sink accident the grace period is about 5 minutes before beam interruption is necessary to avoid surpassing the ASME level D of 977 K for both reactor designs. A total-loss-of-power accident leads to a temperature peak of 664 K nearly 10 hours after accident initiation for a design with the heat-exchangers in the risers. This causes no problems for the core nor the structural materials. The temperature peak during a total-loss-of-power accident, with the heat-exchangers located in the risers, is about 74 K higher. For an 800 MW design the impediment of natural convection during a loss-of-flow accident with the beam on is too severe for an ADS design with heat-exchangers in the riser. A temperature of 2000 K will be exceeded within 30 seconds. Hence, this type of approach is not possible to employ for an ADS of 800 MW(thermal) power. The ADS with heatexchangers in the downcomer stabilizes at 1020 K for the same accident event. However, to locate the heat-exchangers in the risers might be applicable for a critical Pb or Pb/Bi-cooled reactor since negative reactivity feedbacks would shut the reactor off during the accidents considered. During a total-loss-of-power accident the temperature evolution at core outlet is benign for both reactor types. The design with heat-exchangers in the risers performs better and peaks at 750 K after 15 hours compared to about 850 K after 17 hours for the design with heat-exchangers in the downcomers. During loss-of-heat-sink events the temperature evolution is similar for both types of reactors. The proton beam has to be shut off within 3 minutes to avoid exceeding the ASME level D. The natural convection of the 800 MW(thermal) design with heat-exchangers in the downcomer is sufficient to provide adequate cooling for all investigated accidents. This is rather certainly the only approach to cool a larger ADS with the beam on during these accidents. However, the use of water as a secondary coolant is probably not possible. Besides the use of oil, which can only be used for temperatures lower than about 700 K, one could use Pb/Bi in the secondary circuit to avoid any problems with steam or oil ingress into the ADS core. 21
VII. FUTURE WORK
An investigation should be done on the heat removal from heat-exchangers located in the riser through which argon gas passes. This would be important for a design with a simplified flow path. Additional means to improve the natural convection should be also investigated. For example lower pressure drops in the core and the heat-exchangers are important. The design of the heat-exchangers also affects the position of their thermal centers. For instance a more compact heat-exchangers located close to the free surface would increase the distance between thermal centers and increase natural convection.
VIII. ACKNOWLEDGEMENTS
Prof. W. Gudowski for support and valuable comments. The Swedish Fuel Management Company, Svensk K¨arnbr¨anslehantering, for financial support and the Joint Research Centre of the European Commission for providing the computational facilities and the license fee for the computer code.
22
Footnotes 1
Rupture of the protective oxide layer on structural components becomes severe.
2
Fuel failures occur, but the reactor vessel can withstand this temperature for several
hours. 3
American Society of Mechanical Engineers
4
Investment might be jeopardized, but fuel damage or radioactive release is unlikely.
5
Endanger the plant from an investment standpoint, although significant fuel failures
and radioactive releases are unlikely.
23
REFERENCES [1] Cinotti, L., Corsini, G., ”XADS Pb-Bi Cooled Experimental Accelerator Driven System - Reference Configuration”, Ansaldo Nucleare, unpublished (2001) [2] Generation IV Roadmap NERAC Subcommittee, ”Technology Goals for Generation IV Nuclear Energy Systems”, NERAC, (May 1 2001), http://gen-iv.ne.doe.gov/ [3] Juhn, P.E., Kupitz, J., Cleveland, J., Cho, B., Lyon, R.B., ”IAEA activities on passive safety systems and overview of international development”, Nuclear Engineering and Design, 201, pp. 41-59, (2000) [4] Frameatome ANP GmbH, ”SWR 1000”, http://www.de.framatome-anp.com/anp/e/foa/anp/products/s11 2.htm [5] Rubbia, C., Buono, S., Kadi, Y., Rubio, J.A., ”Fast Neutron Incineration in the Energy Amplifier as Alternative to Geologic Storage: The Case of Spain”, CERN/LHC/97-01 (EET) [6] Delpech, M., ”The Am and Cm transmutation - physics and feasibility”, Proc. Int. Conf. on Future Nuclear Systems, GLOBAL99, Jackson Hole, USA (1999) [7] US Department of Energy, ”Report to Congress on Advanced Fuel Cycle Initiative: The Future Path for Advanced Spent Fuel Treatment and Transmutation Research”, (Jan. 2003) [8] Computational Dynamics Ltd., Methodology Volume 3.10a (2000) [9] Hunsbedt, A., Magee, P.M., ”Design and performance of the PRISM natural convection decay heat removal system”, Proc. Int. Topical Meeting on Safety of Next Generation Power Reactors, pp.844-851, (1988) [10] Eckert, E.R.G., Drake, P.M., Heat and Mass Transfer, pp. 299, McGraw-Hill Book Compnany, New York (1959) 24
[11] Wider H.U., Carlsson J., Jones A., ”Beam Shut-Off in ADS Accidents - An Essential Requirement”, GLOBAL 2001, Paris (2001) [12] Adamov, E.O., White Book of Nuclear Power, N.A. Dollezhal Research and Development Institute of Power Engineering, Moscow, Russia (2001) [13] Incropera, F.P., DeWitt, D.P., Fundamentals of Heat and Mass Transfer, John Wiley & Sons, New York, USA, pp.492-493 (1996) [14] Bennet, D.J., Thomson, J.R., The Elements of Nuclear Power, Longman Scientific & Technical, UK, pp.25, (1991) [15] Nuclear Energy Agency, ”Accelerator-Driven Systems (ADS) and Fast Reactors (FR) in Advanced Nuclear Fuel Cycles - A Comparative Study”, ISBN: 92-64-18482-1, Available on the Web at: http://www.nea.fr/html/ndd/reports/2002/nea3109.html [16] Li, N., “Active control of oxygen in molten lead-bismuth systems to prevent steel corrosion and coolant contamination”, J. of Nuclear Materials, 300, pp. 73-81, (2002) [17] He, X.Y., Li, N., Mineev, M., j. Nuclear Materials, 297, pp. 214, (2001) [18] Gromov, B.F., Toshinsky, G.I., Stepanov, V.S., et al.,”Use of Lead-Bismuth Coolant in Nuclear Reactors and Accelerator-Driven Systems”, Nuclear Engineering and Design, 173, pp. 207-217, (1997) [19] Adamov, E.O., Orlov, V.V., “Nuclear development on the basis of new concepts of nuclear reactors and fuel cycle”, Proc. Heavy Liquid Metal Coolants in Nuclear Technology HLMC-98, Vol. 1, SSC RF-IPPE, Obninsk, pp. 24, (1999) [20] Rousanov, A.E., et al., ”Design and study of cladding steels for fuel elements of NPP using heavy coolant”, Proc. Heavy Liquid Metal Coolants in Nuclear Technology, Obninsk, Russia (1998) [21] Novikova, N., Pashkin, Y., and Chekunov, V. ”Some features of sub-critical blankets
25
cooled with lead-bismuth”, Int. Conf. on Accelerator Driven Technologies and applications, ADTTA99, 1999 [22] Kadi, Y., Buono, S., ”The Energy Amplifier Demonstration Facility: neutronic analysis of high-energy beam - target interaction”, Conf. on Heavy Liquid Metal Coolants in Nuclear Technology, Obninsk, Russia, (1999) [23] Pershagen, B., Light Water Reactor Safety, Pergamon Press, Oxford, England, pp.50 (1989), ISBN 0-08-035915-9 [24] Jeong, J.H., Kweon, Y.C., ”The effect of tube rupture location on the consequences of multiple steam generator tube rupture event”, Annals of Nuclear Energy, 29, pp. 1809-1826, (2002) [25] Toshinsky, G.I., Grigoriev, O.G., Efimov, E.I., Leonchuk, M.P., Novikova, N.N., Pankratov, D.V., Skorikov, D.E., Klimov, N.N., Stepanov, V.S., ”Safety Aspects of SVBR75/100 Reactor”, Workshop on Advanced Nuclear Reactor Safety Issues and Research Needs, Paris, France, (Feb. 2002) [26] Ludwig, P.W.P.H., Verkooijen, A.H.M., ”Thermal stresses in a core of a Fast Energy Emplifier cooled with lead under normal operating conditions including reactor scram and secondary pump trip”, ICENES 2000, (2000) [27] Eriksson, M., and J. Cahalan, ”Inherent shutdown capabilities in accelerator-driven systems”, Annals of Nuclear Energy, vol 29/14, pp. 1689-1706, (May 2002) [28] Carlsson, J., Wider, H., ”Emergency decay heat removal by reactor vessel auxiliary cooling system from an Accelerator-Driven System”, Nuclear Technology, 140, No. 1, pp.28-40, 2002 [29] Wallenius, J., Tucek, K., Carlsson, J., Gudowski, W., “Application of Burnable Absorbers in an Accelerator-Driven System”, Nuclear Science and Engineering, (2001)
26
[30] King, T.L., Landry, R.R., Throm, E.D., Wilson, J.N., ”Preapplication, Safety Evaluation Report for the Sodium Advanced Fast Reactor (SAFR) Liquid-Metal Reactor”, U.S. Nucelar Regulatory Commission, 15-1 (1991) [31] Shibli I A, European Creep Collaboration Committee (ECCC), ECCC coordination, ECCC Data Sheets (1999)
27
Table I. Critical temperature for structural materials and protective oxide layers. Characteristic problem
Temperature when problem occurs
Corrosion of structural material
893 K [20]
1
ASME2 level C
922 K [30]
3
ASME level D
977 K [30]
4
The limiting temperature to avoid creep 1173K [31]
5
failure in several hours under the given pressure conditions Melting point AISI316
1670 K [13]
28
Table II. Main characteristics of the ADS-HXR200. Plant Area
Reference Solution
Core power
250 MW(thermal), kef f 0.97 at BOL kef f 0.94 at EOL
Accelerator power
3 MW
Target unit
Pb/Bi eutectic, window type undecided
Fuel
U and Pu MOX
Coolant and moderator
Pb-Bi eutectic
Vessel height
9m
Vessel diameter
6m
Steel, reactor vessel
AISI 316L
29
Table III. Comparison between hand calculation and STAR-CD predictions. Temperature coolant, Hand calculation,
STAR-CD
[K]
[kW]
calculation, [kW]
573
291
316
873
1079
1197
30
FIGURES
FIG. 1. Schematic view of the ADS-HXD200.
31
FIG. 2. Schematic view of the RVACS of ADS-HXR200 and ADS-HXD200.
32
FIG. 3. Schematic view of the ADS-HXR200.
33
ADS-HXD
ADS-HXR
z2 z1
h
h
FIG. 4. Schematic view of displacement of thermal centers after an LOF accident for ADS-HXR200 and ADS-HXD200.
34
FIG. 5. Coolant flow field of ADS-HXR200 one hour after TLOP accident initiation.
35
900 ADS-HXR200 ADS-HXR200 (without bypass) ADS-HXD200
Temperature, K
800
700
600
500
0
5
10
15
Time, hours FIG. 6. Temperature evolution at core outlet during a TLOP accident for ADS-HXR200, ADS-HXR200 without BPRs, and ADS-HXD200.
36
Temperature, K
1100
1000
ADS-HXR200 ADS-HXR200 (without BPRs) ADS-HXD
900
800 0
5
10
15
20
25
Time, minutes FIG. 7. Temperature evolution at core outlet during an LOF accident for ADS-HXR200, ADS-HXR200 without BPRs, and ADS-HXD200.
37
Temperature, K
1000
800
ADS-HXR200 ADS-HXR200 (without bypass) ADS-HXD200
600
400
0
5
10
15
20
25
30
Time, minutes FIG. 8. Temperature evolution in the reactor vessel wall during an LOF accident for ADS-HXR200, ADS-HXR200 without BPRs, and ADS-HXD200.
