UNIVERSITY OF CALIFORNIA Los Angeles Configuration of a Molten Chloride Fast Reactor on a Thorium Fuel Cycle to Current Nuclear Fuel Cycle Concerns A digsertation submitted in partial satisfaction of the requirements for the degree Doctor of Philosophy in Engineering by Eric Heinz Ottewitte 1982 The dissertation of Eric Heinz Ottewitte is approved, Moses Greenfield ClE e AHO&;M Steven A, Moszkowski e Gef/id Ce Pomraning 2 z{flfi% William Kastenberg, Commiitee Chair University of Califormia, Los Angeles ii Table of Contents 1.0 Introduction 1ol Overview of Current Nuclear Politics and Technology 1e1e1 Nuclear Fuel Cycle Problems and Their Politics 1a1e2 Technical Solution: Thorium Fuel Cycle 1e¢143 Technical Solution: Fast Halide Reactor 1e1e3e1 Advantages of fluid~fuel reactors 1elelde2 Introduction to the MSFR 1,2 Historical Review of lMolten Salt Activities Te2e1 American Activity 1e1 Development of the USBR 1.2 Development of the MCFR o2 o2 1e242 European Aciivity Poland/Switzerland, led by Miecyzslaw Taube France 261 242 0243 Engla.nd 2¢4 Soviet Union 1.2.3 Summary MSR State-of-the-Art 2.0 Contemporary Conceins Which Affect Choice of an Advanced Concept 21 Nuclear Fuel Cycle Waste Management 2e.1+1 Characterization of Spent Fuel Wastes 2.1.1.1 Gases Zeiele2 Solid fission products 2elelel Actinides 241+144 Unique biological hazard of plutonium 2e1+2 Management of Spent Fuel Wastes 2.2 Non-Proliferation 22«1 Development of US Policy 2e2¢2 Relationship Between Nuclear Electric Fower and Nuclear Weapons Deveiopment iii 2e3 2,4 2e2e3 2.2.4 2e2e5H 2e2e241 2ele242 2e2e243 Extent of the spent fuel problem Question of reprocessing Attitudes on return of Pu Which Reactor Fuel Contains the Least Proliferation Danger? 202.3.1 2424342 Weapons-—-grade material Unique 233, daughter radiation Preignition ' 238 . . . Pu high heat generation Denaturing and other technical fixes Summary The Significance of Stockpiles 2elelal 2e2e4e2 Peacefnl vs, military explosives What a2bout unsafeguarded production reactors ? Fuel Cycle Vulnerability to Diversion 2424541 2e2eDec 2e2e503 2e2e5e4 2e2e5eD 2424546 Once-~through fuel cycle Solid-fuel reprocessing cycle Molten~salt fuel cycle To reprocess or not to reprocess To breed or not to breed Fusion reactor vulnerability Fuel Utilization 2e301 20342 20343 24344 Earth's Resources Use of Taorium Intercomparison of Reactor Concepts with Regard to Fuel Utilization Use of Both Th and U Reserves Strategic Security- 2edel Present Susceptibility 2adelal 20441.2 2edalel 2a4e1e4 Subnational blackmail Sabotage War Summary iv 2ede2 Potential Remedy with an-MSR 2.5 Summary Design Principals for an Advanced Reactor System 2.5«1 HWaste Management Restraints 2e5¢2 Non-Proliferation Restraints 2659¢2 Fuel Utilization Restraints 2¢5«4 Strategic Securiiy Restraints 2.6 Promise and Unigueness of ithe Molten Chloride Fast Reactor on a Thorium Fuel Cycle 24641 Advantages of the llolten State 24642 Advantazes Stemming from the Exclusive Use of Therium 24643 Advantages of Continuous Reprocessirg 2ebed Advantages of a Very Fast Nevtron épectrum 2.6.5 Spinoff Ability to Digest Existing Spent Fuels 24646 Inherent Disadvantages and Limitations ‘o MCFR 30 Concept Design 3.1 Choosing the Reactor Configuration *3.1.1 Maximizing Breeding by Minimizing Neutron Moderation in the Core 3e1elel BG potential 3e1e1,2 Minimizing neutron moderation 3e1e2 Maxrimizing Breeding by Minimizing Leakage 31421 Effect of an inner blanket 3ela3e2 Size of the outer blanket 3a1e2e3 Hoderating the blanket neutron spectrum {0 enhance capture Je1e264 Neutron reflector and damage-shield Je2 3.1.3 COI‘e a.nd le1ed Blanket Design Choice of geometry Choice of number, size,and spacing of iubes Location of tubes Axial blanket and neutron leakage Method of Reactivity Shimming 3e1e5 Summary Guidelines on Reactor Configuration Reactor Thermohydraulics 3e2al 3.2.2 3e243 3,244 Means of Cooling the lloliten Szlt Fuel Out-of-core heat exchanger In-core heat exchanger across tubes In-core direct contact heat exchange Blanket cooling Power Density in an ISFR - 3424244 13424245 3424246 3e2e247 3424248 Primary 3624341 3024302 3624343 Inherent high power density in FKSFRs Realistic range of power densities High neuiron flux levels and radiation damage in fast reactors Impetus for high temperature operatiou Adjusting power density through criticel mass Reactor design power Reactor power iistribution Summary Coolant Velocity Maximizing velocity apd minimizing pumping power Corrosion dependence on velocity Velocities in similar systems Operating Temperatures 3e2e441 3e2e442 3e2e443 3e2eded 3e2e4ded 3424446 3.2.441 Upper limits on fuel salt temperature Hinimum fuel salt temperature Minimum secondary coolant temperature Resirictions on AT across heat exchanger walls Restrictions on temrzrature differences wiinin the primary circuit {4t,t} in other HCFRs Sumnary vi 3.3 Je245 Design of the Primary Circuit Heat Exchanger 1,2,6 e /¢ 3e2e561 3024542 3e24543 Leat transfer coefficient Minimizing fuel inventory in the heat exchanger Frimary fluid volume in heat exchanger and associated plena and piping Primary Circuit Arrangement General guidelines . Location of pumps and heat exchanger Parallel =mbck-o-nels Tube channels in series . Summary 3.2.7 Reactor Thermohydraulics Summary Choosing the Salt Composition 3e3.1 3e342 3343 Introduction 3.341«1 Neutron spectrum 3.341.2 Chemistry 3e3e1e3 Eutectic melting point 3e3e1e4 Heat transfer parameters 3.341e5 Densities 3e3e1e6 Viscosities 3e341e7 Specific heats 3.3.1.8 Thermal conductivities 3e3e149 Summary Choice of Halogen U W Lo W W wwwwf»wwww NNNN:ONI\)NN \DOD—-JO\:H-PUJN"‘ Neutron moderating effect Neutron absorption Transmutation products Melting point Beiling point and vapor pressure Chemical behavior Cost and availability Density Summary Choice of Actinide Chemical States (Stoichiometry) Actinides inherent to an MCFR(Th) Melting and boiling points Chemical behavior Choice between UCLl, and UCl 3 in the core Chemical states in the blanfet vii 3.3¢4 Choice of Carrier Salt Cation 3e3eD 3.4 Structural 3ede 3ede?2 3ede3 3e3ede1 Neutron moderating effect 3e3ede2 Neutron absorption 2.3e4¢e3 Transmutation products 3e3eded Melting point Je3ed4eH Boiling point and vapor pressure 3e3e446 Chemical behavior 3e3edeT Cost and availability 3e3e448 Density 3¢3s449 Heat transfer coefficient 3e3e441C Summary Choice of Relative Proportions of Actinides and Carrier Salt 343541 Neutron physics of the ThCl 4/11013 ratio in the core mixture Je3e 50 2 Density Je3e5e3 Viscosity 3e3¢5¢4 Specific heat 3e3e95¢5 Thermal conductivity 3e3e5.6 Heat transfer coefficient 3¢3¢5«7 Choice of core mix 3¢3e5.8 Choice of blanket mix 3¢3e5.9 Butectic mix for flush salt Materials General Consideration and Criteria 3.4¢1e1 Plant-life econoxzics 304e142 Corrosion, general 3e¢4e1e3 Corrosion by electrochemical attack Jedele4 Corrosion protection 3edeled Thermal and radiation-induced expansion Chemical Reactions in an MCFR 3040 201 3eda2.2 Reactions with fission~produced mutants The effect of UC14 presence in the core salt Candidate Materizls for an MCFR Overview on metals do alloys Graphite Heat Exchanger considerations viii 3.5 304.4 3eded 30406 3edeT Materials for Core/Blanket Interface (Tubes) iaterials for Reactor Vessel Maierial for a Lead Secondary Circuit 3.44641 Steam generator tubes 3e4e642 Remainder of the circuit Material for a Helium Secondary Circuit Physics of MCFR(Th) and Its Fuel Cycle 3501 3a5e2 3e5e3 3454 3e¢5¢D 3e546 Nuclear Models of an MCFR 3454141 Definition of the reactor geometry 3.54142 Use of spherical geometry for neutronics calculations Neutronic Calculational Methods 3.5.241 Spatial mesh 356242 Angular quadrature 349¢2e3 Convergence criteria Neutron Cross Sections 345+341 Bondarenko 26 group Set 3e5¢342 Twenty-six groups, Pl set from ENDF/B-IV 3¢De3el} Effecis of resonarce self-shielding Transmutation Chains and Fquations 3e¢5¢441 Chains for actinide tragfimutations 345¢4e2 Chain for build-up of U-parented radio- activity Principal Physics Metrics 3e5¢5¢1 BG: actual breeding gain 3e5¢5¢2 BGCX: BG extended to zero neutron leakapge 3e5¢5¢3 BGP: BG potentizl zssociated with the core spectrum 305.5.4 ¢ (COI‘E) 3450505 100KeV 3e5s5¢8 Fuel inventory and doubling time Fuel Cycle Modelling 3¢5¢6s1 Fuel cycle model ix 346 305.7 3e5e8 30549 345610 3e5611 345612 2 Equilibrium fuel cycle «3 On the choice of U removal rates ¢ and b Reactor Design: Configuration Trade Studies 3¢5«7e1 Inner blanket study 3eHeTe2 Optimum outer blanket thickness 3¢5aT7«3 Reflector 3e5¢Te4 Core tube material Reactor Design: Controlling the Core >>°U/234y Ratio 3.5¢8.1 Basis for studying the ratio 3e548.2 Effect of the ratio on spectrum, BG,and Pu presence 3eDe8s3 Effect on reactor fissile inventory and doubling time 3¢5¢8.4 Deducing the ratio on equilibrium cycle Reactor Design: Choosing the Core Salt ThCl4/U01 Ratio 3 3.5;9.1 Effect on BG 3e5¢9.2 Effect on reactor fissile inventory and doubling time 2e5¢9e3 Effect on power density and flux leve§§3 234 34549.4 Effect on actinide fission rates and U/“°™y ratio Reactor Design: Choosing the Core Carrier Salt (¥aCl) Content . 3,5¢1041 Effect on BG 3451042 Effect on reactor fissile inventsry and doubling time : 34561043 Effect on pover,iensiiy and flux levels 3¢5410.4 Effect on EBXU 2°3U ratio Mutation Effects 3e5s11e1 --Significance of §3§inides which emit alphas 3.5¢11e2 Significance of U production 3eDe11e3 Impact of fission product concentration upon neutron physics performance Summary Safety & Kinetics 34641 Fhysics of Reactor Safety and Xinetics W6e1a1 Effective delayed neutron fraciion :6.1.,2 Prompt neutron lifetime, 1, 3 2.6, 3e6e1e3 Temperatus coefficient of relativity 3ebe2 Analysis of Normal Operations 3643 3e¢7 Tuel Processing 3eTe1 3eTo2 3073 374 Reactor startup Reactor control Reactor stability and inventory ratio Reactor shutdown Change in physical properties due %o transmutations of Accident Situations Small leakages Loss of flow Structural failure Emergency cooling Comparison of molten salt reactors to others Precipitation-out of cutectic mixtures Boiling~off of sali mixtures Containment Resistance to extermal threat 0 Molten salt combustion support Principal Salt Reprocessing Methods Chemistry of the heavy elements Solvent extraction from aqueous solution Volatility processes Pyrometallurgiczl processing Molten salt elezirolysis Core Salt Processing 3eTe *je 347 Je7 37 3e7 3eTe 3e7 3.7 3eTe NNI\):\JI‘OI\)N o \ooa - OZW\N YR - I\)N . Blanke% Starter 3eTedel Recovery options Continuous removal of mutant gases Removal of non-gaseous fission products Control of the oxygen levels Removal of sulfur impurities Maintaining the chlorine stoichiometry Storing troublesome fission products by using them as carrier salts Comparison to MSBR reprocessing Materials requirements Salt Processing Fuels Starter fuel for the core xi 3.8 3¢9 147e4e2 Preparation of ThCl, for the blanket 4 3¢TeH Fuel Cycle Summary Safeguards 3,81 National Fuel Inventory 3,8,2 In~plant Diversion Potential 34843 Facility Modification 3.8+4 Desired Radioactivity in Transported Breeding Gain 3¢8.5 Radiological Sabotage Threat Access and Haintenance 3+9.1 Compactness of an IMCFR 3¢9.2 Intrinsic Reliability 34943 lNSRE Experience 3+9+4 Comparison of Primary Circuit Configurations 3495 _Reactor Maintance/Replacement Procedure 3496 EReactor Shielding 3,10 Auxiliary Plant 3.11 3+1041 Power Cycle Options 34104141 Intermediate ligquid coolant for steam cycle 34104742 Coolant for gas cycle 3.10,2 Secondary Hardware: Pumps 341043 rilling, Draiqing, and Dump Sysiems 3010s4 Overall Plant Size Ecofiomics Je11a1 British Study Results 3el1lelel - Capital-costs 3e11e142 Fuel cycle costs Je11e143 Summa.ry 13 34.11.2 Transportation Costs 3e11e3 OQutage Penalties 341144 Other Cests 4,0 Summary 4e1 Pu, Proliferation, Safeguards, Security,and Waste Management 4,2 Reactor Design 4.3 Technical and Economic Feasibility 5.0 References xiii 1938 August 1 1957--1961 1962 1962 1962-1963 1963 1963-1973 1968-1969 1973-1975 197 5~Present VITA Born, Cincinnati, Ohioc Chemical Engineering Cooperative Student with the Aircraft Nuclear Propulsian Proj- ect, General Electric Co., Cincinnati, Chio Ch.E. University of Cincinnati Summer Job as Technical Zditor, Mound Laboe ratory, Miamisburg, Ohio Laboratory Assistant, Phoenix Memorial Lab- oratory, Ann Arbor, Michigan M5. Nuclear Zngineering, University of Michigan, Ann Arbor Member Technical Staff, Atomics Internaticnai, Canoga Park, Califormia Atanics International Assignment to the National Neutron Cross Section Center, Brookhaven National Laboratory - Guest Scientist at Swiss Federal Institute for Reactor Research (EIR), Wuerenlingen, Switzer=- land Scientist at Idaho National Engineering Lab.; Instructor with University of Idaho at Idaho Falls xiv ABSTRACT OF THE DISSERTATION Configuration of a Moltien Chloride Fast Reactor on a Thorium Fuel Cycle to Current Nuclear Fuel Cycle Concerns by Eric Heinz Ottewittie Doctor of Philosophy in Engineering University of California, Los Angeles, 1982 Professor William B. Kastenberg, Chair Current concerns about the nuclear fuel cycle seem to center on waste management, non-proliferation, and optimum fuel utilization (including use of thorium). This {thesis attempts to design a fast molten-salt reactor on the thorium fuel cycle to address these concerns and then analyzes its potential performance, The result features 1¢ A simplified easy-to-replace skewed-tube geometry for the core 2. A ve}&-hard neutron spectrum which allows the useful con- sumption of all the actinides (no actinide waste) Je Reduced proliferation risks on the equilbrium cycle compared to conventional fuel cycles because of the absence of car- cinogenic, chemically-separable plutonium and the presence of 232U which gives a tell-iale signal and is hazardous to work with 4e A breeding gain in the neighborhood of 0.3 XV 1.0 INTRODUCTION 1.1 Overview of Current Nuclear Politics and Technology 1elel Nuclear Fuel Cycle Problems and Their Politics In recent years a number of concerns cover the nuclear fuel cycle . nave arisen. A principal irritant is plutonium (Pu) production in the existing light water reactor (LWR) uranium fuel cycle and in the fast breeder (FBR) extension thereto. Concerned individuals fear that the presence of Pu stockpiles mey promote the proliferation of nuclear weapons, covertly or overtly, among nationzl and subnzational (eege terrorist) groups. In respornse to this concern, President Ford announced that the US would no longer consider reprocessing of LWR fuel to be a foregone conclusion. The alternative of course is to stockpile or to bury spent fuels, thereby creating a problem of actinide waste management, That would appear to be less than an optimum ecclogical alternative as much 238 of the urenium mined (mostly U) weculd never be used: mining trans- forms mountain containing 10,000 partis of embedded uranium from natural to machine - processed state for the recovery of T2 parts fuel, Of that, reactors consume 7 - 10 parts; a few paris change into plutonium and the rest currently becomes other -~ actinide waste, If commercizl FBRs ever appear, the associaied fuel cycle must 21so contend with higher actinides such as americium and curium, Heat source rate and alvha decay toxicity may economically prohibit their fabrication into fuel rods, These and other concerns also influenced President Carier to slow dovm the plans for commercial FBRs. 1e1e2 Technical Solution: Thorium Fuel Cycle The thorium fuel cycle solves some of these problems, By starting with a material which is lower in atomic weight (A=232 vse 238) this fuel cycle eventually produces far less alpha-active waste (especially carcinogenic and weapons-grade Pu) than the 238U/ 233 Pu fuel cycle., Though U from the Th fuel cycle makes justi as 232 good a nuclear explosive, ii 1s always accompanied by U whose dauchter decay productis emit 2 highly-penetratinz semma radiation. This makes 233U hazardous to work withe It also signals its location and transport. 1e143 Technical Solution: Molten Salt Fast Reactor 1ele3e1 Advantaczes of fluid=-uel reactors. Fluid-fuel reactors continuously add fuel and renmove fission products and require no fuel refabrication, Molten-salt reactors (MSR) have been more extansively developed than other fluid-fuel power sysiems. They appear to offer advantzges in limiting the proliferation of nuclear explosives: the fuel cycle inventory of weapons~grade maierial outside the reactor is very small and concentrated at the power plants; little stockpiling or shipping of weapons-grade fissile materials occurs. 1eT14362 lLatroduction to the MSFR, Beginning shortly after World War IT but extending into the TO's several laboratories looked at molten sclt fast reactor (MSFR) concepts, Interest generally followed the fortunes of the thermal MSR and so has waned in recent y=ars, Through- out this time, the inherent absence of cladding in an MSR has hinted of a neutron economy sufficient for breeding, A thermal reactor doesn't breed easily; it requires careful design and continuous reprocessing to minimize nonproduciive neutron caztures in fission products, core sfiructural materials, and control poisonse An MSFR is not so sensitive; it has no moderator and little internal siructure. Cl and limited alkali (4ypically Na) are the lightest materials present. Therefore, the neutron spectum remains fast and fission products absorb far fewer neutrons, Numerous workers have already studied the MSFR(U/Pu) as a breeder, fission product burner, and ulira-nigh, fast-flux fest facility with very hard neuiron spectrume This thesis systematically examines the technical bases of !ISFRs and extends MSFR studies into the thorium fuel cycle, MSFR(Th), seeking 2 system which can maintain criticality in steady=-state oneration without accumulating actinides deirimentally. 238U/Pu cycles differ in their neumtron eco- The Th/233U and the nomicse As a result, interest in the thermal molten sal{ breeder reactor (WSBR) has centered on its special usefulness for breeding on the Th/233U cyclee. Interest in the MSFR has focused on exceptional 2 breeding gain (BG) with the 38 U/Pu cycle [USFR(J/Pu)]: up %o Oeb= 0e8e For an MSFR (Th) even with an anticipated reduction of 0.2« O0e3y, BG= 0Oe3~0.5 may be possible, Though not usually emphasized, the very hard specirum and good neutron economy of an MSFR offer the prospect of usefully consuming all actinide products: sooner or later all the transactinides fission, If some capture initizlly, that only (generally) further deforms the nucleus, thereby increasing the probability of fission, Thus the MSFR offers strong potential for resolving current concerms, particularly on a2 thorium fuel cycle, 1e2 Historical Review of Molten Salt Activities The history of the MSR is as old and complex as the history of nuclear power itself, The ups and downs have followed those of the parent technology but the swings have been if anything more violent, 1e241 American Activities 1e2+141 Develoovment of the SR« In 1947 ORNL began a stuly on the dbhysics, chenistry, and engineering o7 traniume= and thorium=bearinz molten fluorides, The potential for very high temperatures and power density interested the aircraft propulsion project. MSR technology first appeared in the open literature in 1957 (Briant and Weinberg[1]). Next, Bettis, et al. [2,3] and Ergen, et ale [4] reported on the Aircraft Reactor Experiment: 2 beryllium-— moderated, thermal reactor fueled with a UF4/NaF/ZrF2 mix and con- tained in Inconel, This reactor successfully operated in 1954 for more than 90,000 kihr without incident, at thermal powers up to 2.5 MW and temperatures as high as 1650°F. This marked the start of an ORNL program to develop 2 thermal Molten Salt Breeder Reactor (IISBR) for economic civilian power [5]. It included 1. Evaluating the most promising designs 2e Pinpointiry specific development problems 34 Developing materials for fuels, containers, and moderators 4+ Developing components, especially pumps, valves, and flanges suitable for extended use with molten salts a%t 130000 5. Developing sfipplémentary chemical processes for recovering valuable compenents (other than urenium) from spent fuel: 6. Developing and demcfistrafing the maintainability éf afi MSR’system. In 1963, Alexander [6] summarized the Cak Ridge development: l. 1-Z: simplicity of the reacher core and the semicantinucus fuel handling apparatus lead to low capital costs and in- creased piant avoiiability. 2. The simplicity and_ccntinuous nature of fissilie and fertile stream processing methods lead to negligible fuel cycle ccsts in cn-site p-ants. 3., The negative temper-ature coefficient of regctivity inherent in the thermal expansicn of the fuel proviaés safety advant- ages over other reactor cancepts. L. The internally-~cooled reactor offers competitive nuclear performance; the externally-cooled reactor, superior per- formance. Thié effortALéd tb the design,iconstructioh, androperatlon-gf the 8~th Molten Sait Reactcr ixperiment (MSRi). Critical operaticn of the MSRE spanned the period from June 1965 to December 1969, dur- ing which the reactor accumulated over 13,000 equivalent full-power hours of operaticn and demenstrated remarkably high levels of opera- bility, availability, and maintainability. -~ IA recent years, Battelie Nofthwest Lafiorat&fy (BWWL) has made critical experiments on MSBR configurations [7], Several United States miversities have studied the chemistry of molten salts [8]. ANL. identified the need [9] for in-pile corrosion testing at high burne ups to clarify thc effect of noble-metal deposition on container metal, Presently all work has stopped, A current problem was that Hastelloy-ll limited the temperatures of the OHRNL MSBR: the additicn of carbides into the grain boundaries keeps He from forming and swelling there,but at high temperatures the carbides disappear into the grains. E. Zeoroski of EPRI feels that the problem with the MSBR was whether the system parts would hold together very long [10]. 142.1.2 Development of the HCFRe Goodman et al [11] proposed a molten~chloride fast reactor (MCFR) in 1952; Scatchard et al [12] reviewed the chemical problers invelved. In 1955, Bulmer et 21, [13] designed and evaluated = 500 MWtk MCFR, exterrnally cooled, At that time the work was secret and only later decléSsified. Chlorides of sodium, magnesium, uranium, and plutonium made up the fuel salt., The blanket was depleted uranium oxide, cocled by sodium, Thelr report contains numerous irade studies, some still usefue. Bulmer chose chlorides over fluorides to limit neutron moder= ation, However, he felt tkat the strong 35Cl(n,p) reaction would require enrichment in 37Cl. M, Taube [14=16] of the Institute of Nuclear Research, Warsaw shared this view, More recently both Taube [17] and Faugeras [18] have mentioned that fluoride salts may 233 work well with U in a fast reactor in spite of the extra moder— ation, The MSFR work then shifted to Argonne Nationsl Laboratory (ANL). E. Se. Bettis (ORNL) feels that [19] ORNL's look at the MCFR was very superficial since it was outside their thermal reactor charter. They meinly criticized the high fuel inventory tied up in the heat exchanger, Bettis thought that M. Taube had a good idea in direct cooling (in-core, no fuel circulation) with boiling Hg [20], but that insufficient quantities of Hg exist. ORNL looked umsuccessfully at molten lead direct cooling: corrosion proved to be the Achilles Heel, In 1967, ANL swmarized [21] the fuel properties and nuclear rerformance of fast reactors fueled with uranium and plutonium tri- chlorides dissolved in chlorides of alkalis and alkaline-earths, Included were the physical and chemical properties of the fuel, and the heat removal and neutronzcs for one homogenecus reactor and two internally=cooled reactors, The optimum core volumezfor 1000 MWe power proved to be 10,000 liters for each types FEach exhibited favorable characteristics of high breeding ratio, large negative temperature coefficients of reactivity, and low fuel=cycle cost. However, the unatiractive characteristics of large plutonium inventory, large volume, complex design, and container material problems indicated the need for a sizable program 4to develop the MCFR(U/Pu). Recently L. Ee licNeese [22] offered the following comments on the !CFR: 1« Resistance of Mo to salt corrosion seems to depend directly on the oxidation states present, as expressed by the "redox" potentials, This corresponds to the directions of the Swiss chemiczl research, 2 Re graphite vs. Mo as a structural material: graphite is 2 step up in technology requirements. It is also subject to radiation damzse. ORNL stopped the engineering on it for MSBR. 3. Re in=core vs, out-of-core cooling: .inwcore offers lower inventory (SI), but hizh structural radiation damege. The extra shielding for out-of-core is inconsequentiale Though the out—of-core (Pu) version oriers a breeding gain of Q.6 ~0s7 to counteract the extra inventory, the economic mea~ sare of interest (a2t least for U/Pu cycles) is approz= imately 312 BG-1. “n summary out—of-core will be easiest, but may be less eccaomic in terms of fuel cosis, FPission product burner concepts are of littie interest: only 1291 separation has been contemplated. 