38
2500 ADS-HXR200 ADS-HXR200 (without BPRs) ADS-HXD200
Temperature, K
2000
1500
1000
500
0
10
20
30
Time, minutes FIG. 9. Temperature evolution at core outlet during an LOHS accident for ADS-HXR200, ADS-HXR200 without BPRs, and ADS-HXD200.
39
850
Temperature, K
800 750 700 650
ADS-HXR800 ADS-HXD800
600 550 0
5
10
15
Time, hours FIG. 10. Temperature evolution at core outlet during a TLOP accident for ADS-HXR800 and ADS-HXD800.
40
Temperature, K
2000
1500
ADS-HXR800 ADS-HXD800
1000
500
0
5
Time, minutes FIG. 11. Temperature evolution at core outlet during an LOF accident for ADS-HXR800 and ADS-HXD800.
41
Temperature, K
1400
1200
1000 ADS-HXR800 ADS-HXD800
800
600
0
5
10
15
Time, minutes FIG. 12. Temperature evolution at core outlet during an LOHS accident for ADS-HXR800 and ADS-HXD800.
42
3
PASSIVE SAFETY APPROACHES IN LEAD/BISMUTH-COOLED ACCELERATOR-DRIVEN SYSTEMS
Hartmut U. Wider and Johan Karlsson
[email protected] [email protected]
Joint Research Center of the European Commission Ispra, Italy
Abstract For the safety of Accelerator-Driven Systems (ADS) it is important that the proton beam is shut off soon after the initiation of an accident and that the decay heat can still be removed if the regular heat removal fails. This paper presents a new device for passively blocking the proton beam when a Pb/Bi-cooled core heats up during an accident. An investigation was made of the passive removal of the decay heat and also the heat generated by a proton beam that impinges on the Pb/Bi coolant surface (due to the blocking of the beam pipe with Pb/Bi) during a Loss-ofHeat-Sink accident. A Reactor Vessel Auxiliary Cooling System (RVACS) similar to the one used in PRISM is considered. The results show that a small ADS with a blocked-off beam can be safely cooled for a few days. For larger ADSs vessel creep may start after several hours if the beam is not switched off completely. RVACS with novel features can cool larger ADSs with blocked beam for many days. Introduction Accelerator-driven Systems (ADSs) with heavy-metal Pb/Bi cooling lead only to mild power rises and a limited heat-up for reactivity insertions or beam power increases (Wider, 99). In the case of a loss of primary flow (Loss-of-Flow or LOF accident) modern designs with tall and slender vessels (Ansaldo Nucleare, 99) will lead to a sufficiently large natural convection flow that can remove the full power at elevated temperatures. If the heat removal by the secondary system stops (Loss-ofHeat Sink or LOHS accident) and the beam is not switched off, the core will slowly heat up, the rate depending on the reactor power and the heat capacity of the system (Karlsson, 2000). In the LOHS accident, the eventual shutting down of the beam is essential. For all other accidents it is also important to avoid local damage. The emergency decay heat removal also plays a key role in the LOHS accident, but only after the beam is shut off. Reactor Vessel Auxiliary Cooling Systems (RVACS) are particularly attractive for this purpose because they are fully passive systems. Shutting off the Proton Beam As mentioned above, the shutting off of the proton beam is of key importance in an LOHS accident but it is also relevant in other accidents. Switching off an accelerator is, in principal, simpler and faster than the mechanical insertion of safety rods in a critical system. Moreover, the accelerator will be switched off automatically in a station blackout condition that is a major contributor to the possible initiators of an LOHS accidents. The manual switching off of the accelerator based on increased temperature readings will remain an important option. There should also be an automatic switching off based on high thermocouple readings. If these methods fail, a
simple passive device for an automatic beam shut-off would be useful as a last resort. We recommend a melt-rupture disk in the side wall of the proton guide tube that would fail and flood this vacuum tube with heavy liquid metal (see Figs. 1 and 2). This would lead to the blocking of the proton beam and reduce the power to the decay heat level plus the power of the spallation source higher up in the system. We have been granted a patent for such a device (Wider, 98). This melt-rupture disk should be located just above the ADS core in order to sense significant temperature increases early on. Solder material around the melt rupture disk should be chosen such that it softens at a low enough temperature so that the vessel can still be cooled by an RVACS. However, the trigger temperature should also be high enough not to lead to an inadvertent activation. This would require the removal of the beam pipe and installation of a new melt-rupture disk.
Fig.1. Melt Rupture Disk in the side wall of evacuated proton beam tube
Fig,2 Cross section of the Melt Rupture Disk Decay Heat Removal with an RVACS for the Case of the Proton Beam Impinging on the Pb/Bi Coolant high up in the Vessel This part of our study concerns the question whether an RVACS could remove the decay heat plus the heat from a spallation source high up in the vessel. The 3-D thermal hydraulics code STAR-CD was used for this investigation. The pool type primary system and guard vessel from the Ansaldo design of an 80 MWth demonstration facility (Ansaldo, 1999) was used in this study. A Reactor Vessel Auxiliary Cooling System (RVACS) similar to the one used in the American PRISM design (Van Tuyle, 1989) was considered. This consists of a riser channel for the air that is adjacent to the guard vessel and a downcomer that is separated from the riser by an insulating wall. In our LOHS calculations the regular heat removal was completely stopped and the proton beam was assumed to continue. This led to a slow heat up of the core because of the large heat capacity of the Pb/Bi coolant. After 15 minutes the coolant temperature above the core had risen from 400° to 600°C. At that time the melt -
rupture disk was assumed to be pushed into the evacuated beam pipe which is subsequently flooded. Fig. 4 shows that the power then drops from 80 MW to a decay heat level of initially 5.4 MW (at the end of this calculation it is 0.25 MW) and additionally to a continuous 3 MW which represents the maximum beam power of the Ansaldo design. This is now inserted below the surface of the Pb/Bi coolant, high above the core. The spallation source was assumed to be uniform with its radial extent 1.2m and axial extent 1m. The radial extent of this heat source is too large but it allowed us to use coarser meshes. Fig.3 below shows the hot steady state in the vessel after 64 h. The main heat source at that time is the spallation source of 3MW in the darker colors in the upper left corner of the riser section. The Pb/Bi near the cooled wall is colder and therefore denser and moves down faster than the bulk of the coolant in the downcomer. This still adds a considerable natural circulation component to the continued forced convection through gas injection above the core as considered in the Ansaldo design and in this study. Fig.4 also shows that the flow velocity increases initially from the original 0.45m/s due to the early heat-up, but later on it decreases due to the lesser contribution of the natural circulation at the low ∆T in the vessel. The hot spot in the vessel wall is 1163K and located near the coolant surface where the hot coolant impinges on the wall. Above 1173K vessel creep is expected for long times. (Krieg, 99). In a calculation with a more optimized RVACS using an emissivity of 0.9 instead of 0.75, the hot spot in the vessel was only 1077K. A further optimization is possible by including fins in the air riser channel (Karlsson, 99). In an another calculation, it was assumed that the gas lift pumps would also fail (LOF+LOHS). The flow velocity was reduced by only about 50% due to the good
Fig.3 Coolant temperature and velocity distributions in the vessel 64 hours after LOHS initiation
Power Temperature Velocity
6e+7
4e+7
2e+7
0e+0 0.0e+0
5.0e+4
1.0e+5
1.5e+5
2.0e+5
0.460 0.455 0.450 0.445 0.440
Velocity, m/s
Power, W
8e+7
1300 1200 1100 1000 900 800 700 600 500 400 300 200 100 0 2.5e+5
Temperature, K
1e+8
Fig.4 Temperatures, velocities in riser and power in LOHS accident with beam blocking
0.435 0.430
Time, s
natural circulation capability. The critical temperature of the vessel wall was reached after 2 days for the non-optimized RVACS. In further calculations for a 480 MWth PRISM type core and a 7MW spallation source but with the same vessel geometry as used before, the critical vessel wall temperature of 1173K was already reached after about 3 hours with the nonoptimized RVACS. This means that only a major improvement in the grace time for saving the vessel has been achieved through the use of the melt rupture disk. If one used an RVACS with novel options such as water spray cooling of the guard vessel outside and the filling of the gap between the reactor and guard vessel with molten Pb/Bi (Karlsson, 2000), the 480 MWth reactor with the blocked beam and a spallation source high up in the vessel could be cooled permanently. Conclusions A melt rupture disk is a last resort device for passively bringing the power down in accident sequences in a Pb/Bi cooled ADS in which the proton beam is not shut off. A Loss-of-Heat-Sink accident in a small ADS with the beam on, could be cooled for several days with an RVACS if the melt rupture disk were used. For larger systems the grace time until creep in the vessel starts will be several hours. However, a more permanent cooling would be possible with advanced RVACS. REFERENCES - see also: http://itumagill.fzk.de/ADS/; http://nucleartimes.net Ansaldo Nucleare, Jan. 1999, “Energy Amplifier Demonstration Facility Reference Configuration”, Genoa, Italy Karlsson, J., 1999, “Decay Heat Removal by Natural Convection and Thermal Radiation from the Reactor Vessel”, ADTTA 99, Prague, Czech Republic Karlsson, J. and Wider H.U., April 2000 “New aspects of Emergency Decay Heat Removal by Auxiliary Cooling”, accepted for presentation at ICONE8, Baltimore, USA Krieg, R., 1999, FZK, Germany, personal communication Van Tuyle et al, 1989, “Summary of Advanced LMR Evaluations – PRISM and SAFR”, Brookhaven National Laboratory, Upton, N.Y. Wider H. U. and. Schönherr, H., “Beam pipe with safety function for accelerator driven nuclear systems”, European patent No 9811339.7 Wider H.U, Karlsson J. and Jones A.V, Sept. 1999, “Safety Considerations of Heavy Metal-Cooled Accelerator-driven Systems”, GLOBAL’99, Jackson Hole, USA
4
4
Safety Aspects of Heavy Metal-Cooled Accelerator-Driven Waste Burners By H.U. Wider and J. Karlsson, JRC Ispra Summary Accelerator-driven, subcritical lead / bismuth cooled systems have several safety advantages. The critical accident initiators in such systems lead only to a relatively slow coolant heat-up that should be noticed by the reactor operators who will initiate a shutting down of the accelerator. This decreases the reactor power to decay heat levels. If the coolant temperature increase should go unnoticed, passive systems will lead to an automatic shutdown of the accelerator or a blocking of the proton beam. Emergency decay heat removal by natural air circulation cooling of the vessel outside is an attractive option for such a system. If no active or passive beam shut-off took place during a coolant overheating due to a significant Loss of Heat Sink (LOHS) accident, a core melt could eventually occur. Oxide fuel would probably mix with the heavy metal coolant with its high boiling point and circulate in the primary system in a coolable fashion. This type of scenario seems to have happened in a core melt accident in a Russian Alpha submarine with its lead / bismuth cooled critical reactor.
1. Introduction When a new nuclear system such as the Accelerator-Driven System (ADS) is proposed, the early investigation of potentially severe accidents is important in order to point out the areas in which the design could be improved or whether passive devices could be introduced which would stop conceivable accident sequences. For the ADS it has been recognised early on that a complete Loss-of-Flow (LOF) would lead to problems if the accelerator was not switched off. Therefore, a first conceptual design [1] was based on natural circulation cooling and therefore required a rather tall vessel. A subsequent and more detailed design with a shorter vessel introduced the enhanced natural circulation cooling by gas bubble injection above the core [2]. This design still allows the removal of the nominal power for a certain time by pure natural circulation. However, this would still require a complete heat removal by the secondary loops. If these were not working, i.e. in a Loss-of-Heat-Sink (LOHS) accident, the core would slowly heat up (for the design in Ref. 2 by about 200 deg. C in 1000 s). Switching off the accelerator would get the power rapidly down to a decay heat level [3]. As will be shown in this paper this can also be achieved passively. An important initiator for LOF and LOHS accidents is a station blackout – however, in an ADS this would also lead to the shutting off of the accelerator and the decrease of the power to a decay heat level. The remaining problem would be the passive decay heat removal that can be achieved by natural air circulation cooling. Loss-of-Coolant (LOCA) accidents, which are a concern in LWR safety, should pose no significant problem. First, a lead/bismuth (Pb/Bi) cooled system is at a low pressure; second, a pool design has been proposed in which all the primary coolant is in one vessel and there is no primary piping which might develop leaks; and third, a guard vessel surrounds the main vessel in case the latter would leak. Another important safety aspect of a lead/bismuth or a lead coolant is that it is chemically rather inert and does not react strongly with air or water. This is an advantage relative to sodium-cooled fast systems.
2
2. Reactivity Accidents It has also been realised earlier that fast reactivity insertions do not lead to rapid power increases due to the subcriticality of an ADS which acts as a large delayed neutron group [4]. It is still interesting to do more detailed calculations that take into account other reactivity feedbacks besides the Doppler. The European Accident Code-2 (EAC-2), which was previously used for the analysis of sodium-cooled fast reactor accidents [5], was changed so that the point kinetics can take into account an external neutron source. This is only an approximation but it should be good enough to show the key features of an accident in an ADS. The reactor under consideration is the 800 MWe WAC paper design [6] used for benchmark calculations of accidents in fast sodium cooled reactors. Fig. 1a and 1b show the power and reactivity histories due to a 6$/s reactivity ramp with a total insertion of 3 $ (1$= 0.00345∆k/k) in a sodium-cooled ADS with an assumed subcriticality of -5 $. The subcriticality is conservatively assumed to be relatively small and the ramp rate and total insertion to be rather high, particularly since an ADS will probably have no reactivity control but only safety rods. Fig. 1b shows that the axial fuel expansion and Doppler reactivities reduce the net reactivity. That the Doppler reactivity stays negative shows that the reactor is overheated. This is also clear from the power trace in Fig. 1a which is still at 40% of nominal at 20 s. The EAC-2 code did not predict any pin failures, but the conditions were close to it. Another reactivity accident calculation with a 10 cent/s ramp and also a 3$ total insertion, led to a limited number of pin failures. The latter led to some fuel sweepout and enough negative reactivity to get the power back to nominal. To avoid the long-term elevated power and a possible limited number of pin failures, the proton beam should be switched off. For an overpower condition as in Fig.1 this should be done by passive systems (see also under beam power and LOF and LOHS accidents). To get the power back to or below nominal, but not to a decay heat level, shutdown rods could be used. The same reactivity accident with a 6 $/s ramp and a 3 $ total insertion was also calculated for the corresponding unscrammed critical sodium-cooled reactor. This led to a power pulse of more than 1000 times nominal and a nearly complete core destruction. Two aspects explain the difference. First the subcriticality of the ADS and second the positive void coefficient, which came into play in the accident in the critical reactor, but not in the ADS case. In the proposed ADS designs, lead/bismuth is used which has a rather low void coefficient. If the latter coolant is simulated in EAC-2, considerably larger positive reactivities can be inserted without leading to power excursions. 2. Beam Power Accident Since the proton beam intensity can be increased to compensate for burn-up, one can also imagine that the beam power is increased accidentally. Figs. 2a and 2b show the power and reactivity traces of such an accident with an assumed beam power increase by a factor 2 in one second. According to accelerator specialists this large increase is not credible. In this case lead/bismuth coolant was roughly simulated by giving the sodium a high boiling point of 1650°C and 1/10 of the sodium void reactivity. It can be seen in Fig. 2a that the power does
3
not increase too much and levels off at about 1.5 times nominal. Without feedbacks it can be shown analytically that a doubling of the beam strength leads to a doubling of the power. The elevated power in Fig. 2a leads to some pin failures after about 16.5 s. This would not have happened if the beam power was shut off actively or passively The subsequent fuel sweepout stabilises the power at a slightly elevated level. If some more pins rupture, the power would go somewhat further down. According to the meeting on Heavy Liquid-Metal Coolants in Nuclear Technologies (HLMC’98) in Obninsk, Russia on October 5-9th, a partial core melt in an Alpha class submarine showed that oxide fuel mixes with the lead/bismuth coolant and became distributed all over the primary system. From a safety and coolability point of view this would be very good, but to clean up the primary system afterwards would be a difficult effort.
3. Loss-of-Flow and Loss-of-Heat Sink Accidents both with the Spallation Source Active and Switched-off During the Accident Calculations for an unscrammed LOF accident in a critical 800 MWe sodium-cooled fast reactor are available from the Whole-Core Accident Calculation (WAC) group [6]. The LOF calculations which assumed a pump rundown with a flow halving time of 6 s, led to significant short-term power peaks of 200 - 3000 times nominal and nearly complete fuel melting and disruption. A LOF accident calculation for an ADS with the beam active (source on) uses the same reactor set-up and pump run-down as above, but with an assumed subcriticality of −$ 10 and an imposed neutron source that gives the same steady state power as the corresponding critical reactor. Fig. 3a gives the power history and Fig. 3b the associated reactivity contributions. The limited power rise is due to sodium boiling and a positive sodium voiding reactivity of up to 4 $. The power decrease is due to molten fuel dispersal of about 1/3 of the core. Nevertheless, the power does not go down to decay heat levels because the neutron source is still on and more fuel would melt unless some heat removal is restored. The new ADS cooling design of Ansaldo [2] allows a natural circulation cooling of a somewhat overheated core at full power under the condition that the secondary loops are still working normally. With such a design, the above LOF accident scenario that led to sodium boiling, fuel melting and dispersal would not occur. Another way of preventing the progression of the above described LOF scenario and also of LOHS scenarios due to the failure of the secondary loops, is the shutting off of the neutron source before sodium boiling starts [3]. Fig. 4a gives the power history and Fig. 4b the reactivity histories for the LOF case when the neutron source is shut off at 12.3 s. It can be seen that some sodium voiding still occurs but that the flow recovers completely. The core cools down to decay heat levels and this is the reason why the Doppler reactivity goes up. This shutting off of the accelerator could be done manually because the grace time will be long in an ADS with good natural circulation capability. In case this shutting off is not done, semi-passive systems based on thermocouple readings and computer logic should come into play. As a last resort simple passive means of interrupting the proton beam should be used. The authors of this paper recommend a melt-rupture disk at the side of the vacuum pipe through which the protons are streaming into the subcritical core [7]. When this melt rupture 4
disk is pushed inside the pipe by the coolant pressure, the pipe will be unconditionally filled up by liquid metal coolant that will block off the proton beam. An earlier approach by Prof. Rubbia and his team at CERN also relies on the filling of the beam pipe but with a device through which the rising heated coolant overflows into the evacuated proton pipe [1]. As mentioned above a proper treatment of a heavy metal coolant (Pb/Bi or Pb) is not yet available in EAC-2, only a simulation of the key aspects, namely a high boiling point (1650 °C) and a low void coefficient (assumed to be 1/10 of that of Na). This was used in the same ADS LOF calculation as above in which the neutron source was not shut off. It showed that the cladding melting preceded fuel melting and coolant boiling. Thus, molten cladding motion becomes important and should be modelled. Since steel cladding is a neutron absorber, its removal should lead to a power increase and fuel melting. Molten oxide fuel, which has a lower density than liquid Pb/Bi or Pb, should mix with the heavy metal coolant and be carried away by it (as in the 1968 accident of the Russian Alpha submarine). This would eventually terminate the accident without any significant power excursion [8]. In the case of a low melting point metal fuel, the fuel may dissolve in the heavy metal coolant and would then also be safely distributed in the primary system. The fuel / coolant mixing for oxide fuels or the rate of dissolution for molten metal fuel should be experimentally investigated. This could show that a heavy metal-cooled ADS could even cope with a highly unlikely core melting. First out-of pile tests could be done in the KROTOS facility of JRC Ispra.
4. Emergency Decay Heat Removal Another important safety aspect the efficiency of emergency decay heat removal i.e. the decay heat removal for the case when the secondary loops do not work as in the case of a Station Blackout accident. Passive systems based on natural air circulation cooling are attractive. The Reactor Vessel Air Cooling System (RVACS) appears preferable to a Direct Reactor Air Cooling System (DRACS) because the latter may have problems in case of the primary coolant leaking into the guard vessel. Properly scaled experiments are needed to confirm the performance of such a system which has been proposed earlier by the US PRISM and SAFR projects. At the JRC Ispra, a PhD research studentship has started for investigating different features and aspects of an RVACS using the 3-dimensional fluid dynamics code STAR-CD. This research will investigate several measures that can be taken in order to optimise the heat transfer from the guard vessel, for example the optimisation of the gap width between the guard vessel and the collector cylinder, the addition of surface coatings on the outside of the guard vessel to increase the emissivity, fins on the guard vessel or the collector cylinder etc. These measures should not remove unacceptable amounts of heat during normal operation. A way of avoiding this could be the installation of flaps that hinder the air flow during normal operation. The opening mechanism of the flaps could be constructed in a way that they open automatically in accident conditions.