4. The Swiss molten chloride fast reactor designs presume that 2, stream of aboutl one ppm can be continuously directed for reprocessing and returned., For this: a, the chemistry exists (at least for fluorine systems); be the engineering ability exists; ce the smallescale ORNL/MSBR plant was not finished; and de scaling up has not been done and would be a sig- nificant undertakingem— much work to be done, 1242 Buropean Activity 1624241 Pola.nd/Swi'tzerland, Led by Mieczyslaw Taube, M. Taube first, in Poland and later in Switgzerland, looked at a variety of IISR concepts, primarily on the plutonium cycle. Principal studies were on 1« A systém of about four fast CFRs and one MSR "burner® of fission product waste [23]. The overall system con= sumes its own fission product wastes and still has a vositive breeding gaine The burner features a thermal flux trap surrounded by fast reactor molten fuel, The higher the specific power (IW/1 ) in the system, the lower the steady- state fission product concentrztion becomes, 2. Optimization of the breeding retio for an HCTR(U/Pu), theoretically achieving 146 = 148 [241. The 10 IRl 1-1 power density assumed by Dr. Tauve far exceeds the 70 - 80ki 1™ technology of ORNLe Thus, little direct ex- perience applies, Still, L. E. HMcNeese (ORNL) was unable to immediately foresee any intrinsic difficulty wita 10 M7 171 [22]. Presently, the Swiss work has sfopped, as the electorate pushes the laboratory from reactor research towards general energy research and development, 1424242 France, In France, Comme Energe Atomique (CEA) and Electricite de France (EdF) sponsored development of the MSBR as backup to the Superphoenix breeder, EdF emphasized lower powers for near-term feasibility, Their long-term interests included direct-contact in-core molten-lead cooling. CEA pursued higher temperatures for long-term application, Their near-term interests centered on experimental corrosion studies, Fonteray-aux-Roses (CEA) and Pechirney Ugine Kuhlman (PUK) studied several high temperature UMSR comcepts [18,25,26], a Th/ 23%U cycle in fast and thermal fluoride-salt reactors and an MCFR (U/Pu). The MCFR design features in—core cooling (Fig. 1.2-1) and a 7.62 m3 cylindrical core surrounded by a 30 m3 blanket, Table 1.2-I intercompares the main fast and thermal 1Sr. ceneepts of the French. A CZA/FUK subccmpany, SERS, and Carbenne-Lorraine [27] carried out experimental studies in 1974-1976 including 1, Tests of the mechanical performance of ihe materials under tempcrature and radiation 2. Corrcsicn tests in salt at high temperzture 3. Cperaticn of loops at high temperature: one fcr iso- thermal hydrodynamic studies, a non=isothermal cne with small flow for heat exchanse studies, and a third non- isothermal ane for corrosian. In 1977, if the above results were positive, they were to decide on a development program whose cbject would be the canstructicn of a prototype of 25 = 50 Mith, The American and French atomic organizations were also seek— ing to agree on the fields of preparation for a 200 Mie MSBR demo- plant. They were to decide in 1979 on the: construction of this reactor, depending on the progress with high temperature reactors and success of the French Superphoenix 1200 M{(e) Liquid Metal Breeder Reactor (LMFBER)., 10 FUEL REPROCESSING OVERFLOW for FUEL HIONVHOX3 LV 3H z O v - Tm A - -1 = n_ <{ ’ J . o - £ O v = L) z W= u L} ¥ T = e - . L3 - = = W I E 2 = 4 o - m - - =0 a4 S N O Z e COOLANT UClynacl T - s w?flMM1 ARG “fi R S TN RO _l /r/ /j\ TN Mw/V////AH\II _ NSRS /Afi‘n ./n_ _/ i nu,r«hh.,.V,./rw_VVluul/ —— - ;—_—;H‘E o __ ”...Jn. \ E WMWM%.l h//a.f‘u o \'l // T z. 4 A TR /& // _ M/// ////f/ SN _/ L _ J . C‘..%I S R E R Z 3 @x/fli&fi%?/flf//// \ o 75 I 3 \ g w§o Wz v - Z ..hfl._m g - 85 m r O 2l | S F _ 439HVA0X3 LvaH Q O HEAT EXCHANGER FOR BLANKET il i Figure l.2-1 French MCFR Design with In-Care Cooling 11 Table 1.2-I COGMPARISON OF FRENCH MOLTEN SALT REACTORS Fast Thermal CHARACTERISTIC Reactor Reactor Power, Mw Thermal 2050 2250 Electrical 1000 1000 Core specific power w/c 255 22 Efficiency % 49 Ly Fuel, mol % 15 Pu013 71.7 LiF 85 NaCl 16 BeF, 12 Thr 0.3 UF & 4 3 : ja 6 238U 3 Fertile material, mol% 5 013 71.7 LiF 35 NaCl 16 BeF 12 T 0.3 UFh Secondary coolant AlCl3 Na.BF4 Working fluid Dissociating Steam Né)h Gas Core fuel velocity, m/s 2 2.5 Neutron spectrum 0.01 Kev to 10 Mev Thermal Moderator None Graphite Structural material reactor and tubes Molybdenum Alloy Hastelloy=N (Nickel or Iron Added) 1¢2,2,3 England, British workers studied MSRs in 1954 and 1965, lhe MSFR interested them the most and they felt that such a study would complement the U3 MSBR program; A preliminary study of a fast system using the 233U/Th cycle and fluoride salts did not Look encouraginé, sé they refécused on a 238U/Pu cycle and chloride salts. Work on salt chemistry began in 1965; in 1970-1972 the procram extende< “c include other materials aspects. Jerkers at Harwell and Winfrith examined three veriants of £ 25C0 Mie MCFR(238U/Pu) [28-31]: in-core "direct" cooling by molten lead drops, in—core ccoling by blanket salt passing through o tubes,and cut-cf-ccre coo-ing. The direct scheme enccutered too many proclems. The in-ccre cooling cecncept appeared to warrant further study but the fuel inventory did not appear to be as low as first thought; also the high velocities and high pumping pressures presented serious design problems. ‘The corrosion limits and strength of molybdenum or its alloys at reactor temperatures representied a large urknown. With out~of-core cooling, at first the fuel inventory in the reactor circuit was tco high. Hcwever, compzct layouts and higher (but still achievable) heat exchanger volumetric ratings reduced it to within reason. Figure 1.2-2 illustrates the design. The 16 2 -1 13 neutron flux was 3 x 10 ncm ~ s = in the reactor and 3 x 10 An the heat exchanger. 1.2.,2., Soviet Unicn. Very little is known -about Soviet Union . engineering studies. However, extensive salt thermodynamic and physical property studies, evidenced in Section 3.3, suggest a 13 . | Y _“‘r“—] L0 | — ; S Vo — ' i ./‘ ~ :!\_ - 3 }‘ Tt | ~= — » + - 4 . 8 l' x I ; ! | A VaT TR L= Be- } NN~ T SO ‘ H, : ] -~ ‘ ' | p " 1A / \ ~ s ’ - | ==t N\ - e oaim ! S I O b 34 M SCALE | CORE 2 BLANKET ¢ LEAD OUTLET 4 FUEL SALT PUMP—8off It HIGH TEMPERATURE S BLANKET COOLER— 40off RESISTANT LINING 6 BLANKET COOLER PUMP-4o0ff 12 DRAIN LINE 7 ISOLATION VALVE i3 FLOW DISTRIBUTOR 8 LEAD INLET 4 SUPPORT FLANGE Pige 1e2=2 British 6000 Mith MCFR Design; Lead-conled, Oubmof-Core, 14 much larger supporting research program than ir the Western world, A USSR review book on Liquid-salt Nuclear Reactors regards MSRs as reactors of future with compact and promising fuel cycles, They conclude that the developmental focus must be on the extermal fuel cycle, ergo the USSR attention to salt properties, 1e2e3 Summary MSR Statewof-the-Art Both the 2 Mith ARE and the 8 MWth MSRE demonstrated ex-~ tended successful MSR performance but under conditions less strenuous than those considered for a 1000 MiWe or MCFR. The question remains: will a circulating=fuel system hold together long enough to be practical? ore specifically, how will the candidate container materials -« Mo alloys, graphite, or compos= ites with these = hold up at high temperature to corrosion by the variety of mutants and oxidation states which high burnup produces? The answer to such questions will come from in-pile corrosion testing. The French have an on-going experimental program in such engineering studies, but it may be proprietary. Much work needs +to be done in developing and scaling up the reprocessing system. With out~of=core cooling most of the fuel circulates outside the core. This threatens the reactor stability. In- ventory reduction requires compact layouts and high heat exchanger efficiency, 15 2,0 ANALYSIS OF CONTEMPORY CONCERNS ABOUT THE NUCLEAR FUEL CYCLE AND OTHER CONSIDERATIONS AFFECTING CEOICE OF AN ADVANCED CONCEPT 2e¢1 MNuclear Fuel Cycle Waste Management One primary facet of the nuclear waste problem is that reactor operation induces short-and intermediate-lived radiocactivity in materials which had “een stable or only long-lived radiocactive, The 2clution is to altermatively store the activated materials until they decay, or to transmute them back into harmless stable or quasi-stable nuclides, When operation produces nuclides which poison the enviro- ment 2xnd 1ast long, then ecology prefers the added speed of transe mutation, Fuel cycle wastes occur in preparation of the fuel (mill tail- ings), reactor operation (spent fuel and activation of air and water), and reactor decommissioning (activation products of strustural materials), The latter are geperally nonvolatile, bound up in the structural material, and extremely difficult to release to the enviroment, even in accident situations, The mill teilings are in- herently low-ievel, Holdup tanks and stacks with large dilution factors manage the air and water activations, Managzament of the spent fuel poses thz major problem: all others pale in comparison. In the absence of US reprocessing,spent fuel has accumulated recently such as to saturate the utility storage pools.s To prevent loss of nuclear generation, DOE plans to find away-from-reactor storage capacity for 810 MT of spent fuel by 1984 and at least 25,000 NT by 1996, Foreign spent fuel will add to these require=- ments [32]e 16 24141 Characterization of.Spent Fuel Wastes Fissian products and transmuted actinides emit most of the radiatian in spent-fuel, Some of these radioisotcpes exist for long. times, comparable with a human life span. Those with half- life of a few to 50 years are particularly hazardous because they radiate faster than the longer-~lived ones. How lang each element remains in a human or animal depends on 1ts biological-elimination half-life, which may be very long. Some half-lives exceed the life span of our society. A concern then arises that nuclear power might burden the genera?ions to come with accumulated waste, part-cularly the transactinides (Figure 2. 1_1 ) . Relafilve Toxicity . 1 & N w-l" : ] Figure 241w 1 1974 .Bazard from Fuel 1 1 Wastes [33] m*: 4 actinides 134 1™ 17 2.1.1.1 Cases. Spent fuel gases generally include He and T2 from ternary fissicn, Hz, Dz, T2, and He frcm charged-particle-out thresn- cid reacticns, and fissicn procduct gases {Xe, Kr, and Iz). In either a molten salt fuel or in disscluticn of a solid fuel, gases serarate cut easily. The IAEA Code cf Practice for Safe Handling cf Radicnuclides {341 classifies radicnuclices into fcur groups according to thsir racicotoxicily per unit activiiy, primarily upen irhalaticn. £n 40Pk (Th) cculd produce the following nuclides in gaseous form (half-lives in parentheses): 1. Greup I (very high -adictoxiciiy): ncns 5 5 SS 2. Greup II (high radiotoxicity): 301 (3x1C7y, , Lo (8.0 &), ard 231 (20.8 1) : . 1€, 38... 3. Greup IIT (mcderate radictexdicity): ~ F {109.8m), “ Ci _ 8 . 7 . ) (37.2m), E%® (16.72y), kr (76m), B (2.29 1), BT (52.6m), Y% (6.6n), P%e (9.1n) . 9 i \ L. Greup IV {low radictericity): X (12.33 v, 12% ".oxlC7y)) L 1 \ 132 BlL.m (11.9 dj, 7“Xe (5.25 d) o Cf these only 3H, 1291, 85Krm, and 3601 (in order of increasing hazard) have half-lives long enough tc warrant processing: the rest just require short-term holdup before releasé to the atmosphere, Low hazard tritium relatively emits only weak radiation; its maximum allowable concentration is amang the highest far any radicactive material. Furthermore the body excretes it rapidiy end it cannct tioclegicaily ccncentrate in the envircnment, in feood chains 18 or in man. It decays with a 12.3 years half-life to harmless helium and thus poses no Lcng term hazard. It also has applicatien for the fusion prcgram. Nevertheless, it must be manitored and is hard to contain. 8 - SKrm has a similar half-life but is far more toxiz. A4s an inert gas, it is easy to isclate. Both 12912 and 36012 gases might be returned to the reactor as halide salt. There they will most likely transmuie to harmless 3701 and lBOKr. 2.1.1.2 Solid fissicn products. Storage for about a year (TFig. 2.1-2) reduces most fissicn product radioactivity to more manage-— atle amounts. Cf the remaining radioisotcpes the most hazardous are thcse which metzbolize and beccme a part of living crganisms, and decay with half-iives comparable to a human life-span. 2.1.1.3 Actinides. Management of actinide wastes has caused much controversy in recent years [35-49], Their long half-life makes actinides a quasi-permanent burden unless one can transmute or otherwise remove them from the enviroment, Alpha-emitters threaten the most , particularly when ingested. Plutconium is the worst of these: inside the body it seeks out and locates on banes, which makes it carcinogenic. Secticn 2.1.1.4 discusses this further, Most aétinides though usually in zn unwieldy form can also be used for weapons Sectien 2.2 analyzes this. 19 Activity (curie/watt) 10 ) | -14 1 day 10 days 100 days 1 year 10 years 100 years 1000 y. 10 s l | ] ! | ] | T T T T T T 10 10° 10° 1a’ 108 10° 10t time, seconds Fize 2,1=2 Fission Product Decay 20 2.1.1.4 Unique biological hazard of plutonium [50]. Most chemical elements react chemically with biclogical systems, some of them detrimentally. Cthers may be safe in elemental form, but not as compounds. Individual isotopes can also harm living matter by einlt~ ting dangerous radiation. All the isétopes of actinide elements radiate. Plutonium predominates amang these elements because it comprises five quasi-stable isotcpes and constitutes the first significant element above starter uranium on the A scale. The main isotopes of plutonium, in order of decreasing importance are 239Pu, 2L‘OPu, 2hlPu, 238Pu, and 242Pu. This reflects their usual cencentraticns and activity rates. In high-burnup LMFBR fuel, Pu-241 presents the principal short-term hazard (Table 2.1-1), 239by emits 5.15 leV alpha particles which range only 3.6 cin in air and less than L5/f&m in water. In each ionizing collision with air or water meolecules, the alpha leses 35 eV. After about 120,000 collisians, it has lost all of its kinetic energy. It then stops, captures two electrcns, and changes into a peutral helium atom. Noble gas helium does not affect tke human body. Cutside the human body the alpha radiation poses no threat: 5.1 MeV alpha-particles can penetrate the skin from outside only to a depth of hé/Ukm. This is still within the epidermis layer which regenerates very quickly. The short track length of an alpha~particle in living tissue concentrates the energy abscrption in a relatively small volume thereby increasing its affect. Thus to compare toxicity of alpha ¥ N 21 TABLE 2.1=I RADIOACTIVITY IN FUEL 120 DAYS AFTER DISCHARGE FROM A 100-MWD/kg, 1000-MWE FBR Equilibrium ngéag"“ Level Alter Decay . Discha;g_e 2Y2 o Sooline Isotope Mode Half Life |Composition Curies per kg | Kilocuries ' (mM'I?Trhx) of Discharge per M7 9[ ' Isotope Fuel Mix U-232 a 1736y 0.014 2,09 x 10% 0.0 U-237 a |6.754d 3.0 356 0.0 U-238 e |s.5x107y | 909kg 33sx10 | .0 y-239 8 |23.5m 1.2 2 %1077 0.0 Np-236 | @ |22k (57%) | 0.000034 | 5.5 0.0 Np-237 | 8 |2.14x178y| 17 0.706 0.0 Np-2318 | 8 |2.104 0.112 224. 0.0 Np-239 | B |2.35d 179 310, 0.0 Pu-236 | a [2.85y 0.0026 5.04 x 10° 0.0 Pu-238 | @ |89y 246, 1.63 x 107 4.1 Pu-239 | « |24,400 y 40.4 kg 61.38 2.5 Pu-240 | & |6760y 17.2 kg 221. 3.8 Pu-241 | & {13y 2.3 kg 1.13 x 10° 260.0 Pu-242 | @ |379,000y 1.2 kg 3.89 0.0 Pu-243 | B |4.98h 0.05 0.1 0.0 Am-241] @ 453y 176 3,25 x 10° 0.6 Am-242( 8,7 |16k (81%) | 0.064 9.6 0.0 Am-243| « |8000y 96.5 1.85 x 10° 17.9 Am-2e4| 8,y |26m 0.014 2.7 x107° 0.0 Cm-242] a |l162.5d 16.2 2.1 x 10° 34.0 Cm-243] a |35y 0.257 4.2 x 104 0.0 Cm-244| & |18.4y 6.50 7.95 x 10% 0.5 22 with the well-known results for X-rays one must multiply the energy absorptian for alpha-particles by the "Quality Factor", varying from 10 to 50. The most dangerous situation arises ‘when plutaniwn eaters tae human body. Alpha-particles can then penetrate into the tissues to a depth of AS’}Anu Tissues generally are shielded by mémbranes, but these are thirner than l‘}i.m. As water ccmprises mcre than 70% by weight of human tissue, the icnizing penetraticn prucduces hyd-ogen perecxfide tarcugh radio; ysis. Peroxide acts as a very strang poisen inside living tissue: Ithrough a series of chemical reactians it changes the structiurs in the enzymes which catalyze biochemisiry insicde the livingcells Prolongéd irradiaticn by alprha-particles also deforms:nucleic acids, the carriers of genetic informaticn. Cur widest experience concerning biolcgical effects of alpha emitters is with fléRa. Seventy years of handling suggests taat the body burden limit for occupatianal exposure tc it be 0.1 ‘}LCi (O.lupLg). Experiments an dogs and other animal; indicate that Pu produces about five times greater biological effect. Plutonium, like calcium, strantium, radium, and cerium, forms non-saluble phosphates which deposit in the bonas. For these elements the skeletan is the "critical" organ. ,gglufiggigfi binds strengly to the sialprotein on the bane surface; the most abundant component cf bone, callagen, does not bind plutanium to any significant extent. This localizaticn is what makes 239py 23 Q so carcinogenic. The lowest average dose rate of 23’Pu in bones causing death of rats from osteosarcoma was 57 rad (about 15 times less than the lethal dose rate from 9OSr). In contrast thorium and uranium exhibit low radiocactivity. Alpha- . 232U 233 . . . emitters and U also endanger biological systems but they don't "seek bones" like plutonium. This recommends a Th/233U fuel cycle over a 238U/Fu cne. 2.1.2 Management of Spent Fuel Wasties One can either store long—lived spent fuel wastes or transmute them into non-=hazardous products: +those which are stable or near- stable, or which shortly deccy into same (Figure 2.1-3). Studies to date would transmute wast2 in Controlled Thermonuclear Reactors [ 52-55], the Savannah River Plant high-flux production reactors [56,57], power reactors [56=60], accelerators [61], and others [62-64]. Taube [23] and others [hd] have shown that, of the trouble~ 90Sr and 13705 most resist transmitation: some radiation emitters, they have low cross sections (¢ 1 barn) for thermal neutron capture transmutation into 91Zr and 138Ba. However, a thermal flux trap in a fast molten salt reactor could provide sufficiently high flux to reduce the effective half lives from 29 and 30.1 years down to 2 and 9 years, respectively {23]. A solid fuel reactor could not do this because of technical limits on specific power. Sometimes, the same properties which rendsr an isotope hazard- - . . . s 0 ous, also make it attractive as a commercial radiatlon source. 7 Sr and.137Cs are examples. However, commercial demand does not threat- en to exceed supply. 24 Half-l1ife, years >»10%9 -1 10 10 stableLhort nuclides fliving \{.P. F.P. // P \spontareouse / beta cEcay // quasi stable \\ F.P. \\ /, S 9 9 10 10 on o ry 1 O Q 1g transmut long living F.P. 10 10 10 Pigure 2,1=3, Transmutations of Fission Products 25 Actinide "wastes" accumulate when nuclides resiét fission- ing; Figure 2.1-4 indicates the prominent actinides of the Th and U fuel cycles and their fission {,endencies. More explicitly Figure 2.1-5 shows that fission génerally does not predominate over capture until rather high neutron energies are reached. However, &; does exceed o, for 23 L‘U at mich lower energies tha.ri for other actinides. Thus, a verv fast snectrur reactor, esoecially cn the therium fuel evele can uniocuely zvaid accumulatine actinides. According to their median flux energies (Tabie‘ 2.1-I1), fast re- actors must have many more neutrons above the F‘iguré 2¢1=5 crossover points than thermal reactors., MSFRs which minimize neutron mederaticn (elastic scé.tter;ng) will have the hardest neutrcn spectrum of all power reactors. Reactors with highly-enriched fuels (Table 2,1-II) will also exhibit a faster spectrum: en- richment eliminates from the core 23 8U or 23 2Th, which primarily undergo inc].a:.‘bic_ scatter. In considering the reactivity contribution of each nuclide note that the reactivity crossover point in Figure 2.1-5 where N = ’ng:. / O‘c exceeds 1, will occur at even lower energies (roughly where S¢/6. =0.3-0.4). Thus, nuclides usually considered as actinide wastes do not severely detract from criticality or breeding gain potential in a very fast spectrum reactor. 26 232'1}1__,_2331,1_;—-23 Z"u---p--'?'B 5U—-v=-23 6U—=-23 7Np---23 8}:"4:4—--'---23 9Pu fissicn fissicn fissien 238 23%y 2 0py 2lpy L BL2py 203y Pl i fissicn fissim Fig. 2.1-4 Burnup Chains far the <°Th/°>%y ma 2% /23%u Fuel Cycles Table 241-11. Fast Reactor Comparison Fuel Mediun Flux Cycle Reactor -Energy 7 238 U/Pu ANL 1000 MW(e) LIFBR 130 keV AT 1000 MW(e) LIIF3R 180 keV Fast Na ZPPRs 190 keV GCTR Lattice 187 keV 1000 MW (e) GCFR Design 176 keV MCFR Ein-core cooling) 198 keV MCFR {intermal blanket, out-of=core cooling 370 keV Ultra-High Fast Flux 471 keV Molten Chloride Test Reactor Th/233U MCFR (out-ogsgore cooling, high U enriched) 700 keV 27 18.0 1.04 230y Gf/Oc 1 MeV 0.1 100 keV E neutron Figure 2.1-5 Enhanced fissionability of fertile nuclides at higher peutron energies, 28 2.2 Non~proliferation 2e2¢1 Development of US Policy US nuclear energy policy was from the first based on a Keen awareness of the dangerous aspect of nuclear electric power, The Acheson=Lilienthal Revort of 1946 saw a close association between the civilian and military aspecis of nuclear energye. It recommendec international ownership of nuclear explosive materials, The related US proposal to the United Nations failed in part because the Soviet Union would net participate, After that the U5 withdrew into a periocd of secrecy. Eventually we became more relaxed about the development of nuclzar energy for peaceful purposes and abandoned the secrecy. Because we mistakenly thousghi civilian reactor safeguards could be stretched to cover the more danzerous elements in the fuel cycle ~ such as plutonium renrocessing - we allowed plans for the use of plutonium fo go forward unhampered. There is where the damage was done [65]. In 1957 the International Atomic Energy Agency IAEA was estab; lished primarily to monitor the flow of commercial nuclear materials and equipment among member countries, The charter advocates conirol over “excess" quantities of plutonium, but does not restrict their use, The nuclear policymakers of the fifties and the sixties apparently did not realize the security implications of easy access to nuclear explosive material in national stockpiles., The prospect of many nations possessing substantial nuclear explosive materials seemed very far away, A4lso, nuclear weapons were thoughi to be 29 enormously difficult to design and fabricate, The US near-monopoly on the technology, fuels, and equipment for civilian nuclear power activities worldwide seemed to ensure US control of the situation, Fledgling nuclaar power programs seemed to be unrelated to the devel- opment of nuclear weapons., The earlier prescience of the Acheson- Lilienthal group that they had everything to do with it was ignored. There was also some genuine confusion on the technical side. It was once widely thought, for exemple, that you could not make nuclear weapons with plutonium derived from spent power reactor fuel, The Acheson-Lilienthal report seems to have misled many, including the IAEA into believiang you could denature ovlutonium. As 2 result, many of those responsible for protection against military diversion of plutonium were relying upon technological barriers which did not existe On October 28, 1976, President Ford concludéd that "avoidance of proliferation must come before economic interests"e Therefore, we should defer reprocessing until "there is sound feason to conclude that the world community can effectively overcome the associated risks of proliferation.” Two factors contributed to Ford's decision and to the sub- sequent development of President Carter's non=proliferation policy. The first is how the safeguards deficiency impacts international security: we cannot control the plutonium or highly-enriched uranium of other nations should they suddenly decide to abrogate agreements and aprropriate explosive material for weapons. At the same time the extensive zrorth of the nuclear industry, worldiride, is increasing 30 the availability of plutonium, The second factor affecting current policies is the dubious economics of reprocessing and recycle of plutonium in light water reactors, This makes the early introduction of plutonium into international irade as unnecessary as it is dangerous, Presidents Ford and Carter both advocated that we restrict access to dangerous materials, pause in the commitment to plutonium separztion and use, znd search for alternatives to national stocke- piling, Their efforts to persuade our allies and trading partners mct at best, with mixed success, It is hardly surprising that this US policy shift on plutonium produced widespread irritation, alarm, and even the cynical suggesiion fthat the Americans were less inter=- ested in proliferation than in perpetuating a commercial advantage, Both Britain and France have emphasized the security advantages of their reprocessing plans, The British suggest that if reprocessing does not relieve the accumilation of spent fuel, non-nuclear-weapon states will have to develop indigenevus reprocessing facilities: this will certainly pose a serious danger of proliferation, Britain argues that the best way to discourage this development is to confine reprocessing services to the states which already have the bomb [66], Carter subsequently submitted a bill, the Nuclear Non-Proliferation Act of 1977« It would impose stricter rules on US nuclear exports which would block their use for explosives., Opposition to this act argued that whatever relationship exists cannot be reversed and that it is so remote that further controls aren't necessary, Sam }icDowell of DOZ Safeguards and Security has defined non-pro- 3 liferation as prevention of weapons capability in a nonweapons state [67]s The goal of an alternmate fuel cycle would be to deter or reveal diversion such that suitable time would be provided which allowed other countried to negotiate or induce that country to stop,. Zebroski of EPRI has suggested [10] that the primary objective should be to seek improved resistance to diversion from civilian power fuel cycles, The heart of the controversy lies in the access to nuclear explosive materizls which the nuclear eleciric power indusiry pro- vides, Section 2,2,2 discusses +this further, 242.2 Relationshiv Between Nuclear Electric Power and Nuclear Weapons Development The fi.S. generally prodfices plutofiium for weapons purposes in large special purpose reactors and separates it from fhe irradiated fuel in govermment reprocessing plants. However, nuclear power generation electricity also produces plutonium; commercial spent fuel reprocessing planis will separate it, physically creating large stockpiles, Also the same technolozy (and in some cases the same plants) which enrich uranium for fuel can also enrich it further for Wweapons use, The essential point is this: getting the requisite explosive material iz still the most difficult and time-consuming item in the initial vroduction of nuclear weapons,* The operation of civilian nuclear power reactors and plants for fissile fuel geparation and/or enrichment tend to remove this bottleneck, Then if a country has 32 prevared secretly in advance, it can quickly manufacture nuclear warneads once it decides to do sn,. This is now a real threat, not present when the basic internate ional rules for nuclear itrade were formulated twenty years