5
Conclusions Regarding the prevention and mitigation of severe accidents, accelerator-driven subcritical and lead / bismuth cooled systems with a good natural coolant circulation capability and a passive emergency decay heat removal are very attractive. The critical LOF or LOHS accident initiators in such a system result in a slow coolant heat-up. This should normally lead to a manual shutdown. If it did not happen, a passively induced shutdown of the accelerator or interruption of the proton beam should come into play. Since this is the only means to get the power down to a decay heat level, passive beam shut-off systems are important. Fast and rather large reactivity ramp insertions (which are of very low probability in a system without control rods) lead only to a limited power increase. This is due to the subcriticality of these systems. To stop such a limited overpower condition, a beam shut-off or the insertion of safety rods is necessary. The switching off of an accelerator is principally simpler than the insertion of safety rods. Moreover, in an ADS only the beam shut-off gets the power down to a decay heat level. If all active or passive measures failed to stop the spallation source in an accident leading to some core melt, the heavy metal coolant would help to mitigate such an accident. In the case of oxide fuel, a mixing of the fuel with the coolant and its dispersal in the primary vessel is expected based on the Russian experience with a partial core melt accident in a lead/bismuth cooled submarine reactor. Metal fuel may actually dissolve in a heavy metal coolant, but there are questions regarding the rate of dissolution. That fuel may mix or dissolve in heavy metal coolant with its high boiling point could be a very positive aspect for the highly unlikely case of fuel melting. References [1] C. Rubbia et.al. “Conceptual Design of a Fast Neutron Operated Energy Amplifier”, CERN/AT/95-44 (ET) [2] L. Cinotti, “Inherent Safety Mechanisms: Experimental Results in Enhanced Natural Circulation”, Int. Workshop on Physics of Accelerator-Driven Systems for Nuclear Transmutation and Clean Energy Generation Trento, Italy, Oct. 97 [3] H.U. Wider, “Safety of Accelerator-driven Nuclear Waste Burners”, Proceedings of Jahrestagung Kerntechnik, München, May, 1998 [4] H. Rief and H. Takahashi, “Safety and control of Accelerator-Driven Subcritical Systems”, International Conference on Accelerator-driven Transmutation Technologies and Applications, Las Vegas 1994 [5] H.U Wider, “The European Accident Code-2: Overview and Status” Proceedings of the 1990 Fast Reactor Safety Meeting, Snowbird, Utah, USA, Aug. 1990 [6] H. U. Wider et al, “Comparative analysis of a hypothetical loss-of-flow accident in an irradiated LMFBR core using different computer models for a common benchmark problem”, EUR 11925 EN, 1989 [7] H. U. Wider and H. Schönherr, “Beam pipe with safety function for accelerator – driven nuclear systems”, European patent No 9811339.7 [8] W. Maschek, B. Merk, FZK and H. Wider JRC Ispra, “Some Safety Studies for Accelerator -Driven Subcritical Systems”, Second International Topical Meeting on Nuclear Applications of Accelerator Technology, AccApp’98, Gatlinburg, USA, Sept. 98
6
2.0
1.5
1.0 0
5
10
15
20
Time, s
Fig. 1a Power in 6$/s Reactivity Accident Total Insertion 3$, Subcriticality -5 $
1
0
Reactivity, $
Power, P/P0
Power
-1
-2 Net Reactivity Inserted Reactivity Voiding Reactivity
-3
Doppler Reactivity Fuel Expansion React. Fuel Motion Reactivity
-4
-5 0
5
10
15
20
Time, s
Fig. 1b Reactivities in 6$/s Accident with 3 $ Total Insertion
7
Power and Beam Power, P/P0 and BP/BP0
2.5
Power Beam Power 2.0
1.5
1.0 0
5
10
15
20
25
Time, s
Fig. 2a Beam Power Increase Accident
2
Reactivities, $
0 Net Reactivity Inserted Reactivity Voiding Reactivity Doppler Reactivity Fuel Expansion React. Fuel Motion Reactivity
-2
-4
some fuel motion starts
-6
-8
-10 0
5
10
15
20
Time, s
Fig. 2b Reactivities in Beam Power Increase Accident
8
25
1.6
Power 1.4
Power, P/P0
Fuel Motion Starts 1.2
1.0
Sodium Voiding Starts
0.8
0.6 0
5
10
15
20
25
Time, s
Fig. 3a Power in LOF Accident with Source on, Sodium-Cooled ADS
10 Net Reactivity Inserted Reactitivity Voiding Reactivity Doppler Reactitivity Fuel Expansion React. Fuel Motion Reactivity
Reactivity, $
5
0
-5
-10
-15
-20 0
5
10
15
20
Time, s
Fig. 3b Reactivities in LOF Accident with Source on
9
25
Power
1.0
Power, P/P0
Accelerator stops
0.5
0.0 0
10
20
30
40
50
60
Time, s
Fig. 4a Power History in LOF Accident with Source Switched off
5
Reactivities, $
0 Net Reactivity Inserted Reactivity Voiding Reactivity Doppler Reactivity Fuel Expansion React.
-5
-10
-15 0
10
20
30
40
50
60
Time, s
Fig. 4b Reactivities in LOF Accident with Source Switched off
10
5
DECAY HEAT REMOVAL BY NATURAL CONVECTION AND THERMAL RADIATION FROM THE REACTOR VESSEL Johan Karlsson Institute for Systems, Informatics and Safety, Joint Research Centre 21020 Ispra, Italy E-mail:
[email protected] Abstract. Analytical investigations of the natural air circulation cooling at the external vessel of a heavy metal cooled Accelerator Driven System (ADS) are on the way. The preliminary demo design of Ansaldo and the air cooling of the earlier US advanced reactor program are considered. The fluid dynamics code Star-CD is used for the analysis. For the testing of the Star-CD code and the heat transfer correlations for coarse mesh calculations, the FZK Pasco experiments were analysed. Then a search for an optimal decay heat removal rate was made and parameters like the surface roughness, surface emissivity, and the gap width between the guard vessel and the collector wall have been investigated on a simplified geometry only containing the air channel and the vessel wall at constant temperature. Calculations on the complete geometry have been performed at two decay heat generation rates in order to find the equilibrium temperature in the reactor vessel (1% and 5% of normal operation power). Even for the higher rate the temperature in the reactor vessel remains below 1000°C.
INTRODUCTION In several reactor concepts, natural air convection is used to remove the decay heat in accident scenarios. In most concepts, this is the ultimate backup system remaining when the normal heat removal systems fail. Since these systems are totally passive, they have to be very reliable. Two types of natural air convection systems can be distinguished. Firstly, systems that remove decay heat from the outside of the reactor vessel by means of natural air convection and thermal radiation. Two examples are the Reactor Vessel Auxiliary Cooling System (RVACS) for the earlier US PRISM design [1], and the Reactor Vessel Air Cooling System for Prof. Rubbia's Energy Amplifier (EA) [2]. Moreover, a new preliminary design of an 80MWt Demo from Ansaldo [3] also uses RVACS. Secondly, there are backup decay heat
exchangers in the primary circuit having secondary loops connected to natural draft heat exchangers. Here, examples are the Direct Reactor Auxiliary Cooling System (DRACS) in the earlier US SAFR design [4] and in the European Fast Reactor (EFR) design [5]. A criticism of the earlier RVACS systems was that in the case of a reactor vessel rupture the natural air circulation would transport radioactive particles into the atmosphere. Regarding the DRACS, the drawback is that they are not totally passive, but have to be activated to start removing heat. In this research, only RVACS is considered.
All calculations were performed with the fluid dynamics code Star-CD. It was mainly chosen because other researchers have used this code earlier in similar detailed calculations with satisfactory results, for example S.Buono et al. [6]. Star-CD employs a radiation model that is very important for these investigations. Furthermore, it incorporates a two-layer boundary model, which makes good predictions of the heat transfer and of the velocities next to the walls due to a fine mesh in the laminar boundary layer. Other calculations have been performed on a coarse mesh employing wall functions.
VERIFICATION OF STAR-CD WITH THE PASCO EXPERIMENT The intention of the PASCO experiment [7] was to determine the coolability of a composite containment by natural air convection and thermal radiation. A rectangular chimney was used for the experiment, which was heated asymmetrically by one wall. At several locations in the channel, the temperature and the velocities of the air were measured.
The calculations performed with Star-CD were compared to these values in order to evaluate the code and the different heat transfer correlations. The table below shows the results of the outlet heat flux and mass flux calculations for the Two-Layer model, a standard GrPr correlation for turbulent flow and natural convection, and a RePr correlation [8]. The temperature of the heated wall was 150ºC and the emissivity 0.9. For the Two-Layer model and the RePr correlation the temperature and velocity profiles were also in good agreement. Table 1. Comparison between different heat transfer correlations and the PASCO experiment.
Two-Layer
Heat flux [W] (Star-CD) 6700
Mass flux [kg/s] (Star-CD) 0.37
Heat flux [W] (experiment) 6500
Mass flux [kg/s] (experiment) 0.35
0.13(Gr Pr)1 / 3
5400
0.53
6500
0.35
1.22 Re 0 .457 Pr 0.8
6600
0.355
6500
0.35
DESCRIPTION OF THE GEOMETRY The total height of the reactor vessel in the Demo design of Ansaldo is about 8m. The radius of the reactor vessel is 3.25m, the gap between the reactor vessel and the guard vessel is 25cm, and the gap between the guard vessel and the collector wall was assumed to be 18cm. The picture below shows the whole geometry.
Figure 1. Picture of whole set-up
RESULTS FROM PARAMETER INVESTIGATION One parameter at a time was examined in order to find its influence on the air mass flux and the heat removal capacity. The investigations were performed on a simplified geometry of the air ducts outside the guard vessel and the collector wall. The coarse mesh approach with the RePr correlation was used since the fine mesh calculations with Star-CD require long computing times. In the reference case, the surface emissivity is set to 0.75, and the surface roughness is hydrodynamically smooth. Furthermore, no fins are used and the guard vessel temperature is set to 500ºC.
The surface emissivity The surface emissivity has a great impact on the heat removal from the outward facing side of the guard vessel. At a surface emissivity of 1.0, the collector wall reaches almost as high a temperature as the guard vessel. Thus, the surface at high temperatures facing the airflow is doubled. Consequently, the heat removal rate is increased by ~100% when thermal radiation is included in the calculation. This tendency grows more important at higher temperatures since the thermal radiation is dependent on the temperature to the fourth power, Q ∝ Ta4 − Tb4 . The importance of the emissivity with regard to the maximum temperature in the reactor vessel was evaluated by T.Hung for a similar case, and there it was found that an emissivity of e=0.5 compared to e=0.85 make 100ºC difference after 60 hours of air-cooling. [9] Results below
3.0
12
2.0
11
1.0
0.0 0.00
Heat removal Mass flux
0.25
0.50
0.75
Emissivity Figure 2. Heat removal as a function of emissivity
10
9 1.00
Mass flux [kg/s]
Heat removal [MW]
show the heat removal as a function of the surface emissivity.
The surface roughness The heat removal is increased with higher surface roughness since it causes better mixing of the air in the boundary layer. On the other hand, this is counterbalanced by the increasing pressure loss due to the higher surface roughness. This investigation is not finished, but as can be seen below, the surface roughness seems not to have a great impact on the heat removal. 1.785
Heat removal [MW]
1.780 1.775 1.770 1.765 Heat removal
1.760 1.755 0.000
0.001
0.002
0.003
0.004
Roughness [m] Figure 3. Heat removal as a function of surface roughness
Fin pitch The fin pitch is of the greatest importance. The smallest fin pitch investigated was 5.0cm and the largest 60cm. The optimal pitch regarding the heat removal was found to be between 5cm and 10cm, where the heat removed was more than twice that in the case without fins. However, in investigations performed by H.Tzanos where fin pitches from 5cm up to infinity were examined, the optimal fin pitch was found to be the smallest in this range, namely 5cm. [10] The figure below depicts the heat removal and the mass flow as a function of fin interval according to the Star-CD calculations.
19
4.0
3.0
17 16
2.0 15 14
1.0
Heat removal Mass flux
0.0 0
10
20
30
40
50
60
Mass flow [kg/s]
Heat removal [MW]
18
13 12 70
Fin interval [cm] Figure 4. Heat removal as a function of fin interval
Gap width The heat removal increased by about 15%, when the gap width was changed from 10cm to 60cm. Higher temperature differences between the walls and the fluid explain the higher heat removal rate for a larger gap. 30
2.0
25
1.6 1.4
20
1.2 1.0
15
0.8 Heat removal Mass flux
0.6
10
0.4 0.2 0.0 0
10
20
30
40
50
60
Gap width [cm] Figure 5. Heat removal as a function of gap width
5 70
Mass flux [kg/s]
Heat removal [MW]
1.8
Guard vessel surface temperature The heat removal depends very much on the wall surface temperature of the heated wall. As mentioned above radiation also plays a very important role in transporting the energy. Most of the radiated energy from the guard vessel is deposited on the collector wall; the air absorbs only about 3% directly. The figures below show the heat removal as a function of the heated wall temperature. Additionally, all calculations have also been done without radiation. It could be noted that the radiation grows more important with higher temperatures. At a heated wall temperature of 300ºC, the heat removal due to radiation is about 44% while at 700ºC it is 56%. The fraction does not increase more because the collector wall is already nearly as hot as the guard vessel in both cases. 4.0
10.0
3.0
8.0 2.0
6.0 4.0
1.0
0.0 200
Heat removal Heat removal, no radiation Mass flux Mass flux, no radiation
300
400
500
600
700
Mass flux [kg/s]
Heat removal [MW]
12.0
2.0
0.0 800
Temperature [oC] Figure 6. Heat removal as a function of the heated wall temperature
THE FULL CASE CALCULATION Two calculations were performed at decay heat power levels of 1% and 5% of normal operational power. The examinations aimed to find the coolant temperature where the amount of heat removed by Reactor Vessel Auxiliary Cooling System (RVACS) would balance the power coming into the system. A possible scenario where these decay heat levels could occur are when the shutdown rods are inserted with the beam is still on. The decay heat generation from the core alone is less than 1% after about 10 hours.
The Ansaldo design was used in these two calculations. Thus, a height of 8m, a radius of 3.25m, a ‘radiation gap’ between the reactor vessel and the guard vessel of 25cm, and a gap between the guard vessel and the collector wall of 18cm were employed. The surface roughness was hydrodynamically smooth, no fins were used, and the emissivity was set to 0.75. The pressure drop in the core was set to 20000Pa at 0.42m/s, and the pressure drop was set to change in proportion to the flow velocity. The heat transfer correlation Nu = 1.22 ⋅ Re 0.456 Pr 0 .8 was used to calculate the heat transfer coefficient. This correlation was
developed at the Argonne National Laboratory, USA, for the American Liquid Metal Reactors’ programme. [10]
1% AND 5% DECAY HEAT POWER At 1% decay heat power, the maximum temperature was found to be 760K (487ºC). This temperature causes no problems for the structural materials. The melting point of structural material is at about 1400ºC, and the strength of the material would not be impaired below 1100ºC. Consequently, the decay heat at 1-% nominal power is easily removed by the RVACS. At the decay heat power of 5%, the maximum temperature in the reactor vessel was 1253K (980ºC). The figure below depicts how the equilibrium temperature was reached through a single-transient approach. The initial reactor vessel temperature was a guess based on the parametrical study on the heat removal from the heated wall at different temperatures.
774
1300
772
1250
770
1200
768
Max. Temp 1% power Max. Temp 5% power
1150
766
1100
764
1050
762
1000
760 0
10
20
30
40
50
Time [hours] Figure 7. Single-transient approach
60
950 70
Temp [K], 5% power generation
Temp [K], 1% power generation
See below
CONCLUSIONS A 1-% decay heat level is easily removed using RVACS, without any impairment of the structural material. The highest temperature in the reactor vessel, at the time when the RVACS cooling was equal to the power generated in the system was 760K (487ºC). Even at 5-% decay heat generation level, sufficient amounts of heat would be removed by RVACS. The maximum temperature in the reactor vessel coolant was 1253K (980ºC) at this level. Probably no impairment of the structural material would occur at this level either since steel strength is not reduced before about 1100ºC, some long-term creep could be expected though.
In the future, the full case will be examined in order to optimize the heat removal during accident scenarios and minimize the heat removal during normal operation. Furthermore, design measures to avoid radioactive release into the atmosphere will be considered. REFERENCES [1]
Evaluations of 1990 PRISM Design Revisions, Brookhaven National Laboratory,
March 1992 [2]
C. Rubbia et al., Conceptual design of a Fast neutron operated high power Energy Amplifier, CERN/AT/95-44(ET), September 1995
[3]
Ansaldo
Nucleare,
Energy
Amplifier
Demonstration
Facility
Reference
Configuration, January 1999 [4]
Summary of Advanced LMR Evaluations – PRISM and SAFR, Brookhaven National Laboratory, Oct. 1989
[5]
M. Düweke, H.J. Friedrich, The Direct Reactor Cooling-System of EFR Overview and Activities, Proceedings of the 1990 International Fast Reactor Safety Meeting – Vol. 2, American Nuclear Society, 1990, pp. 309-311
[6]
Methodology, Star-CD version 3.05, Computational Dynamics Limited
[7]
X. Cheng, U. Müller, Turbulent natural convection coupled with thermal radiation in large vertical channels with asymmetric heating, Int. J. Heat Mass Transfer, Vol.41, No.12, pp.1681-1692, 1998
[8]
A. Hunsbedt, P.M. Magee, Design and Performance of the PRISM Natural Convection Decay Heat Removal System, Proceedings of the International Topical Meeting on Safety of Next Generation Power Reactors, 1988, pp. 844-851
[9]
T. Hung, S.K. Wang, The Thermal-Hydraulic Analysis of a Passive Heat Removal System in Pool-Type Sodium Cooled Fast Reactors, Nuclear Science Journal, Vol. 34, No.1, 1997, pp. 9-22
[10]
C.P. Tzanos, J.H. Tessier, D.R. Pedersen, An optimization study for the reactor vessel auxiliary cooling system of a pool liquid-metal reactor, Nuclear Technology, Vol.94, Apr. 1991, pp. 68-79
6
6
Application of burnable absorbers in an accelerator driven system Jan Wallenius, Kamil Tuˇcek, Johan Carlsson, Waclaw Gudowski Department of Nuclear & Reactor Physics Royal Institute of Technology 100 44 Stockholm, Sweden E-mail:
[email protected]
Abstract The application of burnable absorbers (BA) for minimization of power peaking, reactivity loss and capture to fission probabilities in an accelerator driven waste transmutation system (ADS, ATW) has been investigated. 10 B enriched B4 C absorber rods were introduced into a lead/bismuth cooled core fuelled with TRU discharges from light water reactors in order to achieve smallest possible power peakings at a BOL sub-criticality level of 0.97. Detailed Monte Carlo simulations show that a radial power peaking equal to 1.2 at BOL is attainable using a four zone differentiation in BA content. Using a newly written Monte Carlo Burnup code (MCB), reactivity losses were calculated to be 640 pcm per percent transuranium burnup, for unrecycled TRU discharges. Comparing to corresponding values in BA free cores, BA introduction diminishes reactivity losses in TRU fuelled sub-critical cores by about 20%. Radial power peaking after 300 days of operation at 1200 MW thermal power was less than 1.75 at a sub-criticality level of 0.92, which appears to be acceptable, with respect to limitations in cladding and fuel temperatures. In addition, the use of BA yields significantly higher fission to capture probabilities in even neutron number nuclides. Fission to absorption probability ratio for 241 Am equal to 0.33 was achieved in the configuration here studied. Hence, production of the strong α-emitter 242 Cm is reduced, leading to smaller fuel swelling rates and pin pressurization. Disadvantages following BA introduction, such as increase of void worth and decrease of Doppler feedback in conjunction with small values of βeff , need to be addressed by detailed studies of sub-critical core dynamics.
1
System concept
Introduction
The burnable poison option was considered for the CAPRA design [11], in order to cope with the high Pu reactivity. The option was however rejected due to degradation of reactivity coefficients needed for safe operation of a critical reactor. On the other hand, in a sub-critical system, one may have enough criticality margin in order to accommodate a positive void coefficient in conjunction with vanishing Doppler feedback. In addition, the absorber will have a positive influence on transmutation efficiencies in terms of fission to absorption ratios, since they shield slower neutrons from being captured by even neutron number TRU nuclides. While already having introduced BA into the core, a differentiated concentration could be used for flattening of the core’s power density distribution. As it turns out, radial power peakings at BOL can be virtually eliminated by an appropriate distribution of absorber pins among the fuel pins in each sub-assembly. Since the absorbers should be burnable, we have chosen enriched boron carbide as a reference material. Boron carbide is used in FBR control rods due to its high reactivity worth in fast neutron spectra. Since sub-criticality is assumed, an additional cost is to be paid in terms of spallation neutrons to initiate the fission cascade. In order to maintain a high multiplication of source neutrons during burnup we substitute the absorber matrix in the inner core zones with depleted uranium, which also is more tolerant to increase in power peaking. Obviously, this diminishes effective TRU transmutation rates somewhat. Even with a successful power flattening strategy implemented at BOL, power peaking will increase with burnup in sub-critical systems, as neutron multiplication decreases. Since power densities are limited by linear ratings, the number of fuel pins and consequently the number of dedicated cores needed to manage the minor actinides will depend on choice of fuel matrix and coolant. In the present study we have adopted nitride fuels, which may enable a doubling of linear ratings, as compared to the corresponding uranium free oxide and metal fuels. Expected problems with high temperature stability of nitride fuels may be addressed by use of nitrogen bonding, which is known to suppress decomposition of actinide nitrides up to temperatures of 2800 K [14, 15]. Production of 14 C may be minimised by usage of 99% 15 N enriched nitrogen, today available on a commercial basis. We have chosen lead/bismuth as a coolant in our present simulations, in order to mitigate the increase in void worth due to massive BA presence in the core. The spallation target was also set to be liquid/lead bismuth, in accordance with European ADS project preferences. One purpose of our design study is to show how one may avoid americium production during the transmutation of transuranics nuclei. Thus the composition of TRU vector, exhibited in Table 1, was assumed to be that of typical un-recycled PWR spent fuel 30 years after discharge (which may be a reasonable value for the ADS introduction time delay). This vector differs from the one assumed for
As was shown in studies made at CEA, recycling of both plutonium, americium and curium is mandatory in order to achieve a substantial reduction factor ( 100) of the radiotoxic inventory sent to geological repository from present nuclear parks [1]. The introduction of a major fraction of minor actinides into the nuclear fuel cycle however raises complex issues regarding reprocessing, fuel fabrication, core behaviour during transients and fuel performance during irradiation. In general, it is assumed that americium (and curium) is to be recycled in fast reactors, due to the superior neutron economy of the fast neutron spectrum, which will allow for higher burnups, and hence lower losses to secondary waste streams. One option being considered for the minor actinide (MA) management is burnout in a second stratum of "dedicated" actinide burner reactors [2, 3]. As the vanishing Doppler feedback in conjunction with a very low βeff would lead to deterioration of safety margins for critical configurations, sub-critical designs have been adopted for these MAburners by a number of authors [4, 5, 6, 7]. Excess plutonium from light water reactors may be recycled in fast reactors. Homogeneous mixing of Pu with the minor actinides in accelerator driven systems (ADS) has also been proposed [8, 9, 10]. Since requirements on Doppler feedback are relaxed, the ADS core spectrum can be much harder than for corresponding critical configurations [11, 12, 13]. Consequently, americium and curium production in the entire fuel cycle is minimised. A major drawback of this approach is the large reactivity loss appearing due to burnout of 239 Pu and 241 Pu, with a concomitant increase in power peaking that may not easily be compensated for by a change in accelerator proton current. Consequently, maximum burnup becomes limited by power peaking factors, rather than by radiation damage to cladding and fuel. The introduction of thorium as a fertile material to keep up reactivity [9] would imply the implementation of an entire new fuel cycle, and must be regarded as a distant future option. The application of burnable absorbers to perform the same task has to the authors’ knowledge not received much attention in the published literature, apart from the suggestion to use 99 Tc in-core sub-assemblies [10]. Hence we have performed an investigation of advantages and drawbacks resulting from a massive introduction of classical burnable absorbers in sub-critical TRU-fuelled cores. First we show how power peakings can be minimised by use of a differentiated concentration of neutron absorbers in a lead/bismuth cooled core. Then, burnup calculations reveal the impact of the BA introduction on reactivity losses and effective minor actinide destruction rates. Finally we discuss how the design can be further improved. 2
Isotope 237 Np 238 Pu 239 Pu 240 Pu 241 Pu 242 Pu 241 Am 243 Am 244 Cm
Mass fraction 0.049 0.019 0.494 0.217 0.033 0.062 0.108 0.016 0.002
Table 1: TRU vector of the fuel used in the present study. The composition represents the TRU discharge of an average PWR after a burnup of 40 GWd/t and 30 years of cooling.
the CAPRA design, i.e. a mixture of Pu from fresh PWR discharges and recycled CAPRA plutonium [16]. The core design is hence to be regarded as a start-up configuration in a two-component scenario [6]. It is true that the Pu vector in a plutonium only recycling scheme reaches equilibrium within 25 years, [16]. However, adding the inevitable americium recycling, the appearance of (quasi) equilibrium TRU vectors is delayed by another 50 years [17, 9], and therefore we find it relevant to start by investigating requirements for start-up cores.