ago. The reason is that the civilian nuclear power industry has grown enorme ouslye. By any reasonable measure (the plutonium production rate or the size of urenium enrichment facilities being wufilized), it exceeds the scale of the worldfs military nuclear orozramse In fact, in most countries the quantities of plutonium is spent reactor fuel, if sep- arzted out and stored, will dwarf any plausible military needs, 2.2.2.1 Txtent of the svent fuel problems Over the next decade Zurope and Japan plan to send 3 G3 (2etric kilotons) of spent fuel tc Winde scale, La Hague reportedly will process about 6 Gg sfient fuel for Japan, the Federal Republic of Germany , Sweden, Switerland, Bel=- gium, Holland, and Austria, This means separating about 75 Mg (metric tons) plutonium, or enough for about 10,000 nuclear weapons. Clearly heavy financial investments ride on the cutcome: when transport charges are included, the Buropean contracts represeni almost three billion dollars in business. The intense internafional competition in nuclear commerce, accompanied by heavy investment and national *¥See the 1975 Encyclopedia Americana article on Huclear Weapons by John Foster, then the Defense Department's RED chief and a former director of the Livermore Laboratory: "the only difficult part of making a fission bomb of some sort is the preparation of a supply of fissionable material of adequate purity; the design of the bomb itselfl is relatively zasy", 33 pride, tends to obscure the proliferation threat, 2.2.2,2 Question of reprocessing. [32,65,66], Current U,S, policy supports the relatively safe fuel cycle activities and discourages the more dangerous ones. A primary U,S,. imperative is to promote come mon rules for internmational nuclear trade, But before formulating rules, much less implementing them, one mmst ascertain just what is dengerous and just how much the spread of nuclear weazons threatens individual countries and world security. Clearly 2 common understand- ing does not yet exist, as witness US opposition to Buropean export sales of pluitoniun reprocessing and uranium enrichment_facilitieso Recent U,S. Administratisns have shunned reprocessing. Many analysts [68] believe this directly increases the risks of both diversion and proliferation, With no reprocessing, the increasing amount of long-cooled spent fuel, and the increasing number of lo- cations with such fuel [69] increase diversion risks, The likeliw hood of "proliferation" increases because concern for a reliable supply stimulates development of independent enrichment capabilities, The capability to develop fissile production now exists in fifty countries, with 10~20 additionzl possible in the next decade [70;71]. A secure fuel supply and reprocessing service through international commerce would economically discourage these countries from devel- oping their own., The economic inhibition is substantial since initial small-scale operations are highly unecomomic relative to ma?ure MAN U= facturing and production capabilities in supplier countries, Section 2.2.5 examines the vulnerability of different fuel cycles to diversion, Pe2e2¢3 Attitudes on return of Pu, Of prime importance to pro- liferation is the question: who gets the bred Pw —the reprocessor or the spent fuel owner? If plutonium eventually displaces uranium as the primary fuel for power reactors, how can indigenous repro- cessing be successfully discouraged outside the nuclear weapon states or a big suppliers club? The goal, long-sought by many, of fuel "independence'" does not equate to simply shifting fuel dependence from the United States to Europe, These commercizl and political pressures, which earlier led to reprocessing ccuid yet force Fu return even witihcut adequate internaticnal rrctecticn against preliferazicn. Presently ncn-Zuratom countries need US permissicn beth to transfer US-supplied spent fuel to cther reprocessoré like “ind— scale and La Hague and to get Pu back. Pu can return cnly "under conditions that...warn the U5 of any diversion well in advange of the Yime" diverted fiaterial could be transformed into weapons. Ingland has pledged the material will be rsturned to ils owners "enly in a form that will reduce the risks" of proliferation. In his report cn the ‘/indscale inquiry, British Judge Parker took refuge in time: 'this matter can be alleviated to some extent oy technical fixes", he wrote. 35 2.2.3 Which Reactor Fuel Contains the Least Proliferaticn Danger? All reactors either produce Pu or 2'33U, or start with fully- enriched 235U. Thus weapans—grade fuel is intrinsically recover- able from all reactor fuel cycles. However, which fuel attracts nationalist and subnational groups the least as a weapon? The answer may depend on unraveling complex interrelated factors, whicn are examined here, 2.2.3.1 Veapons-grade material. For 233U or 235U, weapons grade means isotopic content enrichment in U exceeding 12 or 20%, respeptively. In the broad sense weapcns~grade means any material which can undergo chemical :eparatia to yield Pu or the uranium enrichments menticned above. Thus dilution with any other element including thorium does not lower the weapons grade as chemistry can separate out the U again. At one time it was thought that reactor-grade plutonium (pro- duced in thermal reactors) was not of weapons grade [66]s The U,.S. covernment has now stated unambiguously that they have produced and successfully tested militarily important nuclear weapons, using reactor srade plutonium. Even simpnle weapon designs reliably produce highly powerful (kiloton) explosions. Similarly; the TARA Safeguards Technical Manual now provides the following cuidance: plutonium of any grade, in either metal, oxids or nitrate form can be out in a form suitable for the namufacture of nuclear explosive devices in a2 matter of days to weeks [65]. 36 2.2.3.2 Unique 23 2U Daughter Radiatian. All weapons-~grade fuel can be left in a highly-radiocactive state (by means of fission products). The associated biological hazérd would then discourage its use by subnational weapon makers. Special national labor- atories,could readily clean these up bui ever-present 23 2U inpur- ity in 233U will continue to decay into hazardecus ‘(-emitters (Sec. 3.5.4). In less than a week after high level decontamination, the 233 gamma activity in recovered Th and U becomes so great that fabrication by direct methods can be permitted anly on a scheduled radiation dosage basis [72]. > + 3 3 3 3 £ = 2320 > The characteristic har ‘-gamma signal cf the daughter radiation also reveals the location and transport of this material. 2a2e303" Preigniti_@.‘ The-expiosive yvield of .a f'issicn bomb depends on how long the imploded configuration stays together. Multiplied source neutrons fram spantaneous 21”0Pu fissians cause gearly energy r;alease which produces a premature disassembly. This greatly reduces the' total energy yield of the devics. Cansequently, a military might shun hizh burnup plutonium (high in %%Pu), but a subnational group might still desire it. In pure actinide metal only spaitanecus fissian can con— tribute copious source neutrons. However actinide compounds pro— duce intrinsic preignition-triggers through (o ,n) reactions. Table 2.2-I compares the typical activities from the'se neutron sources. No Th/cycle isctope can match the 2L‘LOPu spantaneocus fissicn rate. 232[] comes closest in half-life (Figure 2.2-1), but its concentration lies magnitudes lower. Thus the neutron 37 Table 242«1 Weavon Preignition Triggers in Pertinent Fuels 232U 231‘U le,opu K half-life (3) 71.7 2.45%10° 6540, & (&/sec/nuclide) 3.1x100° g.ox1™H 3.xc R ol energy (MeV) 543 L8 5.2 SF haif-life (y) 8x10% 2x10%° 1.3x10Mt Ve (n/dis) <2 22 ~2 M5F (n/sec/nuclide) bx107% 2x10™24 . 3x10°17 Typical isotopie 1 1 1 enrichment 1000 5 5 A {n/sec/atom) 6310727 1x10722 1x10"2° fim(n/sec/atom)* \ in F 1x107 18 2%10~17 8x10” in CL x10=20 2x1071 x10~20 in O 2x10~20 1x10~1 x10~20 * Neutron yields per 108 ‘A 5 MeV-alphas are 1200, 11, and 7 for targets of F,Cl, and @, respectively according to Roberts [73] 38 -LIFE LYEARS) SPONTANEQUS FISSION HALF 4 35 3% 37 ¥ 39 40 4 R Fige 2.2-1 Spantanecus fission half-life of even—even nuclides versus Zz/A. 39 production rate from 232U and.23hU in with 233U falls many magnitudes below that from 2l“OPu in Pu. With high C1,F, or O concentration (as in salt or oxide compounds) the (o,n ) neutron rate (Table 2.2-3) eguals or exceeds the spontanecus fissior. neutron rate of metallic Pu (7x 10_20n/s Pu atcm, 20% 2l‘OPu). F is much stronger than Ci cr O, in this regard. This should appreciably reduce the ex— 233 rlcsive thr=at frem U weapons in fluoride, chlcride, or oxide chemical form. The («,n) neutrons will also present a streng actlvation hazard except when heavily shielded. 2.0.0 .0 238?u hign heat cere-ation. 238Pu nas Iound cansiderabie applicaticn as a radioisotope heat source. Because its high heat generation rate will melt practical sizes of the metal, the usual form is PuC,. For a large enocugh uncocled mass, even the oxide can melt. Cxide sources also preduce neutrons through 180 (e\n ) reactians. Together, the neutrons and the heat make plutonium high in 22¢ Pu content cumbersome to work with. . 238 Two principal rcutes produce ~7 Pu, 239y (n,2n) 238y, (2.2-1) 235 (n, ¥) 3% (n,y) P (n,¥) 2B (2.2-2) Route (2.2;1) enhances productions for very fast plutonium reactors; Route (2.2-2), for high enriched 233y or %%y reacters. In high 238 burnup U-fuelled LiRs, Pu isotopic content reaches 1%. In a Th/233U or a high 23%)—enriched reactor, with little or no 2% 238 present Pu should dominate any plutonium separated out. 40 2424345 Denaturing and other technical fixes [32], One technical fiz, the Civex process, has received a fair amount of international attention, It returns pluftonium to the customer in the form of fabricated fuel containing enough radioactive products 4o preclude easy chemical exfraction of bomb material, However, within a few years after leaving the reactor, radioactivity decay forfeits this protection, Since most spent fuel reprocessed in this century will have cooled for even longer than that, the Civex process cannot contribute to the solution of the problems we must worry about now,. Civex also must face other problems, Industry will resist irradiated fuels becafise ther increase occupational and public health exposures, Likewise the owners of the material, especially foreign countries, are not likely to accept the dictates of the reprocessor on the form in which plutonium is to be returned to them, This would also argue against indusitrial recycle of solide 233 form “°°U daughters). The simplest answer to denaturing the BG export from an MCFR will be to take it directly from the core; fission products, Y - emitters.and all. 2024346 Summarye It remains yet to say if there exist any signif- icant proliferation differences among any of the fuels, Ve can categorize them as follows: 1e U (low-enrichment in 230y or 2330) 2a Pu 3, 2%y 41 2334 4. Fu, 235U, and 233U are all weapons-grade, [Fuelling reactors With sub-weapons—grade materials does little to limit proliferation: the reactor will soon produce weapons—grade fuel from the diluert, 238 i.e, Pu from U. Thus, no reactor fuel cycle can avoid weapons— grade material, Also, any irradiated 238Uécontaining fuel consti=- tutes a terrorist threat from TNT-exploSion aerosol-dispersal of carcinogenic Pue All but very fresh 233U exhibits a deadly penetrating daughter radiavicne This hazard discourages the assembly and holding of , 233 U vomb. Hven if onme chamically separates out the daughters, they soon build in againe Thus, in summary,no "proliferation~proof™ fuel exists,but the Th/233U cycle does offer some advantages over the U/Pu one: the presence of 232U daughter radiation and the absence of carcin=- ogenic Pu,. 26244 The Sicnificance of Stockpiles 2e2ele1 Peaceful vs, military explosives, The Nonprolifleration Treaty forbids non-nuclear-weapon states to acquire nuclear exploe- sive devices, whether labelled military or peaceful (they blow up the same way). Peaceful explosive services could still be provided under sirict international physical control, The same rules should apply to something very similar, a dis- assembled bomb, That being so, what about the nuclear explosive material itself? The only answer consistent with the Treaty's prohibition on nuclear explosive devices is = don't label it "peace=- 42 ful" until you prove you can safeguard it [65]. 2.2.4.2 What about unsafeguarded production reactors? [65]s It is contended for example, that no country choosing to build nuclear weapons would turn to its civilian power reactors for the requisite explosive materials; to divert material in this way would risk detec- tion by the IAEA inspectors, and in addition would provide too poor a grade of plutonium to interest weaponeers, Under this self-serv- ing theory, if weapons material is wanted, a special~purpose unsafe= guarded reactor would be built, I% is possible at the moment to do this legally in countries noi pariy to the Nonproliferation Treaty and therefore not subject to inspection of all its indigenous nuclear facilities. This underlines fthe need to exterd the req;iggments of the freaty to nmonsignatory nations by conditioning nficlear trade on acceptance of internafional agreemenfig and inspection on all nuclear activities within importing countries. There is increasing pressure to do this, and a recent bill be- fore the Congress would make this a condition for U.S. nuclear ex~ portse Even if legal, however, the construcfion of a special purpose plutonium production reactor signals a country's intention to build bombs and, in the present climate, risks premeture interception of its attempt to obtain explosive materizl for nuclear weaponse. To avoid interference, a country mighi stockpile separated plutonium from spent power plant fuel openly and legally. A dew- fense establishment can design and fabricate a bomb in privacy; the illegel activity is then confined to a swift, almost one- 43 sten process: appropriation from its storage place of the nec- essary plutonium, fabrication, and insertion into the waiting bomb, It is surely the quickest, cheapest, and least risky route to nuc; lear weapons. So long as individual nations are permitted to kggg nuclear explosive stockpiles they are in effect, in possession of an optiorn to make nuclear weapons almost literally overnight, In other words, from the moment spent reactor fuel is trans- lated into sevarated plutonium and stored, the element of "timely" warning, on which our present safeguards system has been relying, evaporates, The same is true, of course, for stockpiles of highly enriched uranium, It is important to understand that so far as safeguards are concerned a stock of nuclear erxplosive material is ailot more like a2 bomb than it is like a reactors No one would dream of suggest- ing that nuclear explosive devices, regardiess of how labelled, should be exported under international safeguards, The Nonprolif= eration Treaty settled once and for all the notion that nuclear explosives came in two categories - pilitary and peaceful, Under the treaty no such distinction is permitted. Yet strip away the electronics and the conventional high explosives and label the plutonium as intended for peaceful purposes and many nmuclear spokes- man, at home and abroad, will tell you that if subject to occasional inspections it is a perfectly safe proposition: just like safe= guarding power reactors [65]e The worldwide opportunities for diversion are clearly dominated by the existence of large volumes of spent fuel stored in many hun- dreds of locations, and especially long = cooled spent fuel which is not returned to supplier countries [10], The crux of the proliferation issue is not to completeliy avoid Pus. Due to advanced technology, any country can produce it without resoriing to nuclear power cycles, Rather it is to aveid having Pu lie around in larze quantities where diversion is less detectable, In an MCFR(Pu), Pu is always in solution, except for that fuel which is required to start a new reactor. Reprocessing is done within the plant boundaries continously and the Pu then fed back into the system, 45 24245 Fuel Cycle Vulnerability to Diversion The opportunity for national or subnational diversion of weapons—grade material will depend on the size of the fuel cycle inventorye Transport outside secure boundaries and decentraliz= ation of the inventory enhances the potential for subnational diversion: while under transport the material will be in a more easy-to-handle form and probably under less security; decentrali- zation also implies less security. 2.2451 Once=throush fuel cycle. Figure 2,2-1 diagrams the trail of weapons-grade material in a once-through fuel cycle, assuming that the initial fuel elements are low=enrichment U. After gener- ation in the reactor, it passes through about three storage/queue depotss Possibly one more storage depot labelled 'permanent" may appear later, once it is fixed what that will be, Only one out-of plant transfer may be necessarye The inventory in the long term storage depots will grow with time, eventually becoming very large. Although the Pu there is not separated out, a national laboratory could easily do so, One should also note the presence of the enrichment plant: though it doesn't produce weapons grade-material for this fuel cycle, it could, 46 Operations Storage JE A U enrichment plant | Reactor Figure 2,2=1 Once~through Fuel Cycle Legend e e e e o e =2 - = = = intraplant transfer interplant transfer 224542 Solid fuel reprocessing cycle. Im solid-fuel recycle (Fig, 2.2-2) we note that weapons-grade fuel will accumulate in three plants and maybe six storage/queue depots. Four interplant transfer operations occur, counting BG export, It is not hard to envision the equivalent of ten cores being tied up in the various queues,. plantis,. storags,. and transports, 2e2+5.3 Molten-salt fuel cycle, Here almosi all weapons-grade material stays (Fig. 242-3) within the plant boundaries, With out- of-core cooling, the equivaleni of 4-~5 cores inventory is possible, but they remain within the walls and mostly in the primary circuit, Only BG export requires out-ofeplant transport. 47 Operations Reactor , Storage Piping and Heat Exchancger Reprocessing Stockpile of 233U BG for Other Plant Startup gggort of UC].3 - = = - intra plant transfer ======= interplant transfer Fige 242=3 Molten=Salt Fuel Cycle 49 232.5.4 To reprocess or not 1o reprocess, The once-~through fuel cycle, starting with low~enrichment U, seems to display fairly low vulnerability to subnational diversion, but creates a cumulating stockpile available for national use, It also creates a long~term waste management problem and poorly uses our resources, The molien. salt fuel cycle clearly seems to be less vulnerable than the solid-fuel reprocessing cycle. On the present rough scale, the molten salt fuel cycle must be rzted as comparable to the "once=through" in vulnerability, The question now occurs = would you export an MSR with its in-plant fuel cycle capability to a non-weapons state? Or, a different question: which would you export, the "once-through" or the "molten" fuel cycle? Assuming that you could alter the MCFR to make Pu from 238U but that also the non-weapons state keeps its spent fuel from a2 once-through cycle (see section 2.2,2,3), either cycle is bad for national proliferation, This seems 1o say forget about exporting unless you can maintain control: IAEA, US, or guuer, On the domestic scene, the reduced market would definitely favor the "molten" cycle as it solves more problems and makes more fuel economically available, 50 2e2e5¢5 To breed or not to breed, Assuming that the potential 1o breed excess plutonium or highly=-enriched 233U exists, one must next ask - should we? The alternative is fto use the excess neutrons to achieve higher fission product levels, a smaller critical mass, or some other benefit, Breeding obviously provides initial fuel for starting up other reactors, The alternative is to start [CFRs up with Pu from spent LWR fuel, or highly-enriched 235U. But either of the latter two fuels would make it easier to manufacture weapons, Also they would cause the thorium fuel cycle in this reactor to depend on -=he U/Pu fuel cycls, With a system of thermal MSRs the export 233U BG could be dilubed below weapons grade, but 238U diluent fiould evene Th diluent is chemically separable and tuzlly mean Pu production, Based on these considerations, this study seeks a breeding MCFR. Only the first genmeration of MCFR(Th) reactors should then nge@ ?35U or Pu for startun. 2e2e5¢5 Fusion reactor vulnerability. Lest anyone think that fusion reactors are the panacea, note that the current literature is full of fissile breeding ideas; producing weaponsw-grade 233U or Pu, In the US, current thinking regards such a hybrid as neeessary to a near-term economical reactor, In the USSR it is their prime goal, 51 2.3 el Utilization 2.3.1 Barth's Resources 209 (Z = 83) nature offers on earth only the chemical Beyond elements thorium and uranium (Z = 90 and 92). Only these two materials, in present, or feasibly-transmuted form, can sustain a neutron=induced fission chain reaction. Only 0.72% of U is 235U. The enrichment process loses some of that. Present reactors, without recycle, consume just 10% of the 235U before assizning the rest to semipermanent siorage, Thus we presently waste 9Ge 94% of the uranium mined, an affront io the en- vironment. If, instead, the reactor could usefully consume ail of the actinide, considerably less eyesore and mining expense would result. Bconomics might then justify the retrieval of extensive low=grade ores. At a time of energy shortage the present policy seems foolhardy except that 1t reflects the restrictions posed by economics, proliferation concern, and spent fuel waste management. These must first be overcome. Thorium abounds three times as much as uranium in the earth's crust, and offers that much more nuclear fuel energy. The avail- abie uranium ores increase greatly if one includes not only the classical ores { »1000. ppm uranium) but the very abundant granites with 80 ppm uranium and thorium, GCranite is the main constituent of the earth's crust (up to 20 km deep). Extraction from these ores could become economical if fission reactors fully used all the Th and U present. Together with fusion reactors, such a fuel cycle would then open up all the principal world resources of nuclear fuel. 52 2¢342 Use of Thorium Nature has significantly inhibited the use of Th by not pro- viding a thorium isotope which can sustain a neuiron chain reaction, Despite some unique advantages of a Th/233U cycle over a U/Pu one; Th use must still prove its economicse A reactor which yields high BG in the Th cycle would help a lot. High melting points hinder reprocessing of 2 solid Th fuel: Tho, melts at 3050°C; Th metal, at 1750°C. Breeding is very mar- ginal, except with a fast reactor. Reducing non-productive neutron captures (ee.ge to fission products, core structural materials, and control poisons) vecomes crucial, That suggests fluid fuel reactors: they can continuously remove fission products while adding fuel on- ling, and some variants require no core structure. 2e¢3¢3 Intercomparison of Reactor Concepts with Regard to Fuel Utilization Much attention currently centers on optimum uranium usage by candidate reactor concepts, The parameter of Megawatt days of electric energy produced per mined metriq ton of uranium (MWde/ mined HTU) provides a suitable yardsticke Note that the numerator includes energy production from thorium while the denominator ignores thorium consumption, Consider as a benchmark that all the uranium were somehow (ideally) consumed, Then without any energy production from Th, the rating would be 380,000 MWide/mined MTU, The advanced concepts in Table 2,3-I, currently under study in the US, do not come anywhere near this idealistic figure; not even when including energy production from The 23 Table 2¢3=] SOME 30-YEAR FUEL CYCLE PERFORMANCE COMPARISONS [74] Energy Production Reactor Fuel Cycle MWde/Mined MTU Uranium Cxcles: once~through uranium 1860 uranium recycle only, 2270 Pu storage . uranium and single Pu 2650 pags recycle uranium and Pu recycle 2800 Thorium Cycles: 93 w/o 235U, recycle of 3420 235y ana 233y PURs 3 w/o 237U, recycle of 2900 Pu and 233U 3 w/o 235U, crossed 2870 progeny Trecycle Spectrum Shift (Thorium): 93 w/o 235U, recycle of 5550 235U ard 233U 235 3 w/o U23§ecycle of 3450 Pu and U LWBR Thorium Cycle 6800 CANDU Uranium Cycle 5800 Thorium Cycle 10440 HTGR Thorium Cycle: once-through 2620 233U recycle 4240 ¥ 3w/o 235U as U0, core; feed of output Pu into ThO2 with recycle of Pu; feed of 33y %ggo uo, (natural) with recycle of Pu into ThO, and recycle of U into U0, >4 Zebroski [10] concludes that the evolutionary extension of LWR burnup offers the best near-ferm hope for improving usage, However, higher burnup will also produce greater statistical vari- ations in fuel failure, which will limit these extensions, On the long term, an MCFR should come much closer or exceed the ideal limit because all actinides are consumed and it requires only thor= ium feed (except for startup fuel on early reactors), 2e3e4 Use of Both Th and U Reserves One might desire that any new reactor concept be able to operate on both the Th and U cycle, so as to evenitually allow use of all our reserves, MCFR siudies have already indicated large potential breeding gaint on the U/Pu cycle, providing Pu is accept- able, In that event, one could also supply spent U fuel as blanket feed, thereby solving that waste management problem, 95 2.4 Strategic Security Although it doesn't normally get public attention, still strategic security has its place on the 1list of contemporary concerns due to current events: Te 2e 3e de It a fight The lonz-and short-term goals of the USSR Other political conflicts which could lead to world war (viz: Israel-Arab, China vs,..Taiwan or USSR, Arab-Arsd, etc.) The growing=-pains and frustration or have-nots and emerging nations The increasing pelitical turmoil and unrest within America stemming from problems of cities, alienation of extremist groups, and, again, have-nots, might be successfully argued that fat men dofi't look for - "yon Cassius hath a lean and hungry look", Then insecuritiy traces to the failure~to~appear of cheap nuclear electricity, coupled with the disappearance of the cheap oil~fired energy source, Thus the world still hungers for a cheap abundani energy source %o solve its WQES, obvious Wetd like to think the MCFR is it, Until that becomes more though, it would be well {0 secure our US energy supply irom all threat, 2.4.1 Present Susceptibility At present, US energy supplies, key to our might and resource- fullness, are all above board (ground level) for everyone to see and aim at. This includes dams, oil tankers and depots, fossil- fuelled plants, and nuclear reactors. Actually only the containe ment building of the latter protrudes above ground but that still leaves it vulnerable. 