Figure 1: Cross section of the Sing Sing Core. 84 hexagonal fuel assemblies with duct flat to flat distances of 22.0 cm are configured in four fuel zones containing variable concentrations of TRU, depleted uranium (yellow zones) and boron carbide (blue zones). Technetium pins are present in zone two, and grey denotes steel reflector assemblies.
tron cross section library was processed with NJOY [19] to create Doppler broadened pointwise cross section files for MCNP [20] in the temperature range 300 K to 1800 K. A three dimensional geometry setup was constructed, where all 84 fuel bundles of the core were modelled separately. Duct walls were explicitly present, while a smeared coolant/cladding/fuel material composition was used for the duct interior in most cases. MCNP4C with explicit delayed neutron transport was used to obtain k-eigenvalues and βeff , while MCNPX [21] in proton source mode was applied to all calculations of neutron fluxes, power densities, cross sections, and neutron life times at BOL. A pin by pin model was used to calculate the relative fraction of heat deposition in cladding and coolant. Burnup calculations were made with MCB, a continuous energy Monte Carlo Burnup code being developed at KTH
Core design Figure 1 shows the geometry of our preliminary core design. The core is baptized to the "Sing Sing Core" (SSC), alluding to Swedish laws on reactor design. The fuel assembly duct flat to flat distance was taken to be 22.0 cm. The radius of our spallation target then becomes 24.9 cm, with a five mm thick steel container enclosing the target circuit. The active core height is assumed to be 100 cm, with proton beam impact located 18 cm above mid plane. The choice of lead/bismuth as a coolant leads to certain constraints on sub-assembly designs, required for avoiding problems with corrosion of clad and structural material: Pitch to pin diameter ratios have to be comparatively large, to compensate for limitations in Pb/Bi flow velocities (v < 3.0 m/s) imposed by erosion rates of cladding protective oxide films [18]. Pellet, pin and assembly dimensions were configured to keep outer cladding temperatures below 600 degrees C in the hottest channel at EOL and are displayed in Tables 2 and 3. Technetium and burnable absorber pellets have the same diameter as the the fuel pellets, but come without the central hole.
Property Active pin length Pellet inner radius Pellet outer radius Clad inner radius Clad outer radius Fuel density Smear density
Modelling tools
Value 100 cm 1.00 mm 2.40 mm 2.49 mm 2.94 mm 0.90 TD 0.70 TD
We have applied continuous energy Monte Carlo simulation methods to all aspects concerning neutronics and bur- Table 2: Pin and pellet design parameters of the lead/bismuth, nup modelling of the Sing Sing Core. The JEF2.2 neu- nitride fuelled Sing Sing Core. 3
in cooperation with AGH in Cracow [22, 23]. MCB extends MCNP with internal modules for reaction rate and heating calculations in flight. MCNPX was used to write a low energy (E 20 MeV) neutron source on the surface of the spallation target, that was employed for the burnup calculations with MCB. The effect of neglecting the high energy tail of spallation neutrons (1.4% of the source flux entering the core) was estimated to be within Monte Carlo error bars for power densities and actinide burnup.
"!$#%&'()+*-,/.01($2
?A@CB
3$46587:9<;>=
Absorber introduction The absorber ideally should serve the multifold purpose of neutron capture shielding, power peaking flattening, and mitigation of reactivity loss. While other materials like hafnium and europium are good absorbers in a fast neutron spectrum, they are not burnable as boron is. Employing 95% 10 B enriched B4 C, neutron moderation as well as absorber inventory is minimized. Figure 2 displays the absorption cross section of 10 B superimposed on the capture and fission cross sections of 241 Am. It is evident that an enriched boron carbide environment may suppress production rates of 242 Cm, while only affecting 241 Am fission rates to a minor extent. Thus, boron carbide pins were introduced into fuel zones 3 and 4 in proportions yielding the flattest possible power distribution at BOL. Depleted uranium replaced the absorber in zone 1 and 2, in order to improve spallation neutron multiplication. Finally, technetium pins were positioned in zone 2, for transmutation purposes. Table 4 shows a distribution (relative volume fraction) of transuranium nitride, uranium nitride and BA yielding a BOL radial power peaking equal to 1.2. The resulting radial power density along the symmetry axis of the core is displayed in Figure 3. Evidently, power peakings similar to those of critical reactor cores are feasible in subcritical systems. The capture of slower neutrons in absorber rods placed in direct conjunction to fuel rods leads to suppression of the low energy part of the neutron spectrum in the fuel. By introducing the absorber pins directly in fuel assemblies instead of having them in separate fuel bundles (FBR approach), one avoids local flux suppression and is able to utilize the full reactivity worth of the absorber. Figure 4 exhibits the impact of the high boron carbide concentration in
Property Pitch/pin ratio Fuel pins Absorber pins Technetium pins
zone 1 1.95 331 0 0
zone 2 1.785 317 0 80
zone 3 1.785 251 146 0
S-TCU
V"W8X:Y[Z
α
KML6N8O:PRQ
γ
D0E
Figure 2: Absorption cross section of 10 B, compared to the capture and fission cross sections of 241 Am. For high 10 B concentrations in fuel bundles, neutron capture in 241 Am will be suppressed below 500 keV, without significant deterioration of 241 Am fission probabilities.
the outer fuel zones. While the magnitude of the neutron flux above the fast fission threshold (En \ 1 MeV) is similar in zones one and three, a suppression factor larger than three is achieved in the resonance region (En 100 keV). The variation in neutron spectrum over the core is further characterized in Table 5, where median and flux weighted neutron energies are given for each fuel zone. Comparing with the pin data given in Table 4 it is clear that a higher fraction of absorbing material, and a smaller fraction of diluent in the fuel (in our case 238 U) yields more energetic neutron spectra. Consequently, spectrum averaged cross sections for neutron capture in even neutron number nuclides become smaller in outer zones 3-4, and fission to absorption probabilities rise. The spectrum averaged cross sections for capture and fission in relevant actinides at BOL are displayed in Tables 6 and 7. Note the significant decrease in capture cross sections as median neutron energies increase. The probability of fission versus capture as a consequence of neutron absorption in a fuel nuclide is of signifi-
Material Zone 1 Zone 2 Zone 3 Zone 4 Average
zone 4 1.785 319 78 0
TRUN 0.33 0.38 0.63 0.80 0.60
UN 0.67 0.42 0.00 0.00 0.18
B4 C 0.00 0.00 0.37 0.20 0.18
Tc 0.00 0.20 0.00 0.00 0.04
Table 3: Fuel sub-assembly specification of the lead/bismuth, ni-
Table 4: Fuel/BA material distributions given as relative volume
tride fuelled Sing Sing Core.
fractions in each fuel zone.
4
]R^`_badcfeRadgihkjmlAn
wzy
R`Amk" ¡i¢£¡¤z¥§¦ >¨-©< ª-¨kA¥< -«
kwzv
¬¯®±°/²
vdwz~
¬®ª°³
vdw}| vdwm{ vxwzy oqp8rdsut {v
|v
~Rv
vRv
RkAHf
yv
"z
km
`m
z
Figure 3: Fission power density at BOL in fuel rods along the Figure 4: Energy weighted neutron flux spectrum in zone 1 (upper line) and zone 3 (lower line). Neutron capture by 10 B in the outer core zones leads to a suppression of flux levels by a factor of more than three in the low energy region (En ´ 100 keV).
symmetry axis of the Sing Sing Core. Values are given relative to the core averaged power density. Radial power peaking is seen to be less than 1.2 at a k-eigenvalue equal to 0.972.
Nuclide 238 U 237 Np 238 Pu 239 Pu 240 Pu 241 Pu 242 Pu 241 Am 243 Am 244 Cm 245 Cm
cance not only for neutron economy, but also for the production of α-emitting nuclides. Formation of 242 Cm, having a half life of 163 days, will lead to problems with in-pile fuel swelling and pin pressurization as helium gas is accumulated. The appearance of 238 Pu and 244 Cm with half lives of 88 and 18 years, respectively, is likely to induce higher costs for reprocessing and fabrication. Thus one would like to maximize fission to absorption probability ratios for predecessors to the quickly decaying α-emitting isotopes. Table 8 displays this ratio for the Sing Sing Core and compares with corresponding values of CAPRA and Energy Amplifier designs [11, 9]. It is remarkable that the fission probability of the even neutron number nuclides is not only dependent on the presence of an absorber, but also the fraction of diluent material in the core appears to play a major role. In the CAPRA core, diluent steel and 11 B4 C moderator rods are introduced into the fuel assemblies to increase neutron leakage and moderation. In the Energy Amplifier, thorium ( 75% volume fraction of fuel) is used to maintain reactivity. Evidently the concomitant neutron moderation through inelastic scat-
Energy Zone 1 Zone 2 Zone 3 Zone 4
Median 194 keV 246 keV 356 keV 354 keV
Flux weighted 448 keV 522 keV 716 keV 734 keV
E > 1 MeV 10.6% 13.8% 21.8% 22.6%
zone 1 0.239 1.226 0.430 0.382 0.433 0.477 0.371 1.604 1.362 0.458 0.255
zone 2 0.193 0.961 0.332 0.279 0.334 0.421 0.283 1.332 1.089 0.383 0.199
zone 3 0.144 0.663 0.228 0.183 0.234 0.362 0.193 1.000 0.776 0.283 0.139
zone 4 0.156 0.698 0.242 0.198 0.248 0.370 0.206 1.029 0.811 0.294 0.146
Table 6: One-group cross sections, in units of barn, for neutron capture in the Sing Sing Core fuel zones.
tering on thorium eliminates the spectrum hardening benefit of metallic fuel. In particular, the increase in fission probability for 241 Am from 12% in CAPRA and EA core types to 33% in zone 3 of SSC, points towards a strategy where americium is to be burnt in strongly absorbing fast neutron environments, rather than in moderated targets or diluted cores.
E > 20 MeV 0.025% 0.007% 0.003% 0.001%
Neutronics
The massive introduction of a non-resonant neutron absorber like boron carbide into a reactor core leads to degradation of reactivity coefficients. Coolant voiding, for inTable 5: Characterization of BOL neutron energy spectrum in the stance, increases the probability for fission of even neutron fuel zones of the Sing Sing Core. Note that the contribution of high number nuclides. Further, Doppler feedback vanishes, as energy spallation neutrons (En
20 MeV) to the total neutron flux hardly any neutrons ever reach the resonance region. Taat BOL is less than a promille even in the inner fuel zone. ble 9 displays values of these important kinetic parameters 5
Nuclide 238 U 237 Np 238 Pu 239 Pu 240 Pu 241 Pu 242 Pu 241 Am 243 Am 244 Cm 245 Cm
zone 1 0.034 0.346 1.133 1.714 0.400 2.236 0.282 0.264 0.204 0.455 2.367
zone 2 0.045 0.419 1.187 1.658 0.468 2.026 0.345 0.329 0.258 0.543 2.148
zone 3 0.073 0.591 1.349 1.646 0.636 1.824 0.493 0.490 0.392 0.753 1.962
zone 4 0.077 0.600 1.359 1.663 0.646 1.859 0.501 0.503 0.403 0.763 2.003
Parameter Void worth (keff ) Void worth (ks ) Doppler constant β βeff Neutron life time
zone 3 0.47 0.86 0.73 0.72 0.33 0.34 0.73
zone 4 0.46 0.85 0.72 0.71 0.33 0.33 0.72
CAPRA 0.15 0.64 0.37 0.33 0.12 0.10 0.40
-450 pcm
-400 pcm
320 pcm 0.84 µs
300 pcm 0.42 µs
calculated by explicitly simulating delayed neutron transport with MCNP4C. The comparatively small value of βeff /β µ 0.6 in the Sing Sing Core is due to the low emission energies of delayed neutrons (Edel 0.5 MeV), leading to large probability of absorption in 10 B. A comparison with the kinetic parameters quoted for CAPRA designs [11, 25] shows that the Doppler feedback in case of coolant voiding is very unreliable in the Sing Sing Core, whence subcriticality appears to be a necessary requirement.
for the Pb/Bi cooled Sing Sing Core at BOL. The void worth was calculated by voiding the core (including gaps between sub-assemblies) and upper plena from coolant. The source mode value of the void worth obviously is the physically relevant, but we include the eigenmode void worth in order to enable comparison with deterministic codes. As seen, the advantage of comparatively small void worths (typically less than 1000 pcm, for optimal core designs even negative) of lead/bismuth is partially lost when boron carbide is abundant in the core. A positive value of about +3500 pcm for the present SSC design can be compared with +4500 pcm for JAERI:s sodium cooled ADS design [24]. The Doppler feedback was calculated by exchanging cross sections libraries processed at various temperatures in the core. It was checked that the cross sections used yielded correct temperature behaviour, with k-eigenvalue inversely proportional to temperature and Doppler constants in the range of -500 pcm for a simplified model of the CAPRA core [11]. In the very hard spectrum of the Sing Sing Core, however, resonance capture in fuel nuclides is a rare event, and only a small decrease in reactivity with temperature could be observed. The fraction of delayed neutron emission, β, and the fraction of fissions induced by delayed neutrons, βeff , were
zone 2 0.30 0.78 0.58 0.55 0.20 0.19 0.59
CAPRA (2) +2210 pcm
pared to those of the reference CAPRA oxide core (1) and the high burnup CAPRA core featuring minor actinide targets in the reflector (2). Note the unusually small Doppler constant.
the Sing Sing Core fuel zones.
zone 1 0.22 0.72 0.48 0.43 0.14 0.13 0.50
CAPRA (1) +1560 pcm
Table 9: Kinetic parameters of the Sing Sing Core at BOL, com-
Table 7: One-group cross sections, in units of barn, for fission in
Nuclide 237 Np 238 Pu 240 Pu 242 Pu 241 Am 243 Am 244 Cm
SSC +3500 pcm +4400 pcm -13 pcm 270 pcm 160 pcm 0.61 µs
Thermal hydraulics Several important aspects in the thermal hydraulic analysis of accelerator driven system cores differ from those of the classical FBR. Due to the different shape of the power distribution and especially its change over time, one has to be careful in calculating maximum temperatures. Further, with lead/bismuth as a coolant, the lower thermal conductivity will lead to higher temperature gradients within the coolant channel. The range of coolant flow speeds is also more limited for lead/bismuth than for sodium, as erosion of cladding protective oxide films may occur when maximum velocities exceed 3 m/s [18]. Having implemented the oxygen control and monitoring system developed for the russian lead/bismuth cooled submarine reactors [26], in conjunction with use of corrosion resistant, 12Cr-Si ferritic steels, maximum cladding temperatures should kept below 620 degrees C during extended operation [27]. Detailed three dimensional thermal hydraulics simulations of Sing Sing Core coolant flow were performed with StarCD [28], where the Navier-Stokes equation is solved by finite volume modelling employing k ¶ ε turbulence model. Due to the pronounced turbulent behaviour in near wall regions, wall functions are not applicable to describe the boundary condition between cladding and Pb/Bi coolant. The two-layer model used in the present simulations improves heat friction and heat transfer predictions within the boundary layer, a priori calculating distributions of velocities, temperatures, etc. Heat transfer is implemented through the chemico-thermal enthalpy conservation equations.
EA 0.15 0.69 0.40 0.32 0.12 0.11 0.36
Table 8: Fission to absorption probability ratios for quickly decaying α-emitters and their predecessors in the Sing Sing Core fuel zones, compared to corresponding values of CAPRA and Energy Amplifier designs. The benefit of introducing neutron absorbers to reduce α activity and helium accumulation in fuel pins is clear. 6
·¹¸¯¸»º½¼¿¾±ÀÀmÁ Â/ïÁ ÄżÀ}ÆÄÅÁÈÇ<ÉÊ
Ë Ì Ó»Ô>Õ ÍÏÎ ÐÑÒ
are approximately valid, for the fuel compositions at BOL in zones 1 to 4, when Ts 1000K and χ 110 kW/m. Assuming that nitrogen bonding yields ∆Tgap é 1000 K before gap closure [15], we find that the maximum fuel temperature at BOL is located at about 55 cm height in zone 4 pins closest to the spallation target, being 2100 K for an average core pin power of 44 kW. In nitrogen atmosphere, decomposition can be suppressed up to T µ 2800 K for NpN and PuN [14, 30]. This gives the present core configuration a margin to fuel failure of 700 degrees at BOL. Gap closure at about one percent burnup leads to a temperature decrease of 500 degrees in the pellet. Degradation of bonding conductivity due to release of gaseous fission products will be less severe than in helium bonded pins, especially taking into account that helium produced by α-decaying nuclides is also released in large amounts. Hence, temperature margins will remain larger than 600 degrees until EOL, in spite of the increase in power peaking as burnup proceeds. Maximum power peaking, and hence maximum burnup is therefore limited by constraints to cladding temperatures, rather than fuel constraints (provided that radiation damage has not yet become unacceptable). The 620 degrees C limit then imposes irradiation to be interrupted before radial power peaking exceeds 1.40 in zone 2 pins, or 1.75 in zone 1 pins. The issue of decay heat after shut down is less severe for the Sing Sing Core than for typical FBR cores. The larger pin pitches allows for accommodation of a larger heat deposition in the coolant in loss of flow scenarios. Shut down heat due to delayed neutron induced fission is also less by about one third, due to the high probability of delayed neutron absorption in the absorbers. The emission of decay heat was calculated with MCB to be about six percent of full core power at the instant of shut down, going down to one percent within three hours. In any case, cladding melt will occur before fuel temperatures reach the decomposition limit.
Ö}×zØ ÖzÙ}Ø ÖÅÚÛÖ Ü ØÝÖ Ü}Ü}Ü Ü ×Ü Ü Ù}Þ Ü ÚmÞ Þ}Ø}Þ Þ Ü × ÞÝ×}× Þ}Ùzß Þ>Ú<ß ×zØ}Ù ×Ü Ù
Figure 5: Temperature field in a zone 1 coolant channel, assuming 75 kW of fission power in each fuel pin. Out of this, 7 kW is directly deposited into the coolant by prompt and delayed gamma photons, and by neutrons. An inlet velocity v à 2 á 65m â s was adopted in order to keep maximum coolant velocities below 3 á 0m â s.
Pin by pin Monte Carlo simulations showed that the fraction of fission energy deposited directly in the coolant by prompt and delayed gamma photons and neutrons is about nine percent. This effect could be explicitly modelled in StarCD. Similarly, the impact of non-fissile absorber rod presence may be calculated. The resulting coolant temperature field for 75 kW of fission energy release per pin in zone 1 is displayed in Figure 5, for an inlet coolant velocity of 2.65 m/s. A maximum temperature in the interface between coolant and cladding equal to 586 degrees C is found, which is safely below the 620 degrees C limit. The axial temperature rise in the center of the channel is 270 K. The radial temperature gradient increases from about 50 K at the inlet to 120 K at z = 60 cm (where fuel temperatures are maximal). Note that the comparatively low thermal conductivity of lead/bismuth (a factor of six lower than for sodium) leads to higher radial temperature gradients than in typical FBRchannels for identical geometries and coolant velocities. Thus limits to linear powers in the Sing Sing Core are set to 80 kW per pin in zone 1 and 65 kW in zones 2-4. Knowing the linear power χ ã z ä and temperature profile Ts ã z ä on the outer pellet surface, maximum fuel temperatures Ti on the inner surface of the pellet can be obtained from standard relations between fuel conductivity, linear power and geometry. Using the thermal conductivities of actinide nitrides measured by Suzuki and coworkers [29] (The conductivity of AmN was assumed to be 75 % of PuN conductivity) we find that the expressions Ti ã z äåµ Ti ã z äåµ
Ts ã z äCæ 3 ç 95χ ã z ä kW èªã mK ä è ã mK ä Ts ã z äCæ 4 ç 61χ ã z ä kW ª
(1) (2)
Ti ã z äåµ Ti ã z äåµ
Ts ã z äCæ 5 ç 41χ ã z ä kW èªã mK ä Ts ã z äCæ 5 ç 41χ ã z ä kW ª è ã mK ä
(3) (4)
Burnup Limited mainly by cladding temperature constraints a total core power of 1200 MW (including decay heat) was adopted for burnup simulations with MCB. Irradiation was continued without interruption for 300 days, with a 100 day time step where neutron fluxes, power distributions and transmutation rates were recalculated. In order to obtain a sufficient accuracy (one standard deviation less than four percent error in neutron multiplication), the histories of 40 000 neutrons emerging from the spallation target were sampled at each time step. 700 MHz Pentium III double processors running Linux were used for the calculation, yielding time step simulation times of the order of 24 hours. The resulting evolution of k-eigenvalues, required spallation target power, core flux, power peaking and other quantities of interest are summarized in Table 10. 7
Property k-eigenvalue φnê φê f Target power [MW] Power peaking (zone 1) Average flux [1019 /(m2 ë s)] Flux peaking (zone 1)
BOL 0.972 0.877 0.875 20.5 1.08 3.86 1.97
100 d 0.955 0.870 0.867 36.9 1.29 4.14 2.27
200 d 0.936 0.850 0.847 54.9 1.49 4.45 2.57
ure 6. The core averaged transuranium burnup at end of irradiation is 8.7%, with local burnups ranging from 7.3% in zone 4 to 11.5% in zone 1 and 2. The higher actinide density is displayed separately in Figure 7. Note that the comparatively higher reduction of higher actinide density in zone 1 (12.2 %) to large extent is due to conversion of 241 Am into plutonium via neutron capture and α-decay of quickly decaying curium. As helium accumulation in fuel pins is likely to enhance fuel swelling rates, it would be preferable to burn americium by direct fission rather than by conversion into fissile plutonium. Figures 8 and 9 clearly show the higher densities of these troublesome nuclides appearing in the slower spectrum zones 1 and 2. Consequently, a refined core design where americium is removed from fuel zones one and two would appear advantageous. Regarding the reduced TRU transmutation efficiency imposed by the presence of depleted uranium in the inner fuel zones, we note that this is mainly an issue in zone 1, where the slow spectrum leads to a conversion of 5.5% of the present 238 U into 239 Pu. The corresponding figures for zone 2 is 3.4 % of initial uranium content. As can be seen from figure 6, TRU burnup in zone 1 still reaches 11.5%. The burnability of 10 B is manifested in a 9.3% reduction in boron density in zone 3 and 7.8 % in zone 4. These numbers are larger than the transuranium burnups in each zone, and hence substantiates the proposed role of enriched boron carbide as an efficient burnable poison in a fast spectrum. Several authors have proposed to use moderated neutron spectra for technetium transmutation [31, 9]. However, as self-shielding will significantly reduce the neutron flux in the center of technetium pins positioned in a thermal or epithermal spectrum [32], fast neutron spectra may perform
300 d 0.916 0.802 0.797 75.1 1.70 4.78 2.88
Table 10: Evolution of important ADS parameters during 300 days of uninterrupted burnup, imposing a constant thermal power output of 1200 MW. Note that radial power peaking is considerably smaller than radial flux peaking, and stays below 1.75 even at a k-eigenvalue equal to 0.92!