2eliel.l Subnational Blackmaile Should a terrorist group seek a pmeans of national blackmail, the threat of destruction to a dam or a nuclear plant might seem credible. Given sufficient explosire a dam would seem to be a better target, both in size and certainty cf wreaking financial havoc. L¥R destruction would likely cause more perscnal injury. A fossil = fuelled plant destruction would not necessarily threaten public lives., Bombing a city or a poisonous gas {like ohlorine) plant might pose the next largest threats. 2.:.1.2 Sabotage. In time of war or unrest, sabotage to our sys— tem could rear its ugly head. Surprise attacks might often occur. In addition to the above targets, one might then add many more ncn- nuclear possibilities such as poisoning water systems or destroying any energy supply. For sheer havoc, though, destroying all water, supply.to LWR cores as well as their containment would do well. o1 2.4.1.3 War. In event of war, sabotage could be one front of the attack. In addition planes and missiles might be directed at nu- clear plants in hopes of accomplishing what the aggressor didn't want to or couldn't send an atomic bomb to do: perhaps an atomic blast would have destroyed a valuable industrial complex whereas an LWR core release will simply harm and demoralize a lot of people. 2.L.2 Potential Remedy with an MSR The dominant vulnerability of the above energy sources would seem to be the presence of the potential hazard above or near the earth's surface. Moving the nazard lower mitigates the danger. Dams can't go underground. Nuclear power plants can, but it costs a lot. An MSR uniquely offers (1) Continucus removal of the principal hazard from the sys- tem: +the gaseous and volatile fission products. One can store these far underground in a recessed tank for security. (2) Ability to transfer the whole core further underground at a moment's notice to a (possibly natural convection-) cooled holding tank. Thus the MSR may allow the security advantage of underground siting without all the costs and other difficulties. 58 2elie Summar Analysis suggests that an LWR may be a likely target for atiack or threat of attack. The goals of such an attack would be (1) Destruction of the energy production capability (2) Release of gasecus and volatile fission products into the air. This threai to energy and peopie probably exceeds that from a dam which tends to be located farther from people. Also water must follow the lay of the land, which authorities could observe and some - what predict. Transport of fission product gases, they could not very well . At the same time, nuclear plants can be the toughest nuts to crack as they are more compact and self-contained. With a little more hard- ening, such as an MSR allows, they could be America's energy ace-in- the hole and basis for survival. On the long term the MCFR and all other novel concepts should be explored and re-explored for their ability to provide that panacea of a cheap abundant energy source = a far betier means of dealing with yon hungry Cassius than any protective shells one can devise. 59 2.5 Summary Design Principals for an Advanced System 2e9e1 Waste Management Restraints Spent fuel contairs several products whose activity remains high after a years storage, Of these, the alpha~emitters can extenw sively damage intermals when ingested, Plutonium presents a special hazard because it locates on bones: the whole skeleton of man be- comes the critical organe. Although our society has learned how to handle other toxic industrial wastes (arsemic, mercury, chlorine), control is not perfect and accidents have occurred. Thus, we prefer to avoid plutonium production if that does not severely curiail our energy production, A reactor whicn usefully consumes all 2% 197 ) Neutron Enerzy (eV) Fig. 3s 1=3 -Thorium Capture Cross Sectlon ORNL studied the effect of a graphite moderator.in the blanket [53]. ‘Although the fertile capture rate did increse, reducing the necessary blanket thickness, the overall BG decreased (Fig. 3.1-~4) due to simultaneous increase in parasitic ~ neutron capture. Another way of moderating the blanket is tc use a (lighter) 73 Fig. 3.1<4 Influence of a - BG Graphite Moderator in the Zlaenxet upcen £a [13] Thickness of Grapnitie !oderator Section, cnm 3.1.2.L rsuiren raflocter amd domase-snicid. A neulron rellacter cutside the reactor vessel will cbviously reduce L. Few of the re- flected neutrcens will make it through the outer blanket to return to the core. A thermalizing reflectcr would therefore have little effect upcn the ccre spectrum, particularly if the cuter blanket regicn is fairly thick (in mean free paths). Such a reflector weuid enhance neutren capiurs in the blanket (put in carrier sait as well as in fartile). It would alsc decrease Py scmewnat. The net changs in ZG wculd depend cr ncw the ccempeting capture cross sections change as neutron energy dscreases. This 1is expressed by the crcss secticn ratio Q. a O (£) + b Opy (B) 6 (€) 14 where Ha = nhalide: c¢hlorine or fluorine Alk alkali or alkaline earth asb coefficients dependent upon the molar ratios, For a typical 65/35 ThClA/NaCl mix, Table 3,1-I shows the results, At very fast and thermal neuiron emergies Cl parasitic capiure vre- dominates., In the broad intermediate range of 20 ev to 1 lieV Th fer- tile capture is greater, Thus we ovrefer that neutrons in a ThCl 4 blanket slow down, but not all the way to eV energies, This recommends an inelastic or a heavy scatterer more than a light elastic one for the reflector material, This would also recommend a high actinide-to- carrier salt ratio in the blznket, Choice of beryllium as reflector material would: enhance the outer blanket breeding through scattering and (n,2n) reactions. However, beryllium is toxic, relatively scarce and quite light. Graphite, in contrast, costs little. Cu and Ni are well known for their ability to reflect fast neutrons, Fe and Pb reflect almost as well and cost less. All four qualify as inelastic or heavy=elastic acatters. However, reflector studies in Section 3.5.6.3 indicate that with a blanket thickness of 200 cm, the choice of reflector material matters little. Then the low cost and low activation of graphite should predominate.. Borrowing a page from fusion test reactor design, one could reduce damsge to the vessel with an inper liner. Neutrons striking it change direction and lose energy; they are then less 15 Table 3.1 Ratio of Cl to Th Neutron Abscorption in ThClA O abs (b) Group Energy Range Cl Th gngTh 1 6.5-10.5 MeV 0,215 0.012 7.6 2 L.O=6.5 0. 146 0.02 33.1 3 24 5mly 0.057 0.04 6, L lel=2.5 0.020 0,08 1.13 5 OuB=1.4 0.0063 0.14 0e20 6 Oulp=Ge8 0.0021 0.17 T.056 7 0e2=0.4 0.0009 0.19 0.022 8 0. 1~0.2 0.0011 0.27 0,018 9 L6.5-100 keV 0,0031 0,42 0.034 10 21.,5=46.5 0.0103 Ov56 0.084 11 10.0=-21.5 0.0109 Q.75 0,066 12 L e65=10 0.0189 1.35 0.064 13 2.15=1.65 0.0194 2,10 0.042 14 1.0=2.15 0.0221 3.30 0.030 15 L65=1000 eV 0.1670 5,0 0.152 16 215-465 0.3730 11. 0.154 17 100=-215 0.1557 19. 0.037 18 16.45-100 0.3660 28. 0.059 19 21.515.5 0.7081 L47. 0.068 20 10.0=21.5 1.1625 12. 04,0 21 Le65=10 | 1.795 Q.46 1747 22 2.15=4.65 2.771 0.67 18.8 23 1,0-2,15 L4231 0.99 19.4 21 0.465=1,0 6.27L Luhb 19.6 25 04215=0.465 9.33 2.11 20.1 26 0.0252 eV 28.69 7.56 17 .2 likely to damage the reactor vessel. This damage shield should be cheap, suggesting graphite, and might be in partitions for easy replacement. One could also combine damage shield, vessel, and reflector into one thick-walled graphite vessel. cost the least to make or replace. would also be the easiest to dispose of. 16 This might Due to low activation it 3.1.3 Core and Blanket Design 3.1.3.,1 Choice of geometry. Spherical geometry offers 1. Minimum neutron leakage in a critical reactor (thereby maximum BG) 2. Minimum chielding and plant size 3., Minimum critical mass L. Simule theoretical computation The first two factors directly affect reactor design. Minimum critical mass actually matters littie: the fuel utilization issue pertains more to total cycle inventory while proliferation deais with the availability of attractive weapons~grade or car- cinogenic material., The fourth factor affects just the methods used for survey calculations. Three main designs come to mind in spherical gecmetry. The first copies the thermal MSBR: . core and blanket materials, i.e. fissile and fertile, flowing together homogenecusly. The second copies the usual breeder reactor: a minimumesize core surrounded by blanket. The third adds an inner blanket (Figure 3.1-2). Inherent disadvantages with spherical containers may be 1. Difficult fabrication of structural materials 2. Difficult replacement of structurai materials 3. Stresses due to material curvature necessitating a stronger, thicker core/blanket interface L. Non=-uniform flow velocities in the core, possibly creating additional stress. 17 Cylindrical geometry offers choices similar to Fig. 3.1=2: BG in (c) of Figure 3.1-6 may be slightly higher than in (b) when the core fuel is all fissile and no fertile; BG in (a) will probably be lower due to spectrum degradation. Likewise, re- placement of the annular core (c¢) by near—equivalent-area tubes (d) may further increase BG. Either each tube approaches criticality or they approach each other in distance close enough to avoid spectrum degradation. ‘|||I|" ‘lllll" "I%!!l' o0 Figure 3.1-6 Alternate Cyllndrlcal Geometries (Top View) Design (d) also suggests quick and cheap replacement, a simple solution to the problems of radiation damage and corrosion in the reactor. Structural tubes obviously produce no fission products; if made of graphite, they also radiate no gammas. That should facilitate their handiing and disposal. Graphite also costs little., A design with straight tubes would limit re- placement downtime. Design (d) is also easier to make and replace than (e). 3.1.3.2 Choice of number, size, and spacing of tubes. Pro- ceeding with the Figure 3.1=6 (d) concept, Figure 3.1-7 defines some annular arrays of N tubes. The dotted-line c¢ircle passing through the tube locates their centers. To minimize core spectrum 18 degradation the circle might be several mean free paths in diameter so that most neutrons leaving the side don't reach ftubes on the other side, Alternatively tubes might abut one a2nother so that most neutrons leaving a tube either go permanently into the blanket or directly into another core tube, To minimize structural material, N should be small, o o oC oo o 0 o) (22) (8.8) (95) (8.2 o o c o Cn© Op 0o’ N=3 Nc4 H=8& TEYA N=7T Nad Figure 3.1=-7, Annular array of H-tubes A questior certain to arise over core tubes is the interdeven~ dence of their neutronics ani kineticse. One prefers:ejither close or near-zero coupling between them: intermediate loose-coupling could lead to oscillatiions, not quickly damped. 3Behavior as independent reactors (zero-coupling) would probably require multiple control panels and operators, i.c. separate full-power cores sharing the sare blanket, This option might warrant attention as the tubes would share the same reprocessing plant and other facilities as well without affecting each other's operation, Since the reactor physics of each reactor would simply copy that of a conventional_core and blanket reactor, this study will not ireat it separately. Close neutronic coupling implies physical closeness as well (separated by less than one mean free path) and individually far subcritical, Then, shui- down of one tube channel would shutdown the whole plant,. 9 3.1.3.3 Location of tubes. To gain more insight into tube location, consider again the arnular core geometry of Figure 3.1-6(c). Fixing the reactor radius, Figure 3.1-8 depicts designs with inner blanket (IB) diameter ranging from no IB to no cuter blanket. Fig. 3.1-8 Designs with Varying IB Diameters Pertinent metrics include breeding gain, medium flux energy in each region, and critical mass. Section 3.56e1 shows that the largest breeding gain occurs with no inner blanket. This would corfespond.to placing the tubes adjacent to one another in the reactor center. 3.1.3.4 Axial blanket and neutron leakage, Neutrons which leak from the reactor cannot breed. They also require shielding. Lengthening the core helps but an axial blanket stops axial ieakage best. Figure 3.1-9 suggests some designs which accom— plish that by'dafipening the chain reaction in the tubes near top and bottom of the reactor vessel. The overriding concern in choosing between these designs 80 \ ettty S Vertically Horizontally - - Skewed 4/ Skewed Ballooning Side~Entering and —-- - Returning — e b | }— End Diverging Figure 3.1-9. Alternate Designs Which Create an Axial Blanket in the Reactor Vessel 81 should be eame of making and replacing the tubes: otherwise one must design for corrosion and high neutron radiation exposure. The two skewed tube designs should excel in this regard because the tubes are straight. The chain reaction in these would con- fine itself to prolate or oblute spherical gecmetries. The critical geometry for the other three designs, wouid more resemble a cylinder. The next two designs exhibit large bends. Presumably, one- would use straight pieces tc the extent possibie and fasten them to the bends, a cumbersome orocsdure in a radiation environment. The bailconed tube design presents even more difficulty to make and replace. - In conclusicn, the use of a few (N=3 or 4) large skewed tubes should present the least engineering difficulty in their fab- rication, operation, and replacement. It will also minimize neutron spectrum degradation and neutron parasitic capture. 32 3.1.4 Method of Keactivity Shimming The concentration ratio [233U]/[U + Th + FP] provides a suitable though only approximate measure of the ratio of neutron producers to neutron absorbers. As the reactor operatés, this ratio naturally decreases in the core, thereby losing reac- tivity. Replacement by a mix of higher ratio restores the reactivity. In Figure 3.1-10 the blanket breeds 233uc1h and the re— processing plant separates it from ThClh and NaCl (Section 3.7). Addition of 4, Na or Th reduces UClh to UC13: r -+ 5+ UClL ——» HCLl + UCJ.3 Na + UClA —n NaCl + UCl3 Th + 40014 —e ThCll+ + 4UCIL3 The separated 23311013 should immediately mix in a shim tank with highlyeradicactive reprocessed core fuel (for non-proliferation purposes). As shim mix flows into the core, irradiated fuel passes out to the FP remover. If necessary, a ThClh/NaCl mixture could be used to provide negative shim. Figure 3.1-10 | COFC B\thd MSFR(Th) Reprocessing and Reactivity Shimming Scheme ¥ [ Pesitwe S"l.lfl\ T““K P 1 i FP v/Th Removgr ‘;S;::: ::_r 83 3ele5 Summary Guidelines on Reactor Configuration The first priority is to maximize BG. Hardening the ccre neutron spectrum enhances the BG potential but not necessarily the achievable BG. Hard spectrum also increases the fission probability for all the actinides, Spherical geometry maximizes BG by minimizing neutron leak- age but presents engineering difficuities. It also limits the power density: since critical masses are smailer in spherical geometry, power densities rise. Varying the core length in cylindrical geometry provides more choice in the critical mass and power density. The spectrum remains hard. One can also use multiocle, small cylindrical tubes: some spectrum softening occurs there, depending upon their proximity. The optimum configuration may be a small number (e.g. three) of skewed tubes penetrating a cylindrical tank. The effective geometry of the neutron distribution will then closely approach a sphere. 84 3«2 Reactor Thermohydraulics An MCFR will require only low pressures for circulation and repro- cessing removal., Much of the MSBR technology in thermohydraulics should apply as well to the MCFR. 3.2.1 Means cf Ccooiing the Molten Salt Fuel e consider three means of cooling moiten salt fuel (Pig- ure 3.2-1): cui-of-ccre neat exchange, ih--cCre heat exchange acress tubes, and in=-ccrae direct ccatact with a secandary coclant. 3.2.1.1 Out—cf-core heat exchenge., Here the molten fuel flcws cut cf the core through piring to the external heat exchanger, great.y ircreasing the sysi:m inventory. As 3 resuii, mcst de- layed neutrons emanate outs.de the core. This affects reacter control (section 3.6.2.2), and the neutromns act.ivate- the primary circuit and its envirens. The large fuel inventory might pena- lize the economics of a =2 —or Pu-based piant; for an MCFR(Th) the penalty matters less because thorium abounds more, the feed requires no enrichment cr fabrication, and the plant achieves near-fuil actinide utiliza.ticn. Large fuel inventory also works against non-proliferation goals, but not as seriously as if it occurred out=of-plant in transport, reprocessing, or fuel fab- rication operations. The absence of structural material inside the reactor maximizes the neutron economy and minimizes softening the neutron spectrum. The high emphasis an these characteristics in Section 2 prompts us to adopt this method of cooling for this study. 85 ™r- Means O'L_lt—Of—COI'e feat exchanger In-core direct contact In=core heat exchange across tubes rige 3e42=1 Means of Cooling Molien Salt Fuel Scheme [T 3 86 Nelsan et al [21] examined two MCFR(Pu) designs with out~ of-core cooling: his main comment was an economical concern for the large Pu inventory. Lane [ 78] mentioned out-of=core cool~ ing for a high~flux (1016 n sec T cm-z) materials~testing fast reactor. Winfrith examined at least four out~of-core cooling variants, They recommended further study with both He and lead secondary coclants. carlier Taube studies [19] concentratad on in-core conling. Taube cautioned that loss of delayed neutrans out-of-core might hurt reactor control. Later work emphasized out—of-core cool- ing to achieve high flux in owrner and test reactors and high 3G in power reactorse. 3.2.1.2 In=core heat exchanze across tubes. Taube and Ligou [79] 238401 , blanket salt flowing through 23,000 Mo=alloy tubes in the core. A pump analyzed a 2030 MWth MCFR(U/Pu) ccoled by stirred the fuel in the vessel to a speed of 2m/sec to increase the heat transfcr. That much Mo meant large neutron absorption (3.3% of the core total). The core inlet and outlet temperatures were 750 and 793°C. A main question was the weldability of the Mo-alloy tubes. A Winfrith group examined [29] Taube's work.. They found his breeding gains slightly high due to expected deficiencies in the Bondarenko cross sections for 23 8U and Pu. Also they found the physical properties of the salt not to be as good as what he used. These differences would increase tii€ maxdimum fuel salt temperature to 1160°C. There one approaches limits of 87 of corrosion and of sirength of ilo and its alloys, Taube has also suggested tubes of steel-reinforced graphite. Winfrith earlier studied the use of molten lead cooling but limited themselves to an 80000 salt tenperatures Within that re- striction they found poor performance, especially low BG due to neutron absorption by heat exchange materials, dith helium cooling, HWinfrith allowed the salt to reach 97000; the outlet e temperature wes SSOOCg ANL studied two sodium~cocled MCFR(Pu)s [21]. 3e2e1ed In-core direct-contact heat exchange, Here an immiscible £1uid like molien lead [29], boiling mercury [20], or boiling AlCl, [20] mixes with the fuel directly in the core. It then separates out and passes out of the core to a heat exchanger. iThe direct con- tact of molten fuel with molfen coolant affords very good heat trans- fer, elimination of coolant tubes (and cladding), and possibile ex- traction of fission products. The principal problems occur with mixing and separating the fuel and coolant, and corrosion. The need for extensive research to overcome these led to the abandonment of this method of cooling, 3.2.1.4 3Blanket cooling. A Th blanket will generate little heat, Constant bleedoff of the blanket salt (for reprocessing) and re=- placement by cold HaCl/ThCl4 may satisfy the heat removal require- ments,. 88 3,2.2 Power Density in an MSFR 3.2.2.1 Inherently high power density in MSFRs. The absence of struc— tural materials allows high fissile fuel atom density. Absence of nau- tron moderating materials, and limited presence of fertile and carrier salts keeps the spectrum hard. Together, these results produce high v (=96;/ G ) and small critical dimensions. To then produce the same power as a thermal reactor requires high power densities. 3.2.2.2 Reallstic range of power densities. No actual experience with 1SFRs exists. ORNL considered power densities of 5~10 Md/ liter in a study of an MSFR fast flux test facility. Taube's MCFR (Pu) designs range from 0.2 to 11.1 MW/). British designs had 0.36 Mi/l. CRNL operated a thermal molten salt reactor up to 80 kfl/ 1. Peak power densities in existing water-cooled reactors include 1.3, 1.5, 2.5, and 4.4 Mi/1 in m, HFBR, ATR, and HFIR [78], respectively. A coolant velocity of 120 ft/s might perfit 10 MW per liter of core in HFIR [’(8]. For the sodium—cooled fast reactors Melekes Cll-2, Phoenix 250, and FFTF, core power densities average 2.5, 0.46, and 1.0 Md/1. The power per unit volume ccolant for the sodium- and water-cooled reactors above is a factor of two higher., Thus a peak power density of 10 Mi7/1 for a fast molten salt reactor does not appear infeasible, but this st.udy‘should not exceed it without good cause and a lower one would present less engineering difficulty. Flow velocity and heat exchanger size (i.e. cut—of-core fuel inventory) may constitute the real operational limits. 89 3.2.2.3 High neutron flux levels and radiation damage in fast reactors. The follewing integral expresses the power density: Pr) = Xni‘issile Of Cb (B) & (3.2-1) where P(r) the power density at the point indicated by vector r, N fissile = the atom density of the fissile fuel at r, c ;= the fission cross section for the fissile atoms at r, $ = +the energy-dependent neutron flux at r, dE = +the differential of the neutron energy over which the integration is to occur. Because O £ is several magni*udes smaller in a fast neutron spectrum than in a thermal cne, § must increase that much to get the same power - - - i - » density. N fissile I° also higher here, but that leads to small critical volume and higher power density (Section 3.2.2.1) which makes @ even higher. Because of high @ (> 1016 n em 2 sec™?), ancther limit to power density could be radiation damage to structural materials (at thehlan. ket core interface). The number of displacements per atom and the helium production rate gauge radiation damage far intermediate and heavy mass materials. In the past 0.1 MeV and > 1.C MeV fluence served as somewhat coarser (but still valid) éa.figes. This emphasiies ~_tha:t' most 211 6f the neutmné in a fast reactor cause damage wnereas midny don't in a thermal reactor. Coupled with the higher flux levels here, _tha.t bodes a llml'b to the life of s.;:ructura.l materials within the core environs. The solution there is t; perlodically.replacé the skewed tub_es. Cut near the reactor perimeter the 2-meter molten salt blanket has greatly reduced the flux in the pressure vessel down to ~ 101:3 n cm-2 g0 : this gives a 30 year fluence of ~v1022 n cm-2 for which much 5 -1 experience already exists, Keeping the vessel exposure low is ime portant enough to discourage the use of any regular arrays of solid materizl in the blanket through which neutrons could stream, 3e2e2¢4 Impetus for high temperature operation. Heat transfers from the core according to q=m C, AT, (3.2=2) Here g = the heat removal rate = the heat generation rate under equi=- librium conditions m = the mass flow rate, proportional to the fluid velocity (v), density, and the flow area C = the heat capacity of the fluid (energy per unit mass per unit temperature) J&Tf = the temperature increase of the fluid from entrance to exit of the core, Cp and the density are relatively fixed; only v, the flow area, and ATf can varye. Raising v would increase pumping power requirements and corrosione. Increasing the flow area may dictate more ThCl4 or NaCl to avoid supercriticality: that, in turn would soften the spectrum and reduce the BG potential, With regard to increasing Amf, Section 3,3 already requires inlet temperatures of 55000 and higher, Thus an MSFR inherenily generates 91 high temperature heat. This offers advantages (touted elsewhere) as well as problems. The problems may include leakage through seal expan- sion and melting of metal. Also, in numerdus fused-salt systems, high temperature gradients at equilibrium enhance chemical corrosion [80]. Thus, in short summary, increased heat removal requirements will mest likely increase corrosicn and/er soften the spectrum. 3,2.2.5 Adjusting power density through critical mass. Sectian 3.5 sets the minimum critical voiume for an MCFR(Th) at about 60 liters of 0.06 ™. Fer a 10 Gith reactor this would imply a power density of 167 Miyth/l. Even for a 2250 !Wth reactor (1000 Mde) it spells 38 Mdth/l. The power density limits of szction 3.2.2.2 (10 i&/1) dictate a larger critical volume. Cne way to get it is to increase leakage from the core (but not from the reactor) by reducing the core dimensicn in one or more directions. A thin cylinder is one logical solution: above a given length-to=diameter ratio the length can infinitely increase without appreciably changing keff‘ However, space requirements and return piping do impose practical limits on the length. Multiple thin cylinders resolve that dilemma. In regular array (Figure 3.1-7) they approach an annular core geametry with internal blanket (IB). As the IB diameter increases (annulus moves radially outward in Fig. 3.1=-8) the critical volume further increases. 92 In the skewed tube configuration the effective geometry (where the tubes come together) can vary from one of minimal core leakage when angles between tubes are equal and near perpendicular, to one of high leakage when ftubes penetrate the vessel top and bottom or from the vessel sides (Figs 3e1=9), A second way to increase the critical veolume would be to increase neutron absorption by allowing a higher equilibrium concentration of fission products., However, breeding zzin would suffer, A third way is to increase the carrier salt concentiration (de- crease the molar percentage of actinide salt), However, this will sofien the neuiron spectrum and decrease the BG potential, (This would not decrease the BG proliferation hazard of the material: any material which can sustain a chain reaction in a fast reactor, can do so even easier in a bomb configuration,) A fourih method is to add in fertile materizl, This will soften the spectrum somewhat but without parasitic capture of neutrons, In- deed, this option looks very favorable; section 3,3 discusses it further. 93 3.2.2.6 Reactor design power. For a given power density, adding more fuel (e.g. lengthening the core) can increase the total reactor power. This should increase the economics since the capital equipment, includ- ing that for reprocessing, changes mare slowly. However, two or more snall reactors at a plant nay excel over ane larger ane: though each cne requires separate instrumentation and operating crew, while one is down the other still uses the reprocessing system and produces eleciric power. Without significantly coupling the cores, the blankets of sevaral reactors might also abut one another so as to eliminate re- flector and enhance breeding. A popuiar size for intercomparisan of designs is 1000 ifve. 3.2.2.7 HReactor vower distribution. Zonewise, fission of bred 233U should generate more power in the blanket than high-threshold 232Th fission would. The usual concerns regarding power peaking vanish: the fuel is already melted, and burnup is homogenous due to liquid mixing. - 3.2.2.8 Summary. The 2-meter thick blanket protects the vessel from the high fluxes accompanying high power density. The frequency of skewed core tubes replacement will still depend on the flux levels, In- creasing the core volume therefore increases the tube lifetime, The most constructive way to do that will be (1) The use of multiple thin skewed tubes (2) Dilution of the fuel with fertile salt 94 3,2.,3 Primary Coolant Velocity Section 3.2.2.4 pointed out that v, flow area, and AT; are the only variables for removing heat from the core. 3.2.3.1 Maximizing velocity and minimizing pumping power, ToO maX- imize velocity ane desires as few bends as possible. Flow through a sphere like Fig. 3.2-2, unless baffled samehow, would proceed more slowly near the surface than through the middle. This could lead to surface hot spots as well us neck corrosian. The ideal ge-metry for maximum v would seem to be large-diameter pipes as described in section Jelele Fig. 3. 2"'2 Typical Spherical Reactor For canstant mass flow rate the coolant velocity, v , will vary in=- versely as the tube cross sectional area. Pumping power increases as v3; for v {15 m/s, pumping uses 5 = 8% of the power. Friction losses in small pipes will add to this. This recommends a few large pipes for maximum flow velocitye. 95 3.203.2 Corrosion dependence on velocity. How velocity affects cor- rosion depends upon the corrosion mechanism which, in turn, depends on the metal and its environment [80], For corrosion by activation polare ization (Curve B, Fig. l.2-2), velocity has no effect, Corrosion by cathodic diffusion does increase with velocity as in Curve A/Section te Small amounts of an oxidizer, e.g. dissolved oxygen in acids or water, czause this, Easily passivated materizls such as stainless steel and titanium frequently resist this corrosion better af high velocity: there the increased agitation causes an active-tow passive transition (curve A&, sections 1 and 2). Some metals resist corrcsion in certain mediums by forming massive protective films on their surfaces, These differ.from the usual pas~ sivating films in that they are readily visible and Such'less tenacious, Extremely-high velociiies may mechanically damage or remove these films, resulting in accelerated attack (curve C)s This is called erosion cor- rosion, f c Corrosion Mechanism - 8 S A: Cathedie diffusian ¢ B: Activation polarization S| 4 C: Erosion 0 1 v 2 VeloCity e Figure 3.2-3 Behavior of Different Corrcsion Mechanisms 96 3e24,3e3 Velocities in similar systems, In an LMFBR core sodium typically flows at 8 m/se In the HFIR core, the coolant reaches 120 £t/s (37 m/s). In theoretical studies, Taube [17,79] considered velocities of 10416 m/s in the heat exchanger, choosing 14 m/s as optimume. The British [29] varied v from 4~11 m/s in their NCFR heat exchanger. ORNL designed an MCFR heat exchanger with a fuel salt velocity of 6.1 m/s,. 3e244 Overating Temperatures Numerous criteria will limit the primary circuit temperature, Figure 3,2=4 illustrates the discussion. 3.24441 Upper limits on fusl salt tcmperature, A high fuel sali temperature could increase the feasible AT and thereby provide extra hezt transfer driving force, This would then reduce the needed inter- mediate heat exchanger area, and therzto the external fuel inventory. Temperatures above 900°C allow operation with He as the secondary coolant. Even higher temperature would allow MHD topping cycles and increased efficiency irn conventional heat exchangers, However, the temperature musn't exceed the salt boiling point or cause damage to the piping materizl in or out of the core, Temperature increases the rate of almost all chemical reactions, Corrosion fre- "quently accelerates exponentially with temperature once past some threshold value, Refractory metals such as Ho and Mo alloys have high melting point and {thermal conductivity and resist corrosion well, With graphite piping, the fuel salt temperature limit might extend to 150000. (Melting points of graphite or Mo are much higher ,) I 86 temperature 1. corrosion j 2. oxidation Temperature SUpper 1imits on fuel salt B Figure 3424 3. Other chemical reactions fuel L« structural melting Temperature ° ( 5. fuel salt boiling 6 Constraints salt y . excessive vapor pressure in Heat . Exchanger ,70000 _— _ safety margin Design primary circuit AT; across heat safety exchanger =~ AT across core margin - on temperature difference fuel salt‘)'b < across heat exchanger tubes freezing 1. static stresses point 2. dynamlc stresases : 3. migration of elloy con— stituents leading to ///J accelerated corrosion minimum tem-— secondary ] perature for coolant process heat —t~~ applications safety margin | sécondary C L ccolant freeging 7 point End A heat End B exchanger position The British [29] considered 970°C with He cooling and 810 with Pb cooling. Taube studied 860-1100"C and chose 950°C, ANL [21] stayed with 740°C. The French considered 1000-1300°C for high temperature applications but warned of problems to be overcome, 3,2.4.2 Minimum fuel salt {temperature, The minimum inlet temperature must exceed the salt melting point with an adegquate safety margin. Although melting points of actinide chlorides decrease with transmute- ion up the A-scale (Section 3.6.2.3), impurities also build in which might raise the melting point through the formation of complex com= pounds, Table 3,3-VI suggests 2 minimum T of 65000 to allow some fuel range in composition; addins a BOOC margin glives TOOOC. For comparable MCFR(Pu) concepts, the British chose 650-670°C; Taube used 750°C; ANL, 625°C; OZNL, 566°C. 3e2.4.3 Hinimum secondary coolant temperatures. Secondary coolant salts with 2 variety of melting points are available, at least in principle (Section 3¢10.1e1). In practice other problems and properties may limit the choice. For process heat applications, the secondary cocolant temperature must be fairly high, 99 3e2sded Restrictions on AT across heat exchanger wallsz, For fuel salt flowing through a hollow tube,the maximum stress occurs at the outside cylindrical surface of the tube walle It surpasses the wall material strength O (psi), such as the yield strength (YS), when the AT across the wall exceeds o(1-9) _ § [z, 2. (=) AT, = X E _SE[O‘*b In bla 1 at 2 '|n b’a_ - bi_o_z (302"'3) where ¢ = volumetric heat generation rate, Btu/hr in3 k = thermal conductivity, Btu/hr F in = coefficisnt of lincar thermal expansion, F'1 E = modulus of elasticity, psi D = Poisson's ratio a = inside radius, in b = outside radius, in All three terms in this expression are positive; the second numerator term is generally quite small compared to the first, A safety margin should most likely be in ATw itself; this will allow for abnormal transient conditions which produce (1) High fuel salt temperature (2) Low secondary coolant temperature Two materials are of principal concern, Mo and graphite, Table 3,2-1 lists a2 range of properties for graphite, To find the largest AT _, property Set 1 maximizes G and R, while minimizing o{and E, Complementary Set 2 should give the smallest 1}Tw. Table 3,2-I1 shows the results for graphite and 1o under the principal assumptions of 100 Thermophysical Properties of Mo and Graphite Graphite Graphite Material Mo* Graphite** Set 1 Set 2 E(106psi) L8 0.5—1.é 0.5 1.2 & (10571 2.8 1.0-22 1,0 2.2 v (Btu/hr F in) 6.5 T oi=9.7 9.7 T ol ¥ 0.32 0.25 0.25 0.25 0.29-YS (1o3psi) 99.8 —_— — —_ UTS (10°psi) 101. 0524 2.5 0.5 * arc cast, hot-rolied ** electrode quality UTS: ultimate tensile strength YS: yield strength 101 Table 3.2-I1 Caiculated Stress Limits A'I‘w Across Heat Exchanger Tube Walls G- core volume Material IR/CR (10°psi) () At (€) o 0.20/0.25 in* 99.8 0.25 586 99.8 0.5 596 99.3 1.0 601 101. 1.0 601 Le5/5.0 mm%* 99.3 1.2 581 0.20/0,25 inx 99.8 2.0 604, 439 0.25 288 49.9 0.50 289 { LS.9 1.0 250 Le5/5.C am** LS.9 1.0 298 0.20/0.25 in* L45.9 2.0 230.4 Graphite 0.20/0.25 in* 2.4 0.5 1313 Set 1 24 1.0 4317 1.2 1.0 2157 Graphite 0425 Q.25 68 Set 2 0.25 0.5 76 .7 0.25 1.0 81 0.25 Q.5 162 0.50 1.0 166 * Ax = 50 mils #* Ax = 19.7 mils 102 | (1) @ = YS or UTS | (2) A O,20-inch radius tube with 50~-mil wall thickness ( a popular size) (3) g = reactor power/(core fuel salt vcolume of 1 m3 and perti- nent variations thereof) With graphite Set 1 parameters, the siy's *he limit almost. Sef 2 parameters would allow a maximum szw of 162-16600, depending on the exact power densiiy, A safety factor of {two would lower the maximum AT, to 70-80°C. For llo the fixed range of parameters appears to offer more certain- ty. The z2llowable zyrw‘= 60C C seems very adequate, easily accomodating a safety fzctor of two, One sbserves thawu 1e¢ The results do not depent on g which varies inversely with core volume; i.e. the second numerator term in Eqe 342-3 is generally small 2o - W2ll thickness does not set the szw limit, at least dowm to 20 mils 3. The most sensitive quantity is o/«E In conclusion, llo seems to have an adequate safeiy margin; graphite utility may depend on what kind of quality control can be achieved at reasonable cost, The two fluids in the heat exchanger should counter- flow to minimize the A‘I‘w. 103 3e2e4e5 Restrictions on ftemverature differences within the primary circuit, The maximum Zle within the primary circuit will be the difference between core inlet and outlet temperatures, A similar [&Tf occurs betfween heat exchanger inlet and outlet, NS3R experience in- dicates that when (1) The reaction between the fuel salt and a structural alloy constituent stronzgly depends on temperature, and (2) Temperature differs greatly within the circuit, then the constituent dissolves at the high temperature regicns and deposits in colder regions, The scope of this study Joes not permit quantification of this re- striction; however, the poteztial mass transfer does discourage going far beyond the circuit AT, in similar reactor designé’(below). 3e2e4e6 [ AT, T] in other MCFRs, Figure 3.2-5 summarizes the resulis of previous MCFR(Pu) designs, Taube's [17] and the British He-cooled designs [29] are perhaps the boldest in using higher T's and AT's, Many desisms use a lower melting salt (with lower actinide molar con- tent) to operate the fuel salt below 70000. 3elelte] Summary, Based on the above discussion, this study considers (1) A minimum fuel salt temperature of 700°C (2) A referemce AT, = 200°C; advanced, 300°C (3) A reference Amiéx = 200°¢; advanced, 3OOOC (4) A reference LMTD = 150°C; advanced, 200°C The advanced AT's exceed those of earlier studies (except Taube's) and are to be avoided if possible. 104 o] 1000 : C ; e . C{l A : Dl b fwg‘ ' i s [ S o i Temperature 800 Ol &Qa "’ Year 1956 1906 197080 1975 1975 1975 Intermediate Na or ta Coolant Na F 3alt Na Pb Pb He AT fuel (°¢) 167 115 200 150 160 300 % o (°c) 345~ across tubes 120 —— 429 113 189 185 Maxe AT 350~ aoross tubes 167 — 520 140 210 270 * Figure 3,2~5, Review of MCFR(Pu) Heat exchanger Design {Afl% * See IMigure 3,2~6 for clarification of diagrams if necesasary *H defined in Section 3,251 3e245 Desicn of the Primary Circuit Heat Exchanger 3e265e1 EHeat transfer coefficient, The heat exchanger plays z crit- | ical role: virtually all the heat generated in the core must pass through its tube walls into the secondary coolant, Glasstone and Sesonske [81] show that: = 2TR 3 ' q U(2TTR, L &) &T (3.2-1) 1= R ., R @/R) 1 (342-2) U h 2. k i W hi.c. where q = heat transfer rate (cal/s) U = overall heat transfer coefficient (cal/s en® ) R = cutside tube radius (cm) R.= inside radius (cm) of tube through which fuel salt flows k = wall thermal conductivity (cal/s om X) ieCe= Film coefficient of intermediate coolant (cal/s cmZK): this term is generally non-controlling in (2) above L = fupe length N = no, of tubes z&Tw = temperature difference between fuel coolant and infermediate coolant (across the tube walls), Generally one uses the log- AT, — ATa In (BT, /ATa) zero-dimensional approximation, as defined by Fig. 3.2=6, mean temperature difference, LMTD = y for a The heat transfer coefficient for the fuel is Bt M (342-3) 106 AT ///////// AT _ AT f A% :53’ 2> AT a Fnd a: End b: fuel out fuel in 14Ce in ieCe oOut Figure 3.,2-6 Heat Exchanger AT Definitions where k = thermal conductivity of the fuel(cal/som X) D 1} diameter of tubes throush which it passes {(cm) For turbulent flow (Re >1000) of fluids with 0.7< Pr ¢ 700 through circular tubes of L/D:>60, the empirical Sieder-Tate equation gives ¥u = 0,023 ReO'BPr1/3 (/u&[/xw)°‘14 (3.2-4) where "b" refers to {the arithmetic bulk temperaturs, and "w" to the average wall temperature, and where 107 O < D Re = & Pr=—aL R v = linear fuel velocity (m/s) il fuel density (g/cmB) fuel viscosity (centpoises = g/m s) fuel specific heat capacity (cal/gm X) C P To use the above convenient units we must convert centimeters to meters 1 03.2-0. 65 within Re and Pr, Doing this ( x 0,023 = 7.97) and grouping terms gives he (E‘é’%%fi) - 7,97 (vp) 0.8 | 0467 (cp//L)O'B/uw'O'MD‘O'Z (342-5) Equation (3,2=-5) indicases what makes a good film coefficient: high values far/o 'V Xy cp’1é“4 and 1/D in that apfifoximate order of prioriiy. From Section 3.3.5.3 h. is typically 1.2 cal/s em'K. Assuming R .= 0,20 in, B = 0,25 in, and k= 0.32 cal/s em’C, get 1 . 0:25x 25412 1,25 1 T ~0.8x 1.2 0.32 B, (3.2-6) = Bl§3 + 5—%3——.+ smaller term U =0u67 cal/s cn® K The importance of the second term relative to the first in (3,2-2) or (3.2-6) is hy ) h R, Roln(Rd/Ri) ] he R, 1o (Rd/Ri) h/ - R e -_ k ® 0 W W 0.42 here i o e |& N ol%i i 108 Alternately for IR = 4.5 mm, OR = 5.0 mm as in Table 3.2~IT h L o= 1.2 x 0.451In (50/45).= 0.18 h’ 0.32 This shows that the fuel salt film coefficient controls the heat removal: going to a thinner tube will not help much. The significance of Mo would be that it would allow a wider and thicker (therefore more reli. able tube without heat. transfer penalty. For comparison, the ORNL study reported [13] . -1 Us|—t—r 4+ —2 ¢ -t - 0.15 0.30 0.5 1.8 They used the same dimensions as here, but their fuel salt hf was three times smaller: they assumed 17% UCl3 fuel salt and 83% carrier salts {MgCl., & NaCl) while this work presumes high actinide content. The wall 2 material k in the ORNL study was 6.5 times lower (80 Ni/20 Mo there vs. pure Mo here). 3.2.5.2 Minimizing fuel inventory in the heat exchanger. The ofix-of- core fuel inventory affects reactof stability (Section 3.6.2,3}, U inverory costs, and out;of-coré‘criticality.- The -fuel volume in N exchanger tubes 1is 2 vfubes = NLTrRi ' (3-2-7) A reasonably-ccnservative guess of the additional fuel in heat ex— changer plenum and piping appears to be (section 3.6.2.3) ane additional equivalent volumes ‘Then 109 2 OeCe vou‘t-of—core ~ 2 NLT Ri (342-8) Rearranging equation (3.5=1) gives There nat e This or Given sity suffi vance would 2 ¥ LW=q/UR AT fore 2 Voo = By g/ U R, AT, The condition for stability (section 3.64243) is that v » xceed fthree times the core volume, Vc, leCa: vo;,c. < 3 VC means 2 R q<3V U2 AT 2 V,AT, > R “q/ 3R U = 0.67 cal/sec em’ K R;= 0620 in = 0,508 cm R= 025 in 00535 cm = 2250 x 106 watis x 042388y cal = 537.5 x 106 cal/sec watt sec VCATW > 109 m3 K Figure 3.2-7 plots this relation, adding boundaries of power den- {10 MWth/liter and.AEw<:300. A broad reference region lies ciently far from the limiting boundaries so as to not require ad- d design development, To further broaden the reference region require smaller Ri or larger U, 110 Ou‘l'cr flTm“ lelf . in Other Desfim Direction of Neutron Spectrum Dea)f‘ad.a-‘hf:n \ / \ \ . 20+ /\ L (1125 MWy [2) \ 4 \ \ . / \ \ {5 - \ . AR $ Critical \\ ) o nc:rcble Volume | ' ! requ'or\ Vo, (,‘,,_3) 0.5 4 A \ , \ \ / 10— \ e +(2.25 MW, (L) N \ . wz— Ve AT=10q n? K Sfa_b'.\'.k‘ Limi¥ /\ \ ~ 7 oMWy (). . _f. Poq::r&'f)! *‘\, Lumd' Al e i — T Y o 100 2060 36 400 AT, (x) O Figure 3,2-7, Locating the Optimum ATW and VC 111 ' Lzrger U implies higher kw and hf and lower effective wall thick- ness Ax = R In((R O/Ri). Both pure Mo and graphite exhibit high k (Table 342-1)s Low Ax implies keeping the walls as thin as possible, | This dictates a metal with good resistance to corrosion, especially ‘ with 2 high temperature-gradient present. It must also withstand the radiation damage from the delayed neutrons; either that or be cheap and easily replaced. 3.2.5.3 Volume in heat exchanger and associated plena and piping. From equations 3.2-1, 3.2-6, and 3.2-8 0.239 cal 2250 MWth X =t = U(2T R, L N) AT, 100 MH s ' = UR oc AT O 2 W R 8 - v 2:25x239x10° B2 w0t e % 0.67 x AT (K) R, cm? 2 - R‘ VOC(mB) = ..E_g—— X e (cm) ATw(look) Ry, For cases of (R;,Ro) = either (0.20 in, 0.25 in) or (0.45 em, 0.50 cm), get 3.3 oc ATw(loaK) Considering ATW —= 150 K from Section 3.2.4.7, 8et — 3 Voc ~ Zm 112 3.2,6 Primary Circuit Arrangement 3.24641 General Guidelines, Sections 2.5.4, 3.6, and 3.7 and 3,9 show the need for transferring the core and blanket fluids to remote storage tanks, and flushing the primary circuit. Sections 3,6,2 and 3.8 emphasize minimizing the fuel salt inventory. Section 2,5.,5 mentions the importance of reliability and plant factor. 3424542 Location of pumps and heat exchancers., Both pumps and heat exchangers should closely surround the reactor as in Figure 1.,2-2 so as to minimize the fuel inventory in pipinge Shielding from core neutrons, beyond that provided by the blanket, is rot warranted as the primary coolant itself emits delayed reutrons, Placing the pump between the heat exchanger and the ccre inlet imposes a lower overating tempera- ture on it; this should lengthen its service and easeiits designe 3e2e643 Parallel subchannels, As each tube channel (of N total) leaves the core it could split into M>2 subchannels, each with a small pump and/or heat-exchanger., Such redundancy should increase load factor and safety of operation, Small size would also make it easier and cheaper to build and replace the units, and easier to avoid criticality outside the core. However, as the number of pumps increases, the chance for a single pump failure does also, Thus the optimum plant load factor could occur for a smaller number of pumps and/or heat exchangers. 113 Several alternate arrangements are possible (Figure 3,2-8), Plan A schedules a single pump and heat exchanger for each channel, Plan 3 splits the flow outside the core into M>2 parallel paths, each with a punp and heat exchanger. Plans C and D only duplicate pumps., Thus in 3; Cy or D, 2 single pump failure does not terminate all flow, Approaches A, B, and C differ significantily from the multiple cool=- ing loop concept of current reactor designs: the principal N loops have no common container as in Plan D: each circulates independently. Plan D probably is infeasible because of the (1) Desire to compactize space so as to keep down the fuel in- ventory (minimum piping per heat exchanger) (2) Difficulty in avoiding a critical configuration if a single pump fails with plan A, not only must that channel flow cease, but the others' must also: +the whole reactor must shut down less the chain reaction continue to generate heat in the broken channel as well as in the two good ones, 1In plan 3B or C, failure of one pump again affects the whole reactsr: +the flow and power reduce by 1/M in the bad channel and the other N-1 core channels must also due to the close neutronic coupling, If M=2 subchannels in plan C represent regular and backup paths, then each can handle the full flow and one is always ready in backup - just need to switch a valve, PFigure 3.,2-9 shows a top view of such 2 plan, Use of multiple (M22) subchannels especially for pumps, should increase reliability and ability for reduced power operation, but may also increase fuel inventory;auxiliary heating, and friction (due to added bends). 114 7] Plan A Plan B ———a et —— —— o — —— e p— —— — o - — Plan C Plan D Figure 3.,2-8, Alternate Pump and Heat Exchanger Arrangements for Each Core Channel 115 /r\_;_/ Shielding Pumps = Heactor vessel’ . - m . ~ o i 3 - - . DD e L L T *at - P ige 3.2-9 Tcop View cf Plzn 7 itk Altermatz Subcharnals 3o Snie Calls Fig. 3.2-10 Schematic Diagram of Series Cannection of Three Skewed Tube Channels 116 3.,2.644 Tube channels in series, ror a skewed core (recommended in section 3.143¢4) the inlet and outlet of a tube occur on opposite sides of the reactor. To avoid excessive piping and fuel inventory the outlet of one tube shouid connect to the inlet of another, which lies vertically belows This means connecting all N channels in series, Fige 3¢2=10 attempts to describe this. It's as if someone ran a needle and thread through a cylindrical cushion; coming out the top, the thread returns outside the cushion to another entrance point on the bottoms, This continues until the thread reiurns to the starting position. The series conneciion causes failure of one pump or aeat exchanger to shut down the whole system. But this does not differ from plan 4, and one can still employ duplicate subchannels as in.% or C above, Figure 3.,2-11 shows a B-type horizontal M=2 design. Here the tubes peneirate sides rather than the top and bottomes This should simplify the reactor support and reduce the out-ofwcore inventory as well, A physical model would be three pencils lying across each other, penetrating an approximately-square cylinder of (blanket) jello. 3.2.6.§ Summary. In the reference concevi a small number of skewed tubes connect in series, Alternate subcharnels with pumps in each could provide backup, but would add frictvion losses, On-~line backup (to avoid reactor shutdown) would also increase inveniory and auxil- iary heating. Keeping N to 3 or 4 should help the system reliability as well as reduce the time required for repair and maintenance. It also minimizes neutron spectrum degradation and parasitic absorpticn. 117 — — = — R . Bocikup HX & Pump (BHXP) PLAN Cne of Three zZmergency VIieid Lines to Core Salt Safety Tank 1 Blanket ‘ 1 HX Core tube HX L - - - F - — 4 ( T =) ! ‘ Blanket l T Blanket Salt Drain \‘J to Safety Tank ELEVATION VIizd Figure 3.2-11 B-Type Horizontal M-2 Layout of Pumps and Heat Lxchangers. Arrows denote core sait flow. 118 Limits on power density may come from 1e Radiation damage to the high flux 2o Hizh temperature of the fuel coolant causing leakage through a seal expansion, melting of metal, or chemical corrosion 3, High fuel inventory in the heat exchanger, affecting eco-~ nomics,dcubling time, and reactor conirol, Use of replaceable graphite for the material in contect with the fuel coolant throughout the reactor and much of the primary circuit might eliminate much of concerns (1) and (2). 119 3.3 Choosing the Salt Composition The abundance and low neutrcn moderation of chlorine maké sodium and actinide chlorides the preferred molten salts for fast reactors on the U/Pu cycle (Section 1.2). Here we ccncentrate on the thorium fuel cyéle which could have different priorities. 3.3.1 Intrcducticn The top cancerns in chéosing the sélt ccmpositian are neutran- ics (especially neutron spectrum), chemistry, and meiting point. Cther desirabie tréifis are 1. Good heat transfer properties: high thermal canductivity and specific heat cazacity 2. High boiling point to minimize vapar pressure 3. Adequate technological cr laboratory experier;ce L. Low fuel salt viscosity to minimize pumping ccsts 5. Low price and good availability 6., Non—toxic J::.1.1 Neutron spectrum. We desire as hard a neutron spectrum in the core as possible, though not at the expense of BG. Sifice non-— fissile constituents mainly downscatter and capture neutrans, this means maximizing the ratic of 233U to all other isotopes. | In the blanket, a softer neutron spectrum enhances neutron cap- ture by the fertile material. Thus neutrcn moderators might help ‘there, if they don't absorb neutrcns as well. However this would not be good for an inner blanket as it would then soften the ccre neutrcn spectrum. 120 If the core tubes were many mean free paths apart, then each tube would individually approach delayed critical and have a hard spectrum, but their ccupling would be loose or near-zero. 3.3.1.2 Chemistry. Using the molten salt fuel as the primary cool= ant of a nuclear reacter presents many novel problems in both reactor design and system chemistry. The success of the CRNL Molten 3alt Reactor Program shows that it can be done at least up to their pow- er levels. Experience there also revealed that the chemical states often behave as if in equilibrium. This allows much progress tcowards understanding interactions within the salt (chemical stability) and between the salt and its envircnment (corrosion) since many of the equilibria are grossly predictable. The free energy of formation of the compounds, AgG, measures the chemical stability of the salts, specifically their ability to resist forming other compounds which precipitate out. Taube has deduced the temperature-dependent A G for salts of interest (Fig- ure 3.3-1;[82]. Candidate reactants in the system which threaten salt stability l. Atmosphere constituents oxygen, water, and carban diaxide 2. Structural materials (Sectiom 3.4) 3. Fissian products and other mutants Radiolysis also affects salt stability. Chemical corrosicn involves several sciences (Fig. 3.3-2). Physical chemistry and metallurgy describe the physical, chemical, and mechanical behavior. Thermodynamics reveals the spontaneous 121 —400 300 500 1000 1500 2000 TEMPERATURE [°K] Fig. 3.3-1 Free Energy of Formation for Chlorides [82] 122 Electrochemisiry Physical Corrosion <«——————— ietallurgy chemistry Thermodynamics Figure 3,3-2. Factors Affecting Corrosion of a Metal direction of a reactiocn ard whether or not corrosion can occur. Electrochemistry describes electrode kinetics: the dissolution and plating cut of different materials in electrical contact. 3.3.1.3 éutecticrmeltingApoint. Figure 3.3-=3 depicts a typical eutectic behavior: the mixing of two salts lowers the melting point far below that for either salt by itself. The amount of low= ering depends on the molar ratio of the two salts; excepticnally low temperature is possible at the eutectic point, or nadir. This may be of special value for the blanket region where litile heat 1is generated. A fuel mix with melting point well below 700°C would minimize auxiliary heating when shut down and allow a large temperature rise in the core without reaching high outlet temperatures (> 1000°¢). Low melting points of the component salts also will facilitate their initial dissolutione. 123 Carrier salt Actinide salt like NaCl like ThCl4 e — = — = e m - e - 700°~1100°¢C Melting Temperature 300°-700°%¢ 0 50 100 Actinide MNolar Proportion Fizure 3,3-3, Tuteziic [lelting Point Behavior for = ixture Adding 2 third component like a second actinide szlt in Pigure 34,3-4 can further lower the melting point, especially when all three are in near-—-equal proportions., Unfortunately many multicome ponent (ternary or higher) phase diagrams are unkmown or in unavail- able Russzian literature, ~T70°C Figure 3.3-4. Ternary phase diagram for Nac1/9u013/frh014 124 3.3.1.4 Heat transfer parameters. Sectian 3.2.5.1 showed how MCFR heat removal varies according to g 047 ©.33 : h; ~ p O® kfi (.gf ) J//Jw-0.4 / / For survey purposes here let o (bulk temperature ) i//b{(wall temperature ). -~ Then O-B |- 0067 .O'L]-7 0033 - { he /J < /6 Cp The exponent magnitudes measure the.relstive significance of the con- stituent parameters, As cother factors disccurage the use of fluorides in the core, this section studies oniy chliorides. Infermation an density ( fD ) is read- ily available for pure salts and a satisfactory algorithm exists far mixtures. Viscosity (/}1 ) and specific heat (Cp) data also suffice. Thermal conductivities are very sparse. 3.3.1.5 Densities. Desyatnik et al [83] studied molten mirtures of UClh and alkaii chlorides. They found the density of individual salts to decrease linearly with temperature, getting n (LiC1) = 1.8807 = 0317 DKL £ 0.0011 g/ca’ 1000 (NaCl) = 2.1332 = 0.5405 T() +- 0,0010 1000 (xc1) 2.1751 - 0.6060399— + 0.009 1000 (RbCl) = 3.,1069 ~ 0.8799 TE) + 0.0013 1000 125 4 (CsCL) = 3.7726 — 1.0716 ZE)L L 0.0011 f 1000 /4 £ (BeCl,) = 2.276 - 1.10 ]g—é%-L + 0.0011 (MgCl,) = 1.976 - 0.302 T(K) 4+ 0,001l 1000 9 (CaCl,) = 2.5261- 0.4225 TgK% + 0.0011 10 J (SrCl,) = 3.3896- 0.5781 %%l + 0.0011 ‘ T(K) + O (BaCl,)} = 4.0152- 0.6813 0.0011 l — P (PoCLy)= 6.6213 - 2.0536 2L 4 0,012 / 1000 /,-J (ThCL, )= 6.2570 - 2.7030 IK) + 0.015 ; 1000 f O (UCL3) = 7.520 - 2.472 ) /' 1000 ; o (vey,) =-5.6251 - 2,292 ZEL 40,0021 1000 Kinosz and Haupin [ 84] ‘give the density of NaCl and KCl as / (LiCl) = (1.463 + 0.002464 T)/(1 + 0.002 T) ’ (NaCl) = {1.611 + 0.001897 T7)/(1 + 0.001538 7) (KC1) = (1.568 + 0,001619 7)/(1 + 0.001429 T) / (MgCl,)= (1,682 + 0.0044628 T)/(1 + 0.002857 T) /3 (CaCl, )= (2.115 + 0.003768 T)/(1 + 0.002 T) 126 where ¥ = T(K} = 973 = T(OC)- 700. A spot check at 1000 K showed fair agreement for NaCl and KC1, getting /o = 1.596 and 1.552 by Kinosz vs 1.593+0.001 and 1.569+0.001 by Desyatnik. Equivalencing the two formulas one gets Kinosz R (LiC1) = 1.463 (1 - O.-000316 ) f)(NaCl) = 1.611 (1 - 6.000360".') /o(z«:m)' = 1.568 (1 — 0.000396T) F(b1g012)= 1.682 (1 - 0.0002047) /0 (CaCl,)= 2.115 (1 = 0.000218 T) O(LiCl) = 1.461 (1 - 0.0C02967T) Desyetnik /p (NaCl = 1.607 (1 - 0.0C0336 T)" /J (KC1) = 1.585 (1 = 0.000382 T) /o (RbC1l)= 2.251 (1 - 0.000391T) /o (CsCl)= 2.735 (1 - 0.000392 T) /fi(PbClz)=h.623 (1 ~ 0,000444 T) P (ThCl, )=3.627 (1 - 0.000745 T) ) /o (U013)= 5.115 (1 - 0,000483 T /fi (Ucl, )= 3.395 (1 - 0.0006757") The Kinosz equivalencing is by b —_ - b 1+c¢T 14+¢7T The resulting slopes (v coefficients) are expectedly higher than for Degyatnik 127 For a mixture of salts Desyatnik showed [83]} that linear addition of the molar (specific) volumes of each salt P = LI wifp ] gives densities high compared io experiment. Direct addition of densities would give even higher errors, in the same direction. In binary mixtures with carrier salts like NaCl, the maximum deviation occurs a2t about 60 mole % actinide salt. A spot check on a 50=50 UclL 4/1:5.01 mixture at 1000K predicted /c (50/50, 1000K) = 2,91 whereas the measured was 2.871+4¢ 0.,0024, Formation of complex salts, e.g, 3 KC1 + UClB-—-b-K3UC15, causes this deviation from linearity. 3.3,1.6 Viscosities, Section 3.2.5 showed that low viscosity in- creases heat transfer. Fizurs 3.3=5 displays the temperature-dependent viscosity for individual salis [85,86] beginning at their melting points, The viscosity of salt mixtures is gemerally unavailable; Figure 3,3-6 does show viscosity behavior for two salt mixtures: first Mg012 with NaCl [87]; then with ThCl, [88]s (Note that the two sets of pure Mg012 measurements do not agree. The (latter) Russian work is partic- ularly suspect because of other errors found there, Still the data are informative here,) The bumps correspond to formation of complex moclecules, Comparison of these diagrams with phase diagrams Figs. 3.3-7 and 3.3-28 suggests a correlation with eutectic effects which, in turm, depend on complex molecule formation. In general it appears feasible to choose a2 molar composition where the viscosity is close to the low- est individual salt viscosity. Therefore, for ThCl4/Nacl, UClB/NaCl, UCl4/NaCl or combinations thereof we can roughly presume 128 D b b ——— ——— ——— 1 - —— > 4 —— - - . o= L et e e TS IT I T Tt tn S oo e e == f_om T e Tk — g ) 4 5 6 7 8 9 10 11 12 13 14 " Temperature (100°) Figure 3.3-5 Viscosity of Individual Molten Salts [85; 86]. 129 = T TRttt T T - Poo - It e e e e e L (L VI S Cos-d'y (ep) Hsuz 'ThCh; ThCl, mole ofe Figure 3.3-6 Viscosity Behavior for Two Kolten Salt Binary Mixtures [87, 88] 130 900 800 | 700 F Tamp= ature (°c) 600 500 } J.LOO . i . g 0 20 40 60 80 {00 MLl, MOLE % NaCl NaCl Figure 3.3-7. The MgCL,~NaCl Binary Phase Diagram [13] 131 M) = Mgy (B From reference [85] Myacy = 186 x 107 exp [3308/1.987 T(K)] cn. 3.31.7 Specific Heat. Perry"s Handbook [89] recommends for fused salts at and slightly z2bove the melting point: - ( cal ) - 8.1 ( cal{oc ) ) (no. atoms) D c 2tom male molecule - gn} C Molecular weight (gms/mole) Data on chlorides (Table 3,3-I) from the Molten Salts Jandbook [85] supports this formula for 2- and 3-atom molecules, For 4-atom mole- cules, the coefficient increises to 8.5 cal/°C 2tom mole, The next question is the temperature dependence ‘of the specific heat, Table 3,3=II shows how experimental {03_1 - ac, / dTE [90] for some F Salts compare to their E;P°-1 d/o/dT}and coefficients of linear sion gd_ CO expansi P i * p and /Do represent values extrapolated to T=0 0= p fluctuations of the two parameters do not correlate very well however, Kelvin, On the average & parzllels (C de/dT) fair; individual Still this represents the best empirical correlation which could be developed within the scope of this study., On this basis we approximate c, (cal/em C) = cp° [1-al(X) ] wnere Cp = Cp (T} /L1 =T ()] Table 3.3~III calculates some salts of interest. Figure 3,3-8 corre- lates the resulting Z.CP [900°] } so as to deduce the missing value for MgCl (for which o« was unavailable), 132 Table 3.,3=-1 Dependency of Specific Heat on No. Atoms per Molecule [85] t cp N Cp" ( cal(oc ) ( No. atoms) cal/ °c ) Salt mole molecuie molecule mole Qtom LiCL 15.0 2 7«5 NaCl 16.0 2 8.0 KCL 16.0 2 8.0 CsCl 18.0 2 9,0 AgClL 16 .0 2 8.z CuCl 15.9 2 g0 TiCl 1744 2 8.7 Avg. 16.3 2 8.16 NiCl2 2L.0 3 8.0 MgCl, 22.0 3 Te3 C&Clz 23 .6 3 7«9 SrCl, 27.2,26.7 3 9.0 BaCl 26.3,25.0 3 8,6 Ca012 24.C 3 8.0 Hg012 25.0 3 8.3 ZrC.L2 2Le1 3 8.03 Pb012 23 .6 3 7.93 I-InCI2 22.6 3 7.53 FeC'.L2 2hely 3 8.13 Avg. 2442 3 8.06 LaCl, 3747 L 9.4 PrCl3 32.0 L 8 Nd013 35.0 L 8.8 AlCl3 31.2 L 7.8 GdCl3 33 .7 L 8.4 HoCl3 353 4 8.l ET013 33 07 1{. 8.14. BiCl3 343 L 8.6 FeC.L3 32.0 4 8.0 f‘.vg. 3349 4 8047 133 Table 343-I1 Approximate Equivalence Between Specific-Heat Temperature-Dependence and That for Other Physical Parameters, Based con Heasured I Salt Hixfures 105 X CO=- ol 100 doy 10 4P ot eveem Hix Fracticns CE 0 K)d= (0 1)aT sion NaF/ZrF 4 50/50 28.37 23.0 29,06 LiF/NaF 60/10 4o W47 2i.40 27.1 LiF/KF 50/50 32.12 25.70 34.0 LiF /RbF L3/57 36,52 26495 36.5 LiF/NaF/ZrF 4 55/22/23 24,51 23 .54 30.6 NaF/Zth/UF L 56/39/5 | 20.16 22.16 28.3 RbF/Zth/UF L L8/18/L, 3441 21.86 27 .8 LiF/KF/UF, 48/L8/L 51.42 20475 32.4 NaF/LiF/ZrF, 20/55/21/k 31.L46 22,54 29.3 /E, Average 34210 2412 313 134 Determination of C (1) = ¢ [1- o 2(K)] and € (900 °¢) for Chloride Salts of Interest Salt LiCl NaC. KCL RbCL CsCl CaCl SrCl Ba.Cl2 PbCl ’I.’hGl/4 ucl Ucl ‘s (cal/%c mole! 15. 16, 16, 16.2 18,0 23.6 2740 25,6 23.6 L2.5 3k. b2.5 Mol Ut 42,39 58,44 74456 120.92 168.36 110.99 153453 208.25 278,10 373.85 344,439 379.84 Table 343 = III ¢ [T ] p-mp (cal/"¢ £m 0.3539 0.2738 0.2146 0.1340 0.1069 0.2126 01703 0,1229 0.0849 0.1137 0.0987 0.1119 135 T 614 801 776 ( oy oy C C'D ey po C_zm_ 2.96 0.4799 0.3132 3.65 0.450%4 0.257°% 3.98 0.368, 0.196L Le13 0.2263 0,1197 L.O4 0.1701 0.0895 2.07 0.2713 0.2054 2,17 042267 061690 2.20 01688 0.1253 L.10 0.1213 0.0645 5.57 02713 0.0940 Le2h, 041873 0.0941 5.26 0.2049 0.0785 \\ L -‘.__-_ = . ] . 0.2 ~--—————————\ —— ¥~\—alkaline-earth -— L , L~ — chlorides ) e S S cp[%n°c] T, N , o | —_— - Ih \ - . - . 1 (cal/OC %n.)l -—-—*———-—j—'gflo‘:id‘.es s et , : - - i ! | = : { | 1 1 1 1 Molecular Weight Fig. 3.3=8 Correlation of Specific Heat with Salt Molecular Weight to Deduce C_ r900°c] for MgCl, e 136 3.3s1.8 Thermal conductivity. Figure 343=9 displays the measured thermal conductivities of individual salts, all from Russian sources [91- 94]e Por other salts and sal$ mixtures, one must{ rely on phenomenolog-— ical modelss Knowing rooa—~temperature k for a salt does not help in predicting k(T) of the liquid: Figure 3.3~10 shows how k changes with temperature as a sali pesses from solid to liquid state, Gembill ({95] proposed that two distinct mechanisms determine k for fused saltis: 1e Atomic or lattice conduction along the shori-ranzge atomic or molecular order present in liquids 2e Ionic trzusfer by e drift of free charged ions beiween the atoms, Gambill successfully correlated existirg sa2lt mixture (mostly flucrides) data at their melting points by caL/s - O.S O.B 1.8 ki 54 ion drifting predominates and this approach fails (£ mixtures of interest as core salt, Nan generally falls below 3, The atomic 1S Bily or actuzlly negative), However, with fused salt effect of complexing, e.g. 3KC1 + UClj—*-K§1 (IICJ_4)"3 will generally increase Havg’ however, For k(Tmp) of individual chloride salts Gambill's formula differs markedly from experimental Russian data (Table 3,3~IV and Figure 3,3-11), The consistent dependence of the theory correction factors with salt molecular weight (Figure 343=11) suggests (1) A different exponent than 1.8 for M (more like 0.8) (2) A slightly different correlation wita Navg For temperature dependence of k, Gambill [95] derived the following expression 139 "Table 3.3-1V Comparison of Gambill's Formula to Experiment for Individual Chloride Salts / Koo Rme em T al \ g «Pe m.pe /mC ncal Salt ‘Iavg. Lat. (K) M (g/ce) \s K cm)_ (;“F(cm)_ goate LiCl 2 0.258 887 42.39 1.50 8.09 3.61 0ol KCl 2 1049 7L.56 1.54 3.25 2.27 0.698 AbCl 2 988 120.92 2.24 1.78 1.79 1.006 CsCl 2 919 168.36 2.79 1.13 1.29 1.142 MgCl, 3 0.168 981 9s5.22 1.68 3.34 2.45 0.734 CaCl, 3 1045 110.99 2.09 3411 2,67 0.859 5r€l, 3 1146 158.53 2.73 2.12 2452 1.189 BaCl, 3 1236 208.25 317 1.52 247 1.625 140 correction curve TEET T LTI ll.llrll:rlrlrilml!_‘_. e = TTIIITT ‘.”Ha.w_wltx- T T s e e s S - ; TR T o ~|1.]1/]'eyeball average 240 Lo .- - - EEEE - [ S ~ : . - B g - . [ : - o - — - - v . S ae- . - T et b en e = el o - ‘l/ ST T o L T T e e N S Vo s e e T o e oL oTouTTL ;n_LHu.'..,H.‘_A oI o — - / mou - At . — - ——— —d e -l T T I e = A LT LT/ I TTIIIIIT LTI e\ ——— o - ot e = A\ Y= — b e — T e T L Tl I, N — oo i e b g e T e e i L L rm i el T LT N\ T 1&6 ’ Exf;er;ment - to -~ Corr:‘?:’h;n Factor (hEXP /hctk) at T;np. Theo 141 Salt Molecular Weight Fig. 3.3~11 Correlation of Experiment—~to—(Gambill's) Theory Differences with Molecular Weight kt-otal (T )= ®ygea (T2) g‘pfimuc /;ZZD:) [5;2‘3} +(1- &m;c)[m)][ ]/1 The usefulness of this formula depends upon the confidence develooped in it; Cambill had no data with which to test it. Attemots to fit the availapble (Russian) data for individual szlis fail drematically, Inspection of the terms reveals the cause: k. . (first term) ore- dominantly depends on/oi/2 which has a negative T-coefficieat while experimentally dk/dT is posiiives For K: onic (second term) the YT factor overrides the negaetive T-devendence in/o, but only up to a certzin iemperaiure. Overall this approach falls. Perry's Chemical Ingineering Handbook [96] describes numerous other formulae but these generally pertain Yo organics and other non- related fluids; also they either do not freat temperature dependence or they grossly disagree. In Table 3.3~V the electrostatic~bonded alkali and alkaline earth chlorides of intermediate and heavy molecular weight appear to follow %: = 0,024 x I ¢ V(m‘.-I/m°K2) where i1 = molecular weight V = cation valence For highly-covalent UCl4, dk/dT = 0,0023 x MV(mH/sz), a magnitude amaller. For other covalent molecules, e.g; SiCl4, k actually de- creases with T. As PbCl, belongs to this chemical class, its dk/dT w wn 6 W . 104 & Maxmum [+ ' fuel temperature jn this reactor -4 10 - ) -1 10 200 400 600 800 1000 1200 1400 1600 TemperAaTURE [°C] Figure 3.3-20 Vappr Pressure for Metal Chlorides [17] 153 2400 2000 On (& o 1200 800 B, I’ Temperature (oc) Toimeme Alkaldes =mmmmimine sy T} Earths o e ("1 chemi e == '_:r'fi.‘.‘ St (+2 chemi- 7 P TIEE T T 0y state) o i e o e T e . ) S S e e oAl state) = S e et e e e e e e e e e 400 {=isor= i 1= "—r —o ==t B Atomic Number Z of the Salt Cation Fige 343=21 Boiling Points for Halides of Alkalies (Li, Na, ¥, Rb, Cs) and Alkaline Barths (3e, Mg, Ca, Sr, Bz) [97] 154 3.3.2.6 Chemical behavior. This section surveys the chemical stability of the halides in the system (tendency to break up) and their corrosiveness (tendency to chemically attack the structural materials in the primary circuit at the temperature of concern), Figures 3.3-22 compares the stability of some chlofide and fluo-— ride compcunds: the lower the A G, the more stable. NaCl end iir ere particuiariy good. In general chlorides are less stable than fluofides. This pattern of decreasing stability with nalide atomic number continues through bromine and iodine. The smell AC values for MoCl, and MoCly inhibit their fermatien, wiiicn is why Ho is a preferrei siructural materiat in a chloride system. Graphite also resists chemical attack (fcrma%ion of CF, and ocL,) well. The appreciable AC for halides of Fe, Cr, and Ni makes them unsuitable as structural material in contact with salt. Chlorides generally dissalve in H,05 fluorides do not. Inside the molten salt circuit the presence of oxygen or water will produce oxy—chlorides (CCl) which corrode. Cutside the reactor, water sclu- bility may facilitate cleanup of leaks or spills. AG for other halides is less available; they are generally less ’ stable, 155 961 Hymerical Yaluea givan for 40 1100 4 Fuel Componenta -100 nalBOOK -IOOfl (KJInolF) -370 BaF L up § -UF! - b= - %00 ?UFB, ;nP“ -5004 -€00 ~ LIP L0Gux in Struotural Matarials HBP“j b iLe Crf, I~ Fef, C'P5fi ZLLP-ae?z 1oiucion Plsslon ?roaucta aa?3 i fiuP: JDPu 1F Tep 3 zre, ciPp 4 . w Lmtt [~ erp Laly, -4 UOF’ Ser -1 Flgure 3.3-22, Free Energy of Formation + 100 + -100 { 6ulOOOK -200 + (xJ/moiC ) =300 -hoo 4 -5C0 =b0u Puel Components | ucy, ThClH e, — w PuClylpact 'Amcl) — NaCL < Secuctural Materials noc1z1t“°011 [ cciy FP:Clz CrC12~ Pission Producta TcCIJ* Snc12,F CeCl, | 0T LlLl’j SeC1, 4 Hact, "} PdCl2 SeCl TeClz irCk PrCi SmCl2 RuCl CacCl at 1000 ¥ for Fluorides and Chllorides [17] 3e3e2s] Cost and availability. The abundance of seawater provides obvious cost and availability advantages for chlorine over Zluworine, The scarcity and cost of Be for a F/Li/Be mix (necessary to achieve low F melting temperatures) tilts the scale even more in favor of Cl. 3e3e2.8 Density, The power needed for pumping the core salt depends directly on its density; heat transfer Li > (Be, K, Ca) ol Table 3.3-IX Neutron Capture Comparisan of Candidate Carrier Salt Elements [98] IFisgion Core Infinitely-Dilute Cross Section (barns) for . _ Spectrum Spectrum Croup Neutron Energy Be Na - Mg K Ca Li=6 b () (%) 1 6.5 -~ 10.5 MeV 0.030 0.050 0.060 0.330 0.02 2.06 0,000 1.7 0.501 2 LeO = 6,5 0.070 0.005 0.003 0.240 0.01 0.10 0,000 9.2 3.18 3 2.5 = L0 0.095 0.0002 0.0002 0.150 0.00 0.16 0,001 18.6 8.46 L 1.4 - 2.5 0.04,0 0.0002 0.0003 0.065 0,00 0.25 0,001 27« 17.15 5 0.8 ~ 1.4 0.003 0.,0002 0.0004 0,025 0.30 0.003 20. 19.92 6 0.4 = 0.8 0,000 0,0003 0.0004 0.003 0«50 04004 14. 22,03 7 0.2 = 0.4 0.0006 0.0003 0.00} 2.00 0,006 6, 14.42 8 0.1 =02 Y 0.0012 0.0040 0,006 0.95 0,006 2.3 7.86 9 465 = 100 keV 0.0016 0.0005 0.009 0.70 0,005 0.9 4.0 10 21.5 = 46.5 0.0026 0.0005 0.014 0.85 0,004 0.3 448 11 10 = 26.5 0.001 0.0005 0.005 1.20 0,002 0.1 - 0.778 12 L,65~10 0.001 0.0001 0.033 1.0 0,001 0. 0.133 13 2.15=4.65 0,100 0.0002 0.200 2.60 0.001 0. 0.0372 14 1.0 =2.15 ¢ 0,000 0.0003 0.009 3.90 0,001 0. 0.0110 15 465 =1000 eV : 0.005 0.0004 0.013 + 5,70 0.001 0. 1.72-3 FSA(b 0.036 0.0016 0.0017 0.079 0.0013 0.029 0.0022 csaibg 0.018 0,0009 0.0010 0.041 0.0004 0,046 0,0035 Canclusion: (Li,K) > Be > ( Pb, lig, Na, Ca) * in natural lithlum Table 3.3-% @ =-Values for Scme Neutron Reacticns with Candidate Alkali and Alkaline Earth Isotopes [99] Kat . Q-value (MeV) for Target Abund. (n,2n) (n,p) (n,o ) Isotoe () Reaction Reactin Redotim b 7.5 5.7 2.7 4.8 i 92.5 7.3 -10.4 unknown %Re 100. -1.7 . -12.8 -0.6 “Na 0 ~11.1 €3.6 +1.95 23N a 100 —12.4 -3.6 3.9 2hg 79 -16.5 L7 -2.6 Mg 10 =7.3 -3.1 +0.48 25y 11 ~11.1 -7.7 =54 3% 93 -13.1 +0.22 +1.36 % 0.012 7.8 +2.3 +3.9 bl 6.7 -10. -1.7 ~0.10 k0ca 97 ~15.6 0.5 £1.75 tha 0 ~8.3 +1.2 +5.2L Lca 0.65 -11.5 -2.7 +0.35 b3ca 0.14 ~7.9 ~1.0 +2.3 bhoq 2.1 -11.1 —4.5 2.7 4ca 0.003 ~10.4 6.9 unknown 4ca 0.15 9.9 unknown unknown 165 Table 3.3=% (Continued) Nat, O=value (MeV) for Target Abunde (n,2n) (n,p) (ny A ) Isotove (%) . Reactian Reaction Reaction 85 72,17 ~10.5 0.11 0.99 8Tab 27.83 -9.9 3.1 —1.2 Bl 0.56 ~11.8 -0.10 0.68 &gy, 9.9 ~11.5 ~0.99 1.13 &sr 7.0 -84 0.51 3.21 8881' 82.6 -11.1 —lyeli6 ~0.76 Blhes 2.0y 6.9 2.1 6.2 B3cs 100 ~9.0 0.35 L.38 DG, 0.10 ~10.3 0.34 6,76 B2, 0,095 ~9.6 ~0.1,0 5,96 BLB& 201-(» "'9.3 -1028 2-21 B, 6.5 7.2 0.57 6,94 1365, 7.8 ~9.2 -2.0 o2l B%, 7.9 8.5 “4.0 3.77 0Ly, 1.4 -8.2 0.02 8.20 206p,, 2401 -8.1 ~0.74 7.12 207py 22.1 6.7 ~0.65 7.89 208y, 5244 7.4 .2 6.05 2 3 Lo 5. e Te Fe (75%) CLi(n, o)t fi%fl—dae (92.5%) Lil(n, Y)2 “Ee 6 (100%) “Be(n, ¥ )03 lebxl10 ¥ . 105 3 (1003%) 23Na(n, Y)24Na 1-2—;—1‘_?—»241@ " (n, 2n)22Na gif}igl-i'-gzNe (11.01 73)26Mg(n, \(,)27,145' 2:/35_{’.“&2_*‘27 AL (9343%) 39K(n,p)39A §§25§;~;-39K / 5 " (n, )30 ZTEIEL 36, 25 AT (6475%) 41K(n,\/)42K l2hr 4%, /’d— Figure 3,3=24 Priccipal Weutron-induced Reactions Which Produce Impurities and Radioactivity from the Candidate Cations Netural abundances are in parentheses 167 10, 11. 124 13 14. 15« 16, 17 184 19. 204 21, 224 0 d (96494%) 4OCa(n,o< )37Ar _3_4_8_.....3701 8 (72.17%) 2930 (n,20) %m0 4L 2% o €\p Y, 8 i (1, )36 18265 486 ? (fib/ e e BB %y (27.83%) 87Rb(n,2_n) Rb 18,854 5% Kr 89.1, Ve F ,a’r 504 5d 8g' - C e e )G, e B / . ‘ / " (n, p)87KI' 76 m 87 Rb Sy " (n,N)BA'Br 31:8 T - 843'.1‘ & (82.6%) Bsp(a n, ¥)%%zr(a, v )% 2L - ¥ 504 5d /? 1 13 d 6 (1och) '33cs(n, Y ) 34 s(2, ) 3%¢s(a, ¥) 20cs -/—;—?-- 35, 3 ; AE206y ' 23x106y " (n,2n)!3%Cs ——-—-—-—6 474, 132z, ¢ 1325, /2 0" (n, p)133){e _2%\/_(1_._13305 1€8 23 2. 25. 26. 27 . 28, 29, 30. 31. (roag) Pies (m, &) P e P 4 r (11.2%) B37pa (n,p) B7ag 2017, U7, pY (71.9%) 38a (n,Y) P%a B2 ming U9, P Y v o . (nyal) 135Xe _2_'3___, 135Cs Py (1.4%) 20kpy, (n,od) 201Hg 206 (24.1%) 7 Pb (n, k) 2Py i‘i—??—» 2033, (22.1%) 2Teb (n,o) Vg (52.4%) 2%Cpb (n,«) POhg 220 o OO /e Y 169 ically-aggressive AlCl3 as well as a small change in valeacy. Trans— mutaticn of K produces long-lived radioactivity and divalent calcium. The magic proton number of Ca causes the existence of numerous stable and quasistable Ca isotopes. These mostly transmute into themselves, yielding nil radioactivity. Rb produces long~lived SSKr and 9OSr plus considerable short-lived radicactivity and different elements (especially Sr). Sr irradiation 90 principally leads to * Sr. Cs tramsmutaticn products appear innocucus. Ba will directly produce 13705. Protm-magic b will generally produce stable nuclei but highly-exothermic (n, ) reactions will yield chemically different Tl and Hg. 3.3.5.4 Melting point. This section relies on the limifed avazilable fusibility diagrams for salt mixtures [87, 100-322]. Most of these come from terse articles in remote Russian journals; some disagree with one another, This limits us to only qualitative conclusions. Low core salt melfing-point helps avoid high ocutlet temperature (e.ge > 1000 C). Figure 3.3-25 displays the approximate melting points for UCl3 mixes. Alkali chlorides tend to form lower temperature eutec- tics with UCl3 than do alkaline earth chlorides., The bumps signal com- plex molecules like K3U016 (3KC1'U013). Desyatnik et al [100] verify that complex uranium groupings increase in frequency as the alkali- cation radius increases, 170 BeClz ¢ UCY, Se(] oY 800 - &% T00 - $ N\ / - 1/, T +2 5 2, I\ ‘ = = 2 500 - ‘ 0 20 40 60 8b 100 lolar Pez_:'cent Ucl, in IrIJ:rture 3 Fig. 3.3~25 Pusion Diagrams for Binary UC 1T l3 Salt Mixtures Many of the eutectic melting points occur at lowUCl3 molar content. In Contrast; hard spectrum and high BG potential favor high actinide cantent. Table 3.3-XI shows the possible UCl3 molar contents under various temperature restrictions. RbCL excels for a SSOOC limit; KCl or RbCL for 600°C. At 650°C all alkali chlorides show potential, though KCl and CsCl excel. Alkaline earth chlorides don't work until the low temp-— erature limit exceeds 7OOOC. If one wents a high~actinide 30 NaCl/70 UCl; mix then the lowest core salt temperature allowab.e (with a little safety margin) is 7500C. With a 50/50 NaCl/UCl3 mix th: temperature can drop to nominally 650°C. Section 3.3.1.3 mentioned that ternary and higher compoment mix~ tures can further lower the melting point. Figure 3.3-26 presents four ternary diagrams for UClB-UClthkCl, vhere Ak=alkali or alkaline earth. Cne desires to optimize the U()lB/'UCll+ ratio so as to hald down temperature (which affects corrosion) as well as the content of corrosive UClh. Keeping the U/Ak ratio high maximizes spectrum hardness and BG potential. Unfortunately, the isotherms and the eutectic mep. tend awey from hizh gg&B cantent., Ternary mixtures with two AkCl components behave similarly (Fig. 3.3-27). In summary one must tradeoff low melting point (low U-~III content) and corrosion (high Ak content) against large spectrum hardness and BG potential (high U~III content). 172 Table 3.3-XI. Candidate Core Salt Compositions Under Varicus Temperature Restrictions Lowest Temp. UcCl Lowest Temp. UClL to be En=- Molgr to be fn- Molgr countered Carrier Content countered Carrier Cantent _ (%) Salt (%) (°¢) Salt (%) 500 RbCl ~15 700 LiCl 0=560 530 LiCl ~25 NaCl 18~60 NaCl ~ 35 KC1 1068 CsCl 25 RbC1 5-70 RbCl 1545 CsCl LO=70 | MgCl, ~ 38 600 LiCl 10-40 CaCl, 25-55 NaCl 30=45 SrCl, 60-70 KCl 50~60 RbCL 15-55 750 1iCl o-70 CsCL ~55 NaCl 1275 KC1 6=75 650 LiCl 050 | RbCl 0=75 NaCl 20-55 CsCl 0=75 KC1 15-65 MgCl, 0-50 RbC1 8-40 CaCl, 15-65 CsCl L5565 Srcl, 50-80 CaCl, BaCl, 70-75 173 ] 2.'-'1(:( - LL:I & 358 e, \ 335" 52 O Ul Fig . 3 -3‘-26 Ternary UClB/UClL Fusibility Diagrams 174 The blanket salt will generate little heat; therefore, a low m.p. 1 temperature will facilitate its handling. High ThCl, content maximizes BG there. Carrier-salts whose eutectic mixtures with ThClI+ (Fig. 3.3~ 28) seem to qualify are PbCl,, LiCl, NaCl, and KCl. They also form suitable eutectics with the small amownts of UClh present (Fig. 3.3-29). PbCl2 has a sitight disadvantage in that Pb is divalent, further in- creasing parasitic capture by chlorine. Under an arbitrary operating design limit of ane U atom per 10 Th atoms and low Ak content (to avoid blanket power cooling and BG loss) ternary UClA/ThClh/AkCl (Fig. 3.3-30) mixtures do not significantly lower the meiting point: all eutectic points lie at high UClh and carrier salt contents, 175 Melting Poin* Temperature (°c) 900 800 700 600 500 4C0 300 1 20 40 Molar Percent ThCl 4 60 in Mixture 80 100 Fig. 3.3-28 Fusion Diagrams for Binary ThCl4 Salt Systems 176 Melting Point (°0) 700~ 600~ S _,,, \oiififir' 50074 2 % 0\ N 1 400~ 300 : : ; : 0 20 40 60 80 100 Molar Percent Actinide Chloride ThCly UCl, Figure 3,3-29, Pusion Diagrams for Binary Mixtures of Ac(IV) and Carrier Chloride Salts 177 Fige 3.3-30 Ternary UCl h/ThGlL Fusibility Diagrams 178 Addition of a second carrier salt does lower Tmp (down to 325- T 335° ¢) at ~ 20% ThCl, content (Fig. 343=31); however, though 'Li and 4 Pb may not be too absorbing, the high Cl/Th ratio will damage BG and neutron spectrum,. Lict Fig. 3¢3- 31. ThCl 4/1,101/?11012 Fusibility Diagram Fluoride blanket salt presents another altermative, TLip (riz. 3.3-32) could best lower the molting point and thermalize the blanket neutron spectrum without parasitically capturing neutrons. NaP looks next best, However fluorides solidify near TOOOC and eutectic effects do not appear until ThF, molar content falls disastrously (to BG) low 4 to 200 (Fige 3e3-32); chlorides melt much lower. 119 Melting Point Temperature (OC) 1100 1000 900 800 700 600 500 A 400 -+ 300 T v ¥ 0 20 40 60 80 100 Mol % ThF4 in Mixture Figure 3,3-32, Fusion Diagrams for Binary ThF, Salt Systems 4 180 Fig. 3.3-33 ThF LL/L iF/BeF, and ThF L/N aF/ZrF L Fusibility Diagrams LF ?YI:FI LF lfl'hfi THF‘ s TEMRERATURE iry *C CUMFOSITION 1N mate 7, TEMPERATURES ARE IN *°C CCMPUSITION IN mow % Minina 870 £E-618 -443 _— of &S " Siy QA /™ N AN N w° L ey g v A &) R o > he) . N i ‘ Naf -~ = ' o ~ ~ 2F, 290 : T %2 a - 92 181 The eutectic points which minimize KCl content are 5.5 u013/u5 UClh/ 49.5 XC1 for a m.p. of 314°C, or 17 UC1,/26 UCL, /57 XC1 for a mep. of 518°C. With NaCl ane might opt for a m.p. of 339°C at 17 ueL,/ 4.5 UCL, /42.5 NaCl or 432°C at 21.5 UCLy/29.5 UCL, /49 NaCl. The CaCl, ternary offers no attractive eutectic points. Nane of the above points have UClB/UClh> 1; that could spell corrosion troubies (Sect. 3.4.2.2). Still, temperature affecis cor- rosicn as well, so that high UClh courd yet be worthwhiie. Of the five cases mentioned above the mix with mep. 33900 locks the best. Ternary mixtures of ThFh/UFh/AkF behave like their chloride counterpart: no eutectic poimts occur with high ThFh content. Mixing in two carrier salis with ThF, doesn't help either (Fig. 3,3-33): +the well known MSBR salt Flibe Ber/LiF) combines with ThF, or UF, %o reach melting points of 400~500°C, but only at 12 ThF4 (for 50000). Flibe also suffers because Be is scarce and Li produces T unless enriched in 7Li, Figze Je3=6 showed aigher halides %o have lower melting points, but they nave other problems, 3u3ele5 3Boilinr noint and vnvor nressurz, Chlorides of Cz, X, la and Mg all boil in the same range, 1400=1600°C, Their differences are insignificant because UCl4 (792°C) and ThCl4 (92800) pose much lower limits. The boiling points of UCl, and PuCl, are unknown, 3 3 but their melting points compare with those of the carrier salts. 182 3.3.4.6 Chemical behavior. Halogens react vigorously with the al— kali metals to form a very stable binary ccmpound. Salts of alkaline earths also exhibit large free energy for formatien (Fig. 3.3-12). Fb halides are not so stable: O0BW. [13] found PoCl, 4o react with the structural materials available then, They zlso found Zrél2 to cause 2 Snowy Ddrecitiiate. In addition to having high melting points, alkali halides conduct electricity in the molten state, and otherwise behave like electro- valent compounds. Uith the excepticn of lithium fluoride, they readily dissolve in water. 3.3.h7 Cest and availability. NaCl is very abundant and cheap. This reduces both direct capital and fuel cycle costs; it also allows other options such as substitution of fresh for radicactive haCl. 3.3.4.8 Density. The densities of the principal (lighter) candidate carrier salts (Table 3.3-XITI) differ little compared to those of ThCL, UClB, and UClh. Due to differences in molecular weight, a 50-50 molar mix of actinide and carrier salt means about an €515 weight mix, respectively. Thus the actinide salt density controls the final density. 3,3.,4.9 Heat Transfer Coefficient. Section 3.3.l1.4 pointed cut that _ 102 -0y A 03B | OFT h:g /J/ll CF }l Section 3.3.4.8 showed that the actinide salt controls the density. Sectiom 3.3.1.6 suggests thafiyu(fuel salt,hixture):/AA(carrier salt). Fig. 3.3-5 suggests avoiding most of the alkaline earth salts: 183 Table 3.3=XII Density (g/cmj) of Chloride Salts at 800°C Salt | Desyztnik et al [83] Kinosz and Heunip (84] LiCl l.42 1.46 NaCl 1.55 1.61 KC1l 1.52 1.57 RbCli 2.16 CsCl 2.62 PbCl, Lel2 ThCJ.A 3.36 U(.‘.L3 L.87 UClI+ 3.17 BeCl, 1.10 MgCl, 1.65 1.68 CaCl2 2.07 2.11 SrCl2 2.77 BaCl2 3.28 the alkalil chlorides and PbCl 5 offer lox-:ez;/uvalues. In Table 3.3-IIT1, the lighter chlorides offer the higher specific heats. In Figure 3.3=9 they alsc exhibit higher thermal canductivities in the 700-9C0°C range, CsCl is an exceptim. Ignoringja because the actinide density contrels, a heat transfer figure—~of-merit (FCM) for carrier salts is CpO'33 k0'67 L.'O°b7 , 184 The results for Section 3.3.1 data (Figure 3.3-34) and the above discussion recommend the lighter-weight—salts, and all the alkali chlorides. A FCY including the density factor‘fp‘a (Fise 3¢3-35) aczin discourages only the heavier alkaline earth chlorides, CsCl and FhCl, look very good then. In summary, LiCl, MgCl2, and NaCl look to be the most attractive, 2 sive,; however, as i1 generallrs does not include the effects cof actinide in +hz% order. SrCl,. and BaClz}the least. This study is not conclu- caloride a2dmizture, 3.3.4.10. Choice in other studies. ORNL in 1956 [13] chose a 3 NaCl/ 2 MgC12 mix based on the:eutectic point between these 4wo, However, as pointed out above, the advantage of a second carrier salt disappears when the actinide salt predominates. Similarly, ORNL in 1963 [§] and ANL in 1967 [21] reported studies on carrier salt mixtu;es of NaCl, KCl andegClz but with only 30 and 18-50% mole of actinide chloride present, respectively. Taube [17] generally assumes NaCl alone as carrier salt even with low actinide contents. He mentions alsc further study of NaF/ZrF,, but Section 3.3.2.9 precludes F from the core. The British [29] also choose NaCl by itself., 185 - - - l‘ — e ——— e e _ —— —_— — - - —— - —_ Coae — e ——— e ——— e — e — — - ———— e L C(Jf:fh chlorides | alkaline 0.6 0.8 [900°C](ca/g °c)%*3° p . x k[850°¢](H/mix) 07 x)u[85000](cp)_°'47 C 186 Molecular Weight Heat Transfer Coefficient Figure-of-Merit for Chloride Salts of Interest Figure 3-3"340 T T ke 1.2 ST ©OMglly . o _ _ S chiorides __ _ 100 _______X NOCI . 0.8 T T alkaline earth T T T T T T oottt T \chiorides T I T 40 80 120 160 200 240 280 Salt Molecular Weight Figure 3,3=35. Heat Transfer Coefficient Figfiré-of-Merit for Chloride Salts of Interest, Including Density Foll = » [850°c] (/o) *° x C, [900%¢](ce1/5%) "+ x k [850°Cc]W/m )0 €7 x/u,[850°C](cp)"o'47 187 3e3+4+t1 Summarve Table 3.3-XIII summarizes the” intercozparison of carrier salt cations. The goal of keeping the core spectrum hard dis- courages use of Be and Li there because of elastic downscatter; Ca and X dovmscatter the least. lNa undergoes the most inelastic scatter; Ca and ¥, the least. Parasitic neutron capture though small, exceeds by magnitudes for Li and K over that for Ca, Mg, and Na. Ca transmutes the least into radioactivity or troublesome chemicals. Caticn choice on the basis of eutectic melting point depends on the range of acceptable actinide molar contents. Boiling points far carrier chlorides olay no role as they far exceed those for actinide chlerides. . The chenical stability of the 2lxali chlorides (NaCl and KC.L) surpasses that of the aixaiine earth chiorides (i-lgCl2 and CaClz). laCl abounds by far the most in nature and costs the least., Salt density affects pumping power needs but the higher densities of the actinide chlorides dwarf any differences due to carrier salt choice. The light alkalis transfer heat the best. In summary, ¥aCl costs little and exhibits good physical and chemical properties. K and Ca feature better nuclear properties for the core sait but the much higher Cl concentration obscures them. Na and 7Li look best for the blanket sali. Final choicc must weigh these relative advantages plus the location of eutectic melting points near the desired actinide molar fraction. 188 Table 3.3-XII1. nuclear properties elastic scatter inelastic scatter neutron absorption radioactivity impurity mutants overall nuclear physical properties mop- bePe Vapor press. heat transfer coef. chemical behavior salt stability economics cost disposal overall comparison core blanket Intercomparison of Salt Cations rating according to core salt goals good intermediate poor K,Ca Na,Mg,Pb Li,Be K,Ca,B Li,Mg Na,Pb Mg,Ca,'Li Pb,Be Li,K Na 9 Ca,Pb g, Li,Cs Li,Be,Na,K Rb,5r,Ba Ca,Sr,0s K,Na,Be,Ba Mg,L1,Pb Ca K,Mg,MNa,Pb Li,Be K,~a,Li,Pb Mg,Ca Be al. o.k. Li, Mg Na,f,Ca Ba,dr, Pb ,kb,Cs Na,K,Rb,Cs Mg,Ca,Pb Sr,Ba Na Ca,X,Mg,Li,Pb Be (same as radioactivity) Na 7 . K,Ca,Mg,Pb,Sr,Ba Li,Be Na,'Li Mg,Ca,K,Pb Li,Be 139 3.3.5 Choice of Helative Proporticns of Actinide and Carrier Salt . In the core we want to avoid nuclei which neither fission nor breed, but only dovmscatter and capture neutrans. This means maximizing the fissile enrdichment and the ratio of actinide-~to—~carrier salt. It is also important to maximize the fuel salt heat transfer ccefficient at an average fuel temperature of 800 -85000; hf comprises the parameters D ., C and k {Section 3.2.5). 'f’ T/A y p! £ ( ) 3.3.5.1 Feutren physics of the ThC.LL/‘UClz ratic in the core mixcurs. Considerable ThCld inherently resides in between the core tubes as blanket salt 2nd interacts. neutronically with the czore. ~However, as a first approximation, Sectioa 3,951 ignmores that modellinz heter- r 1/ ogeneity as well as that of the multiple core tubes. Thus the ThCl UCL, ratio here refers solely to that within the primary loop. The addition of ThCla to the core salt mixture should 1. Markediy increase the critical core radius, thereby decreasing the power density. This causes a. the core flux level t¢ decrease, which (1) extends the life ¢f the core tubes, and (2) decreases the damage flux to the vessel, which ex- tends its life be. the primary coolant power density in the heat exchanger to decrease, which (1) decreases out-of-core fuel inventory (2) eases heat exchanger stresses 2. Lead to an optimum Th/U ratio where the primary circuit fis- sile-inventory is a mizimum. The 1956 ORNL study [13] obtained a similar result Je Increase internal (core) breeding, which ae reduces the necessary size of the outer blanket be reduces the import of a reactor 4. Soften the neutron spectrum, which 2e decreases the BG potantial of the reactor be 1nhibiis the useful consumption of higher zctinides ce nullifies z smzll part of the flux decrease from the dilution because the averege fission cross sections become smaller, In sunmary there are both pros and cons re Th presence which are difficult to quantify on a ccmmon basis., However, Tualitatively, 3G potential and the useful conzumpiion of the actinide§ argue for low Th concentration ([Th]) while [Th] ——== O could mean too freguent z replacement of the core tubes, Practically, reactor operation could start with high [Th] and wide tubes and later change to low [Th] and smaller tubes as confidence in the tube wall life develops, This ver- satility would probably require only the inclusion of a few extra tube ports in the reactor vessel, 3e3e5+2 Density. This study principally involves U013’ Nall, and ThClq. NaCl molar content may vary only between 30 and 70% to achieve eutectic melting point lowering, Pigure 3,3~36 shows the density for various ThCl4/U013 ratios using the densities and the prescription for their addition from section 3¢3e1e5e For a Pu013/2 UCl3/3.16 NaCl mix, Taube and Davudi [123] listed p =292 g/ce at 850°C; for 15 PuCl3/85 NaCl, 2 = 2.12 gfcc. These values both seem low compared to Fige 343-36 (though different actinide salts are involved). In the same repori these authors quote /cr= 3,526 g/cc for a 1 Pu013/2 UCl3/3.65 ¥aCl mix at 850°C, which seems nighe The 1963 ORNL paper [6] listed a density of 3.08 g/cc at 643°C for a 45 NaCl/25 KCl/3O (PuCl3 + UCl3) mix, Since density decreases vith increasing temperature, that value supports Figure 3.3-36. The earlier 1956 ORNL study [13] used /0 = 25 g/cc for 2 3 NaCl, 2 HgClz, 1 (UCl3 + PuCl3) salt, which also seems consistent, 3434563 Viscesitye 5Section 3e31.6 recommended //% (mi:ture):i/bt(carrier salt) Assuming JaCl carrier salt, -hen Ifrom Figs 3.3=5 at SBOOC //LL= 0.012 gn/s ca For a 1 PuCl3/2 8013/3.65 HaCl mixture, Taube and Dawudi [123] assumed 2 much higher//%-= 0,05 (1-0,001(1=850°C)). Similarly, at 649°C OMIL used 0,0662 + 0.0080 for a 45 NaCl/25 KC1/30 (UC1, + PuCl,) salt in 1963 [6] and 0,10 for 2 3 NaCl/2 MgClz/ 1 (UCl3 + PuCl3) salt in 1956 [13]. However, neither Taube nor ORIL indicate any experi- mental basis whereas the//%.recommended nerc is basad on recens (Russian) measurements [124-125]s This reduction in:/utby more than a factor of three increases hf vy roughly 50% over previous studies, 3e3e5.4 Specific heat, From the study of sfiecific heats for 3- component salts (Section 3.341.7) o 8.1 cal/°C (2m2 + 4m4 + 5m5) c (cal/g C) = X = P yrieDe atom mole m2M2 + m, M, + m.I 44 55 and Cp (T) = Cp, (-7 (X)] 192 [ThCl4]/[U013] Molar Ratio Density (g/cc) Holar Content (%) (TaCl, + UCL,) Figure 3.3-36 Density at 850°C for NaCl/ThCl 1+/UC}.3 Mixtures Based an Janz 5t .Al1 [112] Data and Linear Addition of Specific Volumes of Component Salts. where m2 = mole fraction of the 2-atom alkali carrier salt m. = mole fraction of the 4-atom UCl3 m. = mole fraction of the 5-atom ThCl 4 Moyl s Mo = corresponding molecular weighte 0 - -— cC =¢C 1T K o= % o /01-=T (K)] o = average coefficienti of linear expansion = Z mil)(i 1 -4 q - O(; = 34655 4.23, and 5.57 x 10 °¢™" for WaCl, UC1,, 2nd ThCl 4 Figure 3.3-37 shows the results at the Tmp for each material.. For a Tixed temperature, Cp changes much less with molar composition. Taube and Dawudi {123] assumed c, ; 5.094 and 0,068 x (1-0.0005 (7 -850°C) ) cal/em®°C for 15 PuC13/85 NaCl and 1 PuCly/2 UCL,/3.16 ¥aCl mixtures, fairly consistent with Figure 3.3-37. However for a 1 PuC13/2 U013/3.65 ¥aCl mix 2t 850°C they used 2 rather high value of 0s24 cal/gm C. This agrees better with the 1963 ORVL [6] value of 0u3+ 0403 cal/gn’C at 649°C for a 45 ¥aCl/25 KCL/30 (UoL, + Pucl,) mix and the 1956 ORML [13] valiue of 0.2 cal/an’C at 649°C for a 3 Wacl/ 2 Mg012/1 (ucl, + PuClB) salt, The values used in this work are intermediate to the above extremes, Although they admittedly contain large uncertainty, at least Section 3e3e1e7 describes their basis; Taube and ORNL do not describe theirs, 194 ‘Th{l /Uc13 N 0 Lo o — NaCl Molar rracvion Figure 3.3-37 Specific Heat at T, for NaCl/ThClb/UCl3 Mixtures 3e3e5.5 Thermal conductivity. Section 34341.8 estimated k(T ) = 0,04308 07 0.8/ 18 ¢ cal/s cm K o/~ 7* mp /Omp atomic ' k (7) =k (Tmp) + 6.0 x 107 (T—Tmp) MV and ’ Table XIV shows.the oalculated[k} at T=p_ D ; T=T 4+ 100°C, and T-350°C. The higher k~values seem to occur for high NaCl and ThCl contents, A 195 961 Table 3.,3-XIV Calculation of Thermal Conductivity for NaCl/‘l‘hClL‘L/L_J_Ql3 Mixtures NaCl k ( gg 262%‘ ngizit ThCr%! /Uang Tmp /:n.p : _ a}t at'. Tmp at ( 9) Ratio (C) (g/cc) M N v f atomic Top 4 100c 800°C 30 0 740 h37 295.1 3.40 0.132 1.57 1.94 1.97 50 0 610 4.02 198.9 3.00 0.168 1.69 1.93 2 .26 70 0 540 3.35 142.7 2.60 0.204 2.10 2424 2.52 30 0.5 (713) 3.96 263.2 3.633 0.111 1.61 2.03 2.18 50 0.5 (540) 3.85 204.7 3.167 0.153 1.63 1.90 2.46 70 0.5 (502) 3.25 146.2 2,700 0.195 2.00 2415 252 30 1.0 (700) 3.81 267.2 3.750 0.100 1,68 2,12 2434 50 1.0 (505) 3.79 207.5 3.250 0.146 1.61 1.89 2.58 70 1.0 (4,82) 3.21 147 .9 2.750 0.190 1.96 2,12 2453 30 3.0 660 3.66 273.2 3.925 0.0848 1.81 2429 2472 50 3.0 400 3.83 211.8 3.375 0.1343 1.59 1.89 2.95 70 3.0 L25 3.21 150.5 2,825 0.1838 1.89 2.05 2459 Table XV compares some of these {k} (with T-600°C) to those assumed in other studies, ordered by increasing actinide salt content, ORML {13] and Taube values seem rather high, Close agreement with the other |k{ (z2 x 1073 Cal/s cm K) lends credence to the values assumed in this worxe. Jele5ab Heat transfer coefficient. Yith the above bases, and assuming (1) The alzorithm developed for h,. in Section 3¢2.51 f (2) A salt velocity in the heat erchanger of 14 meters per second (3) An inside tube dizmeter of 0440 inches (4) u= / A (for these survey purposes), 7 well my s 3~ < i X - 3 Thne film coefficient for various mixtures and temperatures becomes h_= T.37 i]A)O'B (10-3)0'67 2 (g/cc)o'8 k (‘IO-3 cal/s cm ¥) T Qe dT Oe2 (1.46) x (1.016) < Cp (ca.l/g K)0,3,3. 0.5367 The results (Table 3.3-XVI ) differ very little with mixture or temperature., Considering the approx:mations made for the input para-~ neters, no significant trend is detected. Roughly hf = 1,2 cal/sec cm2 X The only other velue available for comparison is the ORNL [13] hf = 0438 cal/s cm2 Ke Table 3,3=XVITI shows fthat the factor of 3 increase here is predominantly due to a doubling of fuel wvelocity and an 8-fold decrease in viscosity. The latter is mostly duve to more recent information and higher temperature, and partly contingeni on picking a composition with eutectic-like effects on viscosity (SEC. 3.3.1.6). 197 Table 3.3=XV Comparison of Calculated Thermal Conductivities to Those Assumed in Previous MCFR Studies Entries arranged in order of decreasing carrier salt concentration k Carrier Actinide (IIO-BCal Study salt salt (%) sec cm k ) CRKNL/56 [13] 50 KNaCl 17 (OC1, + 649 el 33 MgCl, ' PuCl3)' ANL/C [21] 82 NaCl 5.4 PuCl, oot 12.6 UCl3 55205 2.14- CRlL/é3 [6] 45 WaCl 30 (Puly+ 649 1.7 25 KC1 UClB) &TL/A [21] 70 NaCl 30 Pucl 682.5 2.1 This work 70 NaCl 30 U013 640 2.2 Taube [17] 55 kaCl 15 PuCL, 850 3.6 30 UCL, 650 2.9 ANL/B [21] 50 NaCl 22.5 PuCl3 615 1.6 27.5 UCLy This work 50 NaCl 50 UC1, 610 1.7 This work 50 NaCl 25 ThCl 605 1.9 25 UCJ.B“ 193 Table 3.3=-XVI Calculated Heat Transfer Ccefficients Molar Composition by (cal/sec en? X) at at o 1 a LhClb/UCIB NaCl Typ + 100 850 C 0 30 1.22 1.22 50 1.20 1.23 70 1.22 1.18 0.5 30 1.17 1.19 50 1.15 1.19 70 1.16 1.11 1.0 30 1.17 1.19 50 1.13 1.18 70 lo1k 1.09 3.0 30 1.20 1.24 50 1.15 1.20 70 1.12 1.06 Table 3,3-XVII Comparison of hf Be-ween This Work and the CRNL/567137 Stud- 199 Effcet of This work/ Parameter _CRNL 56 [13] This “ork CRNL-56 salt Temp (°¢) 732 (582) —_— Yolar % NaCl 83 70 —_— /c(g/cué) 2.5 2.7 ('2—'1)0'8= 1.06 (en/s en) 0.10 0.01L46 (0'012)‘0 47 2.9 c, (cal/gnc) 0.20 0.15 (Efiig)o "33 0.91 -3 10 “cal 12,0. 67 X. (m) Lel 2.12 ( ) 0.61, N‘(Tfi/S) 6.1 14. (ZA.]_ 0. 8: 1094 20 l-lh hf(cal/s m C) 0.38 1.14 (ngé) = 3.00 3e3.5.7 Choice of core mix. A pure fissile fuel would produce the hardest neutron spectrum but an MSR requires carrier salt admixture to reduce the melting point. Adding fertile salt to the core mix helps reach BG potential by reducing leakage through increased fertile capture, but it also lowers the BG potential by softening the core neutron spectrum. Another consequence will be to increase the critical mass of fissile kkuel and the core size, This will be good for power density (Fig. 3.5~15) and in-cors to out-of-core inveniory ratio (for control purposes), but there are limits to economical plant size. An alternative to fertile fuel in the core is to increase the size of the outer blanket. However, this also increases the size of the overall reactor system. The ORNL/56 design limited the actinide concentration to 17 mole precent actinide chloride. Of this, 2/3 of the actinide was fertile material. The prime motivation was to keep the liquidus temperature below SOOOC, to avoid what were then exotic ctructural meterials, The British [29] chose an actinide molar conceniration of 30-4055 the Fremch [23], 30%,following the earlier lead of LNL [217. The. readoas for these choices are not clearly stated, However, we suspect they include ~. . -. l. Keeping the power density low 2. Keeping down the plant fissile inventory 3. Improving the thermophysical properties. This study assumed 70% for most of the physics analyses so as to fully explore the advantages of a hard spectrum: power density is reduced by using a geomeiry with high surface/volume ratio and limited fertile fuel presence (limited reduction in enrichment), By and large (1.086) £ 3 o o -CeCl3 (13.88) - Q -LaCl; (5.78) o . 3 * € 300 -NaCl -ReC1 (1.05) | S -CsC1 (13.79%5) T -SmCl2 (3.73) -SrC1° (5.48) S -BaCl2 (9.50) ) ™ Fig. 3.4=1 The behavior of fission products in the molten chlerides fuel (Yislds given represent products fer 100 Pu fissioned)[17]. 209 For the fissioning of 100 Pu atoms the following balance has been suggested [17] 0,008 segas + 0,003 Brgés +.O.942 Krgas + 1,05 BbCL + 5449 's:~012 + 3,03 YC1, + 21,5 ZrCl 3 3 0.29 NBCl. 4 18416 MoCl, + 0428 MoCl. + 4.01 Tc 5 2 3 , met 31.45 Rumet + 1473 Rhmet + 12,66 Pdmet + 1,88 AgCl + + + 0.66 CdCl2 + 0,06 InCl + 0.324 SuCl, + 0,67 SbCl 2 3 7f65 TeC12 + 6,18 Igas + 21,2 Kegas + 13434 CsCl -+ + 9450 3aCl, + 5.70 LaCl, + 13.98 CeCl, + 4.28 PrCl 3 3 + 1044- PmCl 3 3+ 3.74 SmCl3 + 0,60 Eull 3 + 11,87 KdCl 5 + 0,03 Cd013 + traces of complex compoufids like CsUCL. Q This suggests the production of about 80 kg insoluble metal per Giwth~year, That amount of metal deposition warrants serious attention to prevent resirictions and plugging. Since each Pu atom fissions into iwo product atoms, the average 3 chloride ions ver Pu 2 FP per Pu neutrality. Chasonov showed that for PuCl, the average valency is 3 close to 2,0, indicatinzg a deficiency of Cl ions. However, the re- = 145 Yo maintain electro- valency should be moval of the relatively-inert metals like Ru, Pd, and Te, either through plateout or deliberately by the reprocessing system, can quickly lower the average valency, 3Below 1,5, Cl ions are in excess, producing corrosion. In summary, tube plugging may occur at some locations and fube dissolution by corrosion at others, In~pile testing to high buraups is needed to study these competing effects. 210 Experience on molien fluoride systems indicates that irradiation and fission products do not accelerate corrosion (lMcLain pe800) Molybdenum as a fission product with a yield of lai (out of 200%) of all fission products may remain, in part, in metallic form. When molybdenum is the structural material, the corrosion problems of metal- lic molybdenum or its alloys are strongly linked with the fission product behavior in tnis medium. Fission fragment MoCl, will react with UCl3 to form UCLA plus Mo. Likewise the excess chlorine released will react with the strongest reducing agent present, UClS, to form UClh. 3.4.2.2 The efféct'ofUCl; —resence in the core salfl-tsij. re UC1L, highly corrodes container me-als, but both theory (stability of cor- | rosion product chlorides) and labcratory experience show that dilute—UClh fuel salt attacks little: alloys of ircn, nickel, or refractory metals such as molybdenum resist chlofide fuel and blanket salts, providing the UClB/UClh mix contains no mere than a few percent UCL, . Alloys can also include minor amounts of more reactive metals such as chromium if the design allows for same surface leaching. Extrapolating from CEXL experience with fluoride fuel salts at comparable heat ratings, irradiation should neither release chlorine from the melt nor enhance the attack of the cantainer materials. 211 3.4.3 Candidate Materials for an MCFR The high flux levels in an MCFR suggest high radiaticn damage, However, due to the unique tube configuration the only exposed struc- ture is the core/blanket interface: the blanket shields the vessel, reducing its radiation dose by three magnitudes. Beycnd the high radiation environment of the reactor lie piping, pumps, and heat exchangers for the primary fuel-ccolant. These face exposure to high temperatures, delayed neutron irradiation, and corrc— sicn by salt transmutant impurities. Materials in the secandary and possible=tertiary circuats wilil face a much milder enviraument. However, the amounts of materials there greatiy surpass those ir the primary circuit, thereby sensi- tizing their cost: except for the high temperature helium case, one prefers lower operating temperatures so as to allow cheaper materials. The cost factar, however, must also be balanced with factors of corrosion, thermal stress, and avoidance of secondary-coolant freezing (especially for active liquid metal) on the intermediate heat exchanger tubes. Table 3.4=IIT 1lists the most promising materials for use with the different fluids. Strength cansideraticns dictate the temperature iimits. Corrosion tests under MCFR conditions have generally not been carried out yet. Materials which best resist corrosion by chemical attack should be those with small free energy of chloride farmation (Fig. 3.4-1). Low vasor pressure (high boiling and melting points) will also allow chlorides which form to remain as protective coverings. From avail- able information (Figse 3.4-1 and 3.4-2) Mo is a strongcardidate..- 212 Table 3.,4-I1T, Materials for Various MCFR Fluid Fnvironment Approximate Material Fluid Candidate Materials Temperature Limit € Molten Chloride Stainless steel 600 Salt Hastelloy N 7CO Molybdenum or TZM - 1000~1200 (?) Graphite 15C0 Helium Nickel alloys 9C0~1000 Molybdenum or TZM 1000 Graphite 1500 Lead Fecralloy or (Croloy 500 Molybdenum 1000 (?) Superheated Steam Stainless st=zel or high 550 nickel alloys Lower Temperature Ircn alloys L50 Steam and water 3.,4.3.1 Overview on iletals, High nickel alloys ard refractory metals appear to be the only metals compatible with fuel and blanket salt, Superior sirenzgth and resistance to corrosion and radiation embrittle ment at high temperature recommend molybdenum alloys (particularly TZH) for contact with the circulating fluid. There is some incentive for using molybdenum alloys for all the components (including pumps) which contact the salt., However massive molybdenum outside the core would require precautions against its external oxidation: e.ge enclosing the reactor vessel and other sensi- tive components in a hot box purged with an inert or slightly reducing atmosphere, An alternctive to this would be to use duplex material: a protective Mo layer elecirodeposited or plasma sprayed onto certain 213 UcCls oPuCla 9 ~ SALT-FUEL COMPONENTS 1500- MgCle NaCl J Z2rCla .C O 2. 10004 L - Q/Q/ z MaClx = 1 1 2 ] WCl, 3 PtCl, 0 — 500- AuCls ] 0 100 200 300 . =000 -t Free Enthalpy of Formation AG'm [}c}'.molCfl Fige 3e4=2 Desistance-of Metals to Chloridization Corrosion [17] 214 steels, For the less severe conditions of the reactor vessel and the salt ducts returning from the heat exchanger, Hastelloy-N c¢r advanced devel- opments of that alloy might suffice, However, having decided to use melybdenum for the core/blanket membrane, a change to a dissimilar metal elsevhere in the salt circuit could cause galvanic corrosion. Also, nickel alloys resist molten lead (a2 candidate secondary coolant) very poorly compared with molybdenum; they would probably be subject to attack when the lead leaked fthrougn the heat exchanger into the salt, 3044342 o 2lloys. Mo offers high temperature sirength, zood irra- diation resistance, and high thermal conductivity: promising traits for 2 material in contact with fuel and blankev salts at hizh temper- atures, However, fabrication of light molybdenum sections requires welding and heat ftreatment: this technology needs development, but it is already proceeding for other applications, Also, the high specific cost of molybdenum is acceptzable only on a2 limited basis: +the vessel piping, eic, should use materials which are cheaper and more easily fabricated. Candidates include Mo/Fe, Ho/Mli and Ti/Zr/Mo (TZ1) alloys. With Mo/Fe indications are that the swelling for Fe has already peaked at the operating temperatures, while that for Mo occurs 2t much higher temperatures, The use of molybdenum or its alloys may permit high fuel salt temperatures (above 100000). This can reduce fuel salt inventory through higher power densities and allow gas turbine cycles and/or process heat applications using helium as an intermediate and final coolant, 215 Mo has an appreciable absorption cross section for neutrons. Such reactions also produce significant amounts of technetium. This changes the nature of the structural material and makes it radiocactive. Figure 3.4=1 and the discussion in Section 3.4.2 indicated Mo to be strengly corrosiaon resistant. The most significant reaction from the standpoint of ccrrosicn shouid be (Fig. 3.3=13) UClb, + Mo ———i < UCJ.3 + Mt:JCLL2 UCi 4 will be a major censtituent in the blanket from the neutron transmu- fation of ThCla. Free chloride from fission of U in the core will combine with UC].3 to form UClL. It also reacts directly with Mo: Mo + 012 — Mo()l2 Traces of oxygen or water will produce molybdenum oxide, €.ge. MoCJ.2 + H20 — MoQ + 2 HCL With all of these reactions, mechanical properties will change. 3.4.3.3 Graphite. Both carbon and graphite resist corrosion and conduct heat well enough to warrant exdsting uses in heat exchangers and pumps. A resin-bonded graphite has found wide- application in the chemical process industries. Both America and France have used graphite with Li.F/BeF‘2 molten salt: Mce=- Donnell~Dougl as markets a high-strength, woven graphite for this; the French company PUK sells a product with graphite connected to heat exchanger material [27]. The high thermal conductivity results in excellent thermal 216 shock resistance. Still one must exert care as carbon and graphite are weak and brittle compared with metals. Tensile strength varies between about 500 and 3000 lb/in.z, and impact resistance is nil. Abrasion resistance is poor. Highe= temperature stability is good. and they can be used at temperatures up to 4000 or 5000°F if protected from oxidation (burning). Silicon-base coatings (silicides or silicen carbide) and iridium coatings are claimed to give protection up to around 2900°F. Based on the existing experience as a structural material for molten fluoride thermal breeders and on periodic replacement of tubes, we assumed graphiie for the physics studies here: 2 cm thick for the tube wall anc 4L cm thick for the vessel wall, 1.4.3.4 Heat exchanger considerations. As mentioned in sectian 3.3.2.5 low vapoer pressure of the primary and secondary fused salts permits thin-walled heat exchangers. These are highly efficient in heat transfer, thereby decreasing necessary area and fissile volume out-ofecore. However, assuming close pitch, they will require development and testing to resist the thermal and vibrat onal strains over a long working life. Refractory metals are attractive as they can accommodate high tem—- pefatufeslfihiie»still affording the good heat transfer of a metallic material. A strong candidate is Mo, It does have an appreciable cross section for neutron absorption: +this affects delayed-neutron activation but not neutron ecanemy or radiaticn damage there. Of greater cancern is the welding and joining properties of Mo, its cost, and its oxidation. 217 3,4.4 Materials for the Core/Blanket Interface _Finding an interface material that can withstand a 1016 n cm-'2 éeo;1 flux over an acceptable period of time is indeed a challenge, It is greatly eased by assuming periodic replacement of the few tubes: an action far less complicated thar the conventional replacement of fuel elements; just drain the core and blanket and uncouple the tubes outside the reactor, This is especially feasible if one uses a cheap material like graphite. Section 3.9, Access and Haintenance, discusses this further. Still the tubes may experience a cumulative dose of 1023 n cm-2 in just a few months, Also irradiation of graphite produces "stored energy'. ledeH HMaterials for Reactor Vessel The reactor vessel will experience (1) Low pressure (2) Temperature higher than the blanket salt meliing point (3) Radiztion damage from a 10'3 1 em™? sec”! flux. A suitable material must meet these conditions; it should also mini- mize (1) Induced radioactivity so as to maximize access (2) Cost Candidates include graphite and prestressed concrete (of low= activating compositions), 218 3e4.6 lMaterial for a Lead Secondary Circuit 34446,1 Steam generator tubes, Mo alloys normally resist lead cor- rosion well, bul mizht not if a steam generator lezked in small amounts of oxysene Dissolving Mg or Ca as a getter in the lead would help lower the oxygen level, but might not suffice, One could also clad the molybdenum with an alloy which resists "oxygenated" lead; Fecraloy, 2 ferritic sfeel containing 2luminum, forms a protective alumina film under these conditions, A third, untried alternative, might be that Fecraloy alone can resist both steam and oxygepated lead, Forming a duplex with staine less steel could overcome ifs poor high temperature strength, The Dossibllity of m2ii-g the neat exchanger tubes of Hastelloy- ¥ haes only slignt attractions: it would still need a Fecraloy or molyb- denum external cladding to resist the lead, Y,0e642 Remainder of the circuit., The ducts carrying the lead to and from the steam generators and zlso the steam generator shells might be molybdenum, protected from external oxidation either by a stainless steel cladding or by an inert atmosphere. An altermative is Fecraloy, externally clad with stainless steel for strengthe. 219 3ede 7 Material for a Helium Secondary Circuit The main change relative 1o the lead-cooled design is bigher.tcm=- peratures, parallel to a core temperature increase from 810 to 970°C. This suggests the need for molybdenum alloys, especially on mechanical grounds, However, even though there is no reason to doubt their cor- rosion resistance at this temperature, any supporting evidence is come 2letely lacking, British studies showed {that design variations might lower the sa2lt pump operatinz temperature from 800°C %o 67000. This would ease the engineering and the selection of materials, Materials selection for helium cooling should ve much easier than for lead cooling but so far has been little studied: unidentified orob- lems could become just as ch:;lenging as with lead-cooled sysiems. Presumably a helium cooled system could draw on the extefi;ive work already done with respsct to HTRs and gas turbines: The MCFR version could actuwally ride on the future HTR program,. HTR programs have recognised that, in practice, helium is not an inert coolant since iraces of air, moisture and other oxidizing gases can cause corrosion and traces of carbonaceous ges can car- burice structural materials, Thoush much work remains it does seenm likely that it will be possible to specify 1e The impurity levels acceptable in the helium 2, MNethods of controlling them 3o What intentional additions if any, should be made, for example, to render the helium reducing, Selected nickel alloys could be used for {the turbines and other circuit components, This would restrict the need for molybdenum in the coolant circuit to heat exchanger tubes and similar applicztions, 220 3.5 Physics of MCFR(Th) and Its Fuel Cycle 3.5.1 Nuclear Models of an MCFR J.5.1.1 Definition of the reactor geometry. Section 3.l1.3 conclud— ed that the best MCFR arrangement might be a small number of skewed cylindrical tubes. The chain reaction will center where the tubes converge. However, as nc fuel salt boundaries exist along the tube ares, criticalitiy boundaries blur, High levels of flux and power generation in the primary salt may extend far along the tubes. This will depend on individual tube subcriticality, distance of separation between tubes, and angle of their skew. For example, if they are near critical by themselves and are distant from one another, then their fluxes will follow a cosine distribution over the total distance between bends. ‘However, ine creasing the intertube'coupling will cause the fluxes within the tubes to fall off more steeply beyond the tube convergence region. Th concentration and other fuel salt parameters will also affect neutron distributions because they moderate the neutrons which short- ens their mean free path. Fluxes will also extend into the blanket in all directions but with much shorter relaxation lengths. The choice of tube arrangement will depend upon a detailed trade study of effective core size: extension of the chain reaction out along the tubes will increase exposure of the pressure vessel to neutrons emanating from the tubes; making the core size small re- duces stability of the system to perturbations. 221 3.5.1.2 Use of spherical geometry for neutronic calculations. Neutronics analyses of complicated geometiries require expensive Monte Carlo calculations. For.the present survey purposes, it should . suffice to use simple one-dimensional spherical geometry featuring five zones (Fig. 3.5=1): core, core tube wall, blanket, vessel wall, and reflector. . —7 __——core L7 — reactor tube vessel II wall wall E core blanket | reflector Fig. 3.5~1 Cne-dimensional Spherical Model All the tube wall material gets lumped into one spherical shell. The validity of this approach, vs. uniform material distribution in the spherical core, will depend upon the number of actual tubes. For only three or four tubes, the current thinking, it should do quite well. Taking the tube wall thickness to be two cm facilitates compar-— isons with MCFR(Pu) studies. The vessel wall is four cm thick; the reflector zone, 4O cm. A check calculation with 20 em reflector- thickness showed no significant changes when the blanket thickness was 200 cm (quasi-infinite). éection 3.5.7.2 studies the effect of blanket thickness, settling on a value of 200 em for most cases. The critical core radius usually ranges 25.-= 50 cm. 222 Section 3.3.5.6 gives the densities of individual salts and an algorithm for combining them. Densities for graphite and nickel were 0.1 and 0.09 atoms cm * barn™l. A primary difference between an exact calculation and those in spherical geometry is likely to be the spectrum softening and other effects of the ThClA/NaCl blanket salt intermediate between the fuel salt tubes. Depending on the clearance and skew angle between tubes, a smeared model of such a spherical region would probably include 5= 25% blanket salt by volume. In this regard, the calcilations may be better cdone in infinite- cylinder gecmetry. That remzins to be shown however,iand cylindrical calculations cost more. In the meantime, spherical c;itical U masces (U) or ¥slumes Gan be corrected to -cylindrical geometry by use of - shape factors: U @/D) = U, /SFE/D) Fige. 3.5=2 displays some experimental and calculated data. A sub=- set of these (Table 3.5I) was fit with a least squares program %o produce the foilowing algorithm for use in Section 3.5.5 studies: log SF(L/D) = 64399 x 107> log(L/D) - 0.6976 (log(L/D))* - 90820 X 10-3 223 1.0 - O —Q s\ ™ 4 \‘\Jn\ 0.87 Q A o A £ 0.7 o 3 O ) = 0.6 4 4 7 A 0.5 + O OO]+ ! ] 171 I { , ] l f Ool 0-2 Ool:l. 006 008 1 2 L/D Fige 3.5~2 Shape Factors for Cylindrical Fast-Neutron Systems [123] Legend D high-density metal system O low=density metal system _ 0O ZPR~III 1l:1 critical assemblies Table 3,5~ Experimental Shape Factor (SF) Data [1281 Assumed for This Study L/ SeFs _L/D_ SeF. 0.217 04297 0.9 0,96 0,257 04565 1.0 096 03 0.7 1425 0,94 0ud 0,82 Ted 04 91 0.5 0.89 167 0.87 0.6 0,94 2 04825 0e7 0,96 2 04825 0.8 0,96 8,79 - 0429 225 Je5.2 Neutronic Calculational Methods Neutronic analysis of small fasi reactors like MCFRs generally requires transport theorye. Mcdeling the reactor as a one-=dimensional symmetric sphere permits use of the well established and efficient ANISN code [129]. 3.5.2.7 Spatial mesh. A spatizl mesh of 1/10 of a2 mean free path £cr interval would give quasi-exact results in transport theory calculations, Computer storage and cost generally prohibit this, A mesh of one mean path (mfp) per interval suffices for most cal- culations, In MCFR(Th) calculations, the neutron mean free vath typically ranges from 1-10 em (Teble 3,5-II), With a core radius of 30-80 cm, a 200 cm blanket and a 40 cm reflector, a uvniform one mfp interval mesh would mean a lot of intervals, PFurthermore, most would lie outside the core, leading to possible numerical instability in the iterative transport theory calculations, Consquently a coarser mesh was used (Table 3.5-II), especially outside the core, This worked fine for most of the survey calculations; 2 few select effects needed finer detail, 3¢5¢2+.2 Angular quadrature, Most studies were done with an 34 angu- lar quadrature, 88 check calculations on the smallest of spheres showed increased neutron leakage as expected, but only increased the volume by 1.5%, while decreasing BG by ~ 0.01, 3,5.2.3 Convergzence criteria, Initial survey calculations satisfied a keff convergence of 0,01 and a critical radius search convergence of 1¢%. Select effects required 0.001 k pp = cOnvVergence, 226 Lee Table 3.5 Spatial Mesh in MCFR{Th) Neutronic Calculations Core | Vesgsel Core Wall Blanket Wall Reflector Typical mfp (cm) 12 3 10 2 1 Typical dimension (cm) 30-80 2 200 L LO Renge of dimensions (cm) 21~136 2 100-200 4 20,40 Survey No. Intervals 10 1 10 1 b (10/16)* Int. Size (cm 3-8 2 20 4 10 " v (mfp 0.25-0.67 0.67 2 2 10 Detail No. Intervals 20 1 go 1 14 (20/1,6 )% Int. Size (em 1.5 2 o7 4 3 " n (mfp 0.1 0.7 0.7 2 3 Multizone No. Intervals 10+10 2 10+10432 1 545 (20/65)* Int. Size (cm; (Re5,vb1l) 1 1. 3. 5. 4 3.5 " " (ulfp (0.2'Vbl) 003 O.l 0.3 0.5 2 3.5 Intermediate No., Intervals 10 1 25 1 L (10/31)* Int. Size (cm; 2.7 2 8 L 10 " " (mfp 0.23 0-7 008 2 10 * (no. ¢ore intervals/no. non-core intervais) 3e5.3 Neutron Cross Sectians This study required neutron cross sectians for Th, 233U, 23l‘U, 235U, 2330 fission products, C, Na, Ni, and Cl. Data for 236U, Mo, F, and other carrier salt catlons were desirzble, 3.5.3,1 Bandarenko 26 group set [98]. - The Bondarenkc neutron crbss sectlon library long ago =stablished itself for fast réazctsr caleu- lations, Testing has revealed deficiencies in the Pu isotopes, 235U, 238 and U, but these nuclei by and large do not appear in an MCFR(Th). Less is known in general about Th and 233U crdss sections; they are. cur- rently undergoing extensive reevaluation for ENDF/B-S; For survey ournoses nhere, the 3ondarenk: values in Pl-cor-ecied PO form (trans- port cross sections) should suffice, : Nuclei fiiésing from ihe library are Cl and F. Early in the study, K substituted for Cl: this was conservative as K absorbs more neutrons. Final calculations used Cl and FPl sets derived from ENDF/B - IV. 3.5.3.2 Twenty-six group, P1 set from ENDF/B-IV. Available late in this study was a 92-group Pl set of cross sections gener- ated by AMPX [130] from ENDF/3-IV, containing most of the materials of interest here. This was too large and expensive to run so the set was collapsed to 26 groups similar (Table 3.5-III) to those in the Bondarenko library to allow cross—comparison and -usage. The list of materials included not cnly Cl but also F. Missing however were Na and C. Comparison of their nuclear properties indicat~ ed Mg would represent Na well, especially since Na content is small campared to Cl. For C, the Bondarenko set was used ; C ic also a minor 228 5 : SRS EGREBwe~onrwbro Table 3.5III Lower Boundaries of Energy Group Bondarenko Set 10. 5 MeV 6.5 MeV L.0 MeV 2.5 MeV 1.4 MeV C.8 MeV O.4 MeV Q.2 MeV C.l MeV LE.5 keV 21.5 keV 10.0 k2V L.65 keV 2.15 keV 1.0 keV 465 ev 215. eV 100. eV L6.5 eV 21.5 eV 10. eV L+65 eV 2.15 eV 1.0 eV 0.465 eV 0.215 eV thermal 229 149 6.07 3.68 2423 135 0.82 AMPX-Generated EXDF/B-LV Set MeV MeV MeV MeV MeV MeV MeV MeV MeV keV keV keV keV keV keV eV eV eV eV eV eV eV eV eV eV eV constituent. 3.5.3.3 Effects of resonance self-shielding. Résdnance—regicn cross sectims for the two sets described above were averaged assuming the materials to be infinitely dilute, i.e. ignoring resonance self-shielding. In 233U most of the reactiviiy effects from fission and capture self-shielding, if present, would cancel each other. Self-shielding for 232Th capture, if present, could affect 8G. The Bondarenko-calculated self-shielding factors for capture (Table 3.5I7 ) show virtually no self-shielding in the top 12 grcups, which cover well ths core and blanket spectra. %“hat little . 232 does occur in grcup 12 fcr 233U is similar. Thus the infinitely-dilute calculations Tn, Doppler Broadening partly amelio— rates. appear correct to first and maybe second-order approximaticn. 230 Table 3.51V Bondarenko Self-Shielding Factors for B2 98] "C at Uo cqual to }l I, at O, equat to ‘g t st G5 equal to 1! l 10+ I 15 2l I I ' ! r ; l 300 ‘ 1.00‘ too! 08 'oe7 1o T1uo {098 & uo 1,00 0,99 500 ! 17001 vool 100 noy § o fioo oo | 10 | 100 1,00 20 | LW % LOUG Lo Lo tuo LU0 e T Lo L Log ) tu L R © 300 LU0 LO0T 0.7 0.02 T b0 099 004 100+ 0,99 1097 a0 00 L0000 00 L Lo 09 L log ! 1,00 0.0 2000 1,00 | !.OUi 1,00 in.slv HERCIRRE | Luo 100 100 5 e | o S v i WO Lo 1,00, 0,95 'o S0 41 1097 |oss 11,00 099|093 Codw L : OO 098 0.~ o outd "y Y LLAo » w99 | 0w 2000 11,001,001 0,99 %().-..l L 100 1o iu,gs ; 100 | 1,00 | 0.99 ' : ! , E ; :] ! OG0 Lou 0T 002 02 T tae 093 T05) b0 ) 0,93 ¢ 6,96 POU Tu v Yy 002 sty 097 0S8 L0 099 uEs PII0C LG bt 099 o2 (L 1 ,IO,SZS <100 100 1 0,94 Ly | | 300 1 1.00.097: 087 (0,45 i 1,00 {086 :0,59 , 1.00 | 0.94 | 0.7 900 11,00 0981 0,92 {053 ) oo |09t (0,62 oo {097 | o's0 2100 [ 1,00 Tow | 0,96 30'64 | Loo juve 0,65 0 | 095 | 0,86 | L i o 300 {1,00:0921 058 1020 034|050 [0.29 1092 ) 0.65 ! 0.4 | 900 100 095, 0,67 1025 087 1058 10,30 | 0,95 [ 071 | 044 | 2100 ‘ 1,00 10,46 i 0,74 lu31 {091 1066 10,32 40,9 | 0,77 48 i | ‘ ! 300 | 1.0010.01 1 0.59 10.21 | 0.82 1044 10,23 {090 | 0,61 { 0,36 o000 | 1,06 10940 0,68 1026 T 057 |0.51 |0.2¢ §093 | 0,66 | 0.39 | 2060 LU0 006 076 035 09l 10,62 1027 0,95 0,73 | 0,43 N L | | 300098086045 1003 0R2 1037 10,22 2091 | 0,64 | 045 L 900 1099 6l 0054 10T Cosr tuks 10023 Y04 | 070 ¢ o030 | 2100 099094 062 vl i (.94 5U,Su 0,24 40,9 10,71 0,30 , R 300 {0,9610.72| 0.27 100837 0.61 029 10,22 i 0,85 | 0.64 | 0.5 900 : 0,97 .0.76: 0,30 ! 0.0857 0,70 10,30 1022 [ 0,88 | 0,65 | 0.54 2100 {098 0,82 | U3 10,099 | 07D 1034 10,23 |09 | 069 | 055 i ! ) i | I © 300 11.0011.00 | 1,00 | 100 4 1.00 : 100 11,00 § 1.00 | 1,00 | 1.00 900 + 1,00 - Lo 1,00 100 1 16O 100 1I.UU 21,00 11,001 100 2100 | 100 100 100 Luo "t e g 1,00 1,00 ) 1,00 | 1,00 o 1 300 [ 09110551 0,16 0058 0,37 L0080 0,043 . U538 | 0.22 SOU OO 06T 0 0 0TeE 07 0 T00 T 067 1 026 10,14 i 2100 | 0.96 {0761 0,31 0,097 i! G0 1004 0 077 | 0,36 | 0, i 1 ! i 231 e ————— 3a5.4 Transmutation Chains and Equations 345461 Chains for actinide transmutations Figure 3.5-3 details the Th cycle burnup chein as it pertains to reactor criticality. 233Pa branching ~ by neutron capture to 234y or by‘fia- decay to 233y ; is very important in a thermal reactor where 234U only absorbs neutrons. However, in 2 fast HCFR(Th) spectrum 234y mostly fissions (Figure Zel=3) ;-- much more so than any other fthreshold-{ission isotope. It does not significantly moderate the spectrum or otherwise detract. 33 232Th: 233 2 Production of ~~~Pa follows directly from 233 Th decays quickly enough to ignore i1ts dresence, Pa decays at the rate of 1n 2/27.0 days = 3.0 x 10-7 disintegrations per second per 233Pa 210Mma With g = 10" 5 en? s in core (25% of +ime), ~ O oui-~of-core, and a core-spectrum-averaged cross section of 0.3 barns (Table 3,5-V), 233Pa(n,Y) reactions oceur 0.7 x 107 3-1, more than two magnitudes 233 slower than Pa decay, In the blanket the greatly~reduced flux lowers the reaction rate even further, A fhermal reactsr with<3é200 = 39b and f = 104 5 en™® s 0,25 yields a similar rate of 1 x 1077 s [1313. In summary, 234U production from 233Pa parallels that in 2 thermal reactor; but a low-threshold fission cross section raises it to fissile fuel status and also destroys it faster. Therefore ihe reduced presence of 234U and the 233U;like reactivity behavior of 234U in an IICFR obviate the need for separating out 233?& to let it decay as in an USBR, Therefore it suffices in these survey calcula~ tions to igmore the 233p, (n,¥ ) branch, i.e. to construe all 232 233 Th captures to pass directly to Us 232 £ee 232 e 23y Py 23 1;1>a ’__J 233% _____] 2310 et Figure 3.5-3 Th Cycle Burnup Chain for Neutronic Calcul ;236Pu ‘ 238Pu_ e 239Pu 1___237Np ____J 23 5U 236U (n, ¥, - LNy 2n) p ations Table 3.5V MCFR{Th)-Spectrum Averaged (ross Sections for Some Obscure Reactions Neutron 6.5—;0.5MeV LeO=b.5 2 5=440 1.&"2.5 QuB=1.4 Q.4=0.8 0-2—0011- 0.1-0.2 L6 o 5=100keV 21.5—46-5 10.0=21.5 465=10.0 2 415=4..65 1.0=2.15 1,65-1000 eV 215 465 17 100=- 215 18 46.5-100 19 21.546.5 20 10 =21.5 21=26 E;CRE:C3FSE:ESq)03«JChvwx-th)h' Core average Cross Sectigé {tarns )* > g;zztrum g;zgtizm Group Energy 230Th(n,)j 33Pa(g,73 33Pa(an) (%) %) 0.006 0.01 1.2 0.501 0.0LC3 0.005 0.02 0.8 3.18 0.272 0.0066 0.05 0.8 8.46 0.835 0.01 0«13 Ca7 17.15 2.78 0.2 0.3 Oal 19.92 6.82 0.034 03 Oe 22.03 16.5 0.04L 0.3 1L.42 18.5 0.05 Ouds 7 .86 17.1 0.12 Q.6 LeO 1Lel Qo2 lel 1.48 1l.1 0.32 2.2 0.778 Te71 Qo 5e5 0.133 2.70 0.65 5¢3 0.0372 0.521 l.1 ) 0,011 0.84L2 1.7 1.5 nil 0.1522 0.5 10 1.49x10_ 25 1.87x10 IQ—O 1-L083"l+ 95 3.954 nil 0.035 0.29 C.53 0«12 0.79 0.83 Blkt.average * cross sections eyeballed from @NDF/B-V. In calculating burnup (fuel depletion and replenishment), this study also omitted the 232 3 Th(n,2n) and 233U(n,2n) paths as well as the cal- c e i 236, ... culation of isotopes higher than 3 U (Fig. 3.53). These approximatiaons should not affect the conclusions of this survey with respect to BG or criticality. The resulting chain was 234 232Th 433U 23LU 235U 236U Each arrow denotes an (n, ¥ ) reaction. The corresponding equations, which describe nuclide activity up through 235U, are d,/dt = Gy @ Th = (0, &) Uy (345.bm1) dUh/dt = (S &y Uy = {o,4) U, (3+544=2) dUS/dt = (64 9 U, - 21h(n,2n) 7x10 1x107 22 = 2 - ~-10 A oy T (YY) 4xad”? 26 ~/ 231 . - -10 11 llxfi 3 Pa(n! () 7x10 7 6x10 ~ 13 ‘gumfi 2339a(n,2n) 107 2x107P 23 So? 2Hq(n,2n) 20107 a0 ) = 232 -3 =10 73 3 2 11 1 Pa{n,abs) 2x10 7x10 A\ 233, [/ 4~ =7 -7 /,3 3 » 13 pa’ b ) 3.0x10° " 3.0x10 3 232, -8 w17 /;522 o Y (n ,abs) L4x10 1.x10 » oo 22 (o) 3.1x10720 3,1:007%° T Pm,Y) 7,100 6510712 o 57 23pa(n,abs) 1.1x107° lx10 0 S o7 . . . Sal . # 7 ® ¢ are production coefficients and j(i; are cestruction coef=- ficients. Their use in the differential equations helps in visual checking and bookkeeping of terms 243 The probability ratio R of path <3 OTh(n, Y )23 lTh to path 23 2Th (n,2n)231rh (Fig. 3.5=5) is R=_ Soo ‘oo o2N 'NOZ For N Mmuo = 1{)—6 (1 ppm) mentioned above, get R = 107 x Zi-lf);g - 107 in core 7 x 10 c o0k SELTT |60 i bramket 1x 1079 Thus the = OTh(n, ¥) 231rn path can be ignored. In the destruction of N5, )\13 = 3.0 x 10~ s~L far exceeds ¢0'13: in the core by a factor of nez- 30; in the blanket, by three magnitudes. In the destruction of N22, 622;1) far exceeds }\22 and t@:lus becomes the governing time constant (of about 10 mos.) in the approach to 232, eq— wilibrium concentration. 244 Taking La Place transforms of Equations 3.5.4=4 through <6 23/”23 * 0(11 1t n/}iw /& 22 77 {/‘Q * g3 Moy + Ky 111*“13%132 SQ'{22 -42 22 = d 22 T s4 -'_)22 s My - 21 - & 22@{- 02 =2 U 11 . e T 1S ] s /0( fl,( /p - /513/}( 13 42 = s+/} * O2Y/]{ 02 - .-.\ J . Assuming k. constant = N, for 02 and 23, then 4101 (s) = é /VLR < . i s .eru (s)=3 4123 ) and substituting, get | olya, 1 ° i o e ——Jqn 2l K g 2 S+ Py s S s+ Oy " + o3 [ ° o2y ’flc,, ] } . /]'1 , o—_— C S 4-/—1"\3 ' S 245 Now initially one would have o o 2’\3 (n A/lll, Then l c '){H ) < Q = — ‘x23q£l = Doy 4Z /(27_ s (s+042) { ’ s~/ ez L Fia c = 'XOLY ’z(' Sf'_f),_; cl Normalizing 232U to 233U cancentration cne has N - At B U‘@ ) N2y T New ot nt Xy "(cz:\j ( N75 ','G)u -/317_ 4’/‘57_2 2 —/Jn € + e ——— ./b\\ /'931 /-17-1 —fill - ez : Aat Pt L o1z Yozy (M‘/N'u)?\ Ny -foia + /Ju_& P - /5.3 < ® /5‘3 /57—1 /b.r, '/bn, . \‘&—fl?u.l" - fu l-e”i%ll.r' | — d_flh t —_— - . 2 2. /3 - P i__e"‘fil)_t] (l - e——fllit } ' Nez [ /5’-!- /3 + Ky O(O'z.\{ (’_._;.C__ > —_— N /9 3 - I‘GLL by oo (2 ) N3 246 3 —( 2 and H(_rg’i,az): &(.5,_)~C(1) get : hJ-Zj. - 0(2_5 G"(_bz& + \i” ;XOZI'{ (ifi) ‘+ (,”6“1 ./321.) 23 + oy Negy ('“:]'a._;\ H (,5a3) ‘,5;;) (3;5°4-7) e can now eXplore r122/1\:23 buildup as a function cf 1\102/'1\123, ie€e 0T/ Jas £ (2 00) To calculate the radicactivity we construct and seolve the decay equations for the 2320 daughters in Figure 3.5~4. 221’&3, 220&:1, 216 2121, . . Po, and b all decay relatively instantaneously, so they can be skipped. For a fuel ocutside the reactor with a given 25 2U con= centration one then has | 60.6 m 2321] _71.1'_.,.228'[-1-1 .l_.?y‘ha_;zl%i 0.50 208T1 S 06 MeV 6006:& 0. 64 For vurvose of simplification, we take the 232U concentration here as constant, The differential equations are then dN5g = >\22 N2 = Mog Nog dT 247 \ >0 Fos = *32 Va2 > ol m N i == 0.36 \/\32 Ny, - \)\18 Xg where N,, = cecncentraticn of 322'[} Tog = 'f '_'. 338% 1732 = " " 21 %i 33 Ng = " " 2087y 7\2_2 = 1n 2/72y = 3.1 x 19710 s;'1 | )\08 = 1n 2/1_.9y - 1221070 & )\32 = 1n 2/60.6m = 1.9 x 1074 s"{l )\18 = 1ln 2/3m=3.9x10-3 5_1 Taldng Laplace transforms Q .s’”,g "4’(,3 = 0.36 )JL ’v!sz _)lB 44!8 Mg = —L {%;+oa@haflfl} S+ hig sfl(fl "4/[; = )«oe q‘{og - >\3’L 4{32_ 248 T = o5 o+ s S Mag - /L(Dg = A My = Ao Nos w5l ey | S /Y\YL - ’flz?_ = = e /Qu_ Ny rr\ Mg 1 S+ )«ZL . -t o ) 2 ° Ao ° # e - S+ e g%" SOk [%52 v 22 (M S-l— )\31 S+)Ot © + >z1. 447_2 )] K = Mo oMy S+ Mg (s+hg)(s+ Afi-> + D2k ),31 Xoe oz | 036 V32 do N2z /Ul?: (54-),5) (s+)3,_>(s +)og) T 3(S+)‘\B>(5+)‘52)(5+>:‘:> There are an infinite set of possible situations and correspen- ding solutions associated with different reprocessing schemes. Cf most direct interest is the situation where a o 'vl(go =My