The decrease in k-eigenvalue is 5600 pcm, accompanied by an increase in required spallation target power with a factor of 3.7. The growth in proton beam current is larger than what naively would be inferred from the loss in keigenvalue, due to the simultaneous decrease of the source neutron efficiency φiì , defined as φiì í
i ¶ 1 Mext Mfiss ¶ 1
(5)
where Mfiss é 1 èªã 1 ¶ keff ä denotes the fundamental mode i neutron multiplication. Mext describes either the multiplication of the spallation source neutrons according to: n Mext é 1 æ k0 æ k0 î k1 æ k0 î k1 î k2 æ î}îÝî
(6)
where k0 is the external source neutron multiplication factor, and ki (i>0) is the generation dependent multiplication factor, or the fission multiplication given by: f Mext é 1 æ ν¯ N f
&
(7)
where ν¯ is the average number of neutrons released in a fission, and N f is the number of fissions induced per spallation neutron. The reasons for the source efficiency being smaller than unity are a larger probability of axial leakage out of the target for spallation neutrons and a relatively large moderation of neutrons within the spallation target, leading to a higher fraction of parasitic capture, than is the case for the average fission neutron. This is a price that is paid for achieving flat power power density, being inevitable if thermal striping of fuel assemblies is to be avoided. Power peakings remain within the limits set by thermohydraulics as discussed above, with neutron fluxes not deviating much from typical FBR-values, except for pins and duct walls closest to the spallation target at EOL. Radiation damage in fuel cladding and ducts is thus not expected to be a problem, since the irradiation time is much shorter than in the typical FBR-cycle. The transuranium density evolution is exemplified in Fig-
ïñðÈòôóRõdöi÷ømùAú
ûü§ýþ
ûü§ýþ ûü§ýþ
û¹üý1þ[ÿ
"!"#%$
Figure 6: TRU density evolution in the Sing Sing Core operating at 1200 MW thermal power. Values are plotted as fractions of initial nuclide density in the individual fuel zones. A 8.7 % TRU burnup is achieved after 300 full power days.
8
KC G&D D DG&I H DG&I L DG&IKJ DG&IE DG&I D DG&H H
MN
¢¦¨§ tvuxwyz{}|K~ Q
¢¦Q¬ ¢¦«ª ¢¦¨¥ ¢¦Q¤ ¢¦©¢
< 8O@5QPR?
')(+*-,12 ')(+*-,3 ')(+*-,10 ')(+*-,/. 457698;:=< >?"@BA E D D F D D
CD D
¢x£K£
Q B o¡ ¤££ ¥£K£
Figure 7: Evolution of higher actinide (americum + curium) den-
Figure 9: Evolution of 238 Pu density in the Sing Sing Core, rela-
sity in the Sing Sing Core. The high reduction rate in zone 1 is mainly due to conversion into plutonium following neutron capture in 241 Am.
tive to its initial concentration in each fuel zone. The high growth rate in zone 1 is due to larger capture to fission cross section ratios for neptunium and americium.
equally well. In the present configuration, 4.8% of the 300 kg technetium present in zone 2 is transmuted during the 300 day irradiation period, corresponding to an effective half life of 12 years. Comparing to the 15 year half life obtained in moderated fast reactor assemblies [31], we conclude that fast flux transmutation of 99 Tc is at least as efficient.
tening power distributions, mitigating reactivity losses, increasing fission to absorption ratios in minor actinides, and lowering helium accumulation in fuel pins. Implementing a four zone differentiation in concentration of enriched boron carbide and depleted uranium we obtained a BOL radial power peaking equal to 1.2 at a k-eigenvalue equal to 0.972. Fission to absorption cross section ratios for 241 Am as high as 0.33 has been achieved in fuel assemblies where 37 % of the fuel pins had been replaced with absorber pins. Operating the core with a constant thermal power of 1200 MW for 300 days, a transuranium burnup of 8.7% for unrecycled TRU discharges from light water reactors was achieved. As a consequence of the boron introduction, helium accumulation in fuel pins due to formation and decay of 242 Cm could be significantly reduced, compared to absorber free core designs like CAPRA and the Energy Amplifier. The concomitant reactivity loss of 640 pcm per percent transuranium burnup in the present core configuration (19 pcm/day) is to be compared to 730 pcm per percent in absorber free cores burning similar TRU compositions. The EOL radial power peaking was 1.70 at a k-eigenvalue equal to 0.916, substantially lower than the corresponding flux peaking. The introduction of nitrogen bonding of the fuel pins allows to maintain a margin to nitride decomposition of more than 1000 degrees as linear powers increase up to 77 kW/m in zone 1 at EOL. Measurements of americium nitride stability in nitrogen atmospheres need to be made, however, to verify the assumption made that AmN decomposes at the same temperature as NpN and PuN. The obvious drawback following upon a massive absorber introduction is a positive void coefficient in conjunction with a minimal Doppler feedback. The choice of lead/bismuth as coolant instead of sodium somewhat mitigates this effect. Reactivity losses are still high enough for
Conclusions The introduction of burnable absorbers into sub-critical systems has been shown to have the multifold benefit of flat-
TgjST Tg&T i Tg&Th Tg&Td Tg&T U
a Ycb UdU \ Ze_XQfR^
)k l+m-n1r k)l+m-ns k)l+m-n1q k)l+mon/p
ST T
WX7Y9Z[=\ ]"^"_%` U T T V T T
Figure 8: Concentration of 242 Cm in the Sing Sing Core, relative to initial 241 Am nuclide density in each fuel zone. The significantly lower accumulation rate in outer fuel zones is directly due to the introduction of enriched boron carbide pins in these parts of the core.
9
the required spallation target power to rise by more than [9] C. Rubbia et al. Fast neutron incineration in the energy a factor of three. Such large availability of proton current amplifier as alternative to geologic storage. Technical margin may be a serious safety threat, if accidental inserReport LHC/97-01 (EET), CERN, 1997. tion of full proton beam power at BOL should occur. These problems, will be addressed in more detailed studies to be [10] F. Venneri. Disposition of nuclear waste using subcritical accelerator driven systems. Technical Reperformed on the Sing Sing Core in the near future. port LA-UR-98-985, Los Alamos National Laboratory, 1998.
Acknowledgements
[11] A. Languille et al. CAPRA core studies - the oxide reference option. In International Conference on FuThis work was financially supported by SKB AB (Swedish ture nuclear systems, GLOBAL 95, page 874. ANS, Nuclear Fuel Ltd), EC (The European Commission), KTC 1995. (Centre for Nuclear Technology) and SI (Swedish Institute). The authors would like to thank C. Broeders, F. Venneri and [12] G. Gastaldo, M. Rome, and J.C. Garnier. CAPRA core W. Maschek for comments on the use of BA in accelerator optimisation by use of 11 B4 C. In International Condriven systems. ference on Future nuclear systems, GLOBAL 95, page 1324. ANS, 1995.
References [1]
[2]
[3]
[4]
[13] H.M. Beaumont et al. CAPRA core studies, high burnup core - conceptual study. In International Conference on Future nuclear systems, GLOBAL 97, page M. Delpech et al. The Am and Cm transmutation, 137. ANS, 1997. physics and feasibility. In International Conference on Future nuclear systems, GLOBAL 99. ANS, 1999. [14] H. Matzke. Science of Advanced LMFBR fuels. NorthHolland, 1986. S.L. Beaman. Actinide recycle in LMFBRs as a waste management alternative. In First international conference on nuclear waste transmutation, page 61. Uni- [15] M. Mignanelli, R. Thetford, and D. Williams. Fuel performance modelling of nitride fuels. In Inversity of Texas, 1980. ternational Conference on Future nuclear systems, GLOBAL 99. ANS, 1999. H. Murata and T. Mukaiyama. Fission reactor studies in view of reactor waste programs. Atomkernenergie[16] H.W. Wiese. Actinide transmutation properties of Kerntechnik, 45:23, 1984. thermal and fast fission reactors including multiple recycling. J. Alloys and Compounds, 271-273:522, D.G. Foster et al. Review of PNL study on transmu1998. tation processing of high level waste. Technical Re-
port LA-UR-74-74, Los Alamos National Laboratory, [17] M. Salvatores, I. Slessarev, and A. Tchistiakov. Anal1974. ysis of nuclear power transmutation potential at equilibrium. Nuclear Science and Engineering, 124:280, [5] T. Takizuka et al. Conceptual design of transmuta1996. tion plant. In Specialist meeting on accelerator driven transmutation technology for radwaste, page 707. LA- [18] N. Novikova, Y. Pashkin, and V. Chekunov. Some 12205-C, Los Alamos National Laboratory, 1991. features of sub-critical blankets cooled with leadbismuth. In International conference on accelera[6] M. Salvatores et al. Long-lived radioactive waste tor driven technologies and applications, ADTTA 99, transmutation and the role of accelerator driven (hy1999. brid) systems. Nuclear Instruments and Methods A, 414:5, 1997. [19] R.E. MacFarlane and D.W. Muir. The NJOY nuclear data processing system, version 91. LA-12740-M, Los [7] B. Carluec and P. Anzieau. Proposal for a gas Alamos National Laboratory, 1994. cooled ADS demonstrator. In International conference on accelerator driven technologies and applica- [20] J.F. Briesmeister. MCNP - A general Monte Carlo tions, ADTTA 99, 1999. N-Particle transport code, version 4C. LA-13709-M, Los Alamos National Laboratory, 2000. [8] H. Takahashi et al. A fast breeder and incinerator assisted by a proton accelerator. In Specialist meeting on [21] M.B. Chadwick et al. Cross section evaluations to 150 accelerator driven transmutation technology for radMeV for accelerator driven systems and implementawaste, page 552. LA-12205-C, Los Alamos National tion in MCNPX. Nuclear Science and Engineering, Laboratory, 1991. 131:293, 1999. 10
[22] J. Cetnar, J. Wallenius, and W. Gudowski. MCB - a continuous energy Monte Carlo Burnup code. In Fifth international information exchange meeting, page 523. OECD/NEA, 1998. [23] J. Cetnar, W. Gudowski, and J. Wallenius. Transmutation calculations with Monte Carlo continuous energy burnup system MCB. In International conference on accelerator driven technologies and applications, ADTTA 99, 1999. [24] T. Takizuka et al. Studies on accelerator driven transmutation systems. In Fifth international information exchange meeting, page 383. OECD/NEA, 1998. [25] H.M. Beaumont et al. Heterogeneous minor actinide recycling in the CAPRA high burnup core with target sub-assemblies. In International Conference on Future nuclear systems, GLOBAL 99. ANS, 1999. [26] B.F. Gromov et al. Use of lead-bismuth in nuclear reactors and accelerator driven systems. Nuclear Engineering and Design, 173:207, 1997. [27] A.E. Rusanov et al. Developing and studying the cladding steels for the fuel elements of the NPIs with heavy coolant. In Heavy liquid metal coolants in nuclear technology. IPPE, 1998. [28] Computational Dynamics Ltd. Methodology Star-CD version 3.10, 1999. [29] Y. Suzuki and Y. Arai. Thermophysical and thermodynamic properties of actinide mononitrides and their solid solutions. J. Alloys and Compounds, 271273:577, 1998. [30] S. Pillon et al. Preliminary assessment of targets and fuels dedicated to the minor actinides transmutation in the frame of the CADRA program. In International Conference on Future nuclear systems, GLOBAL 99. ANS, 1999. [31] J.L. Kloosterman and J.M. Li. Transmutation of Tc-99 and I-129 in fission reactors. Technical Report ECNR-95-002, ECN, 1995. [32] R.J.M. Konings et al. Transmutation of technetium in the Petten high flux reactor. Nuclear Science and Engineering, 128:70, 1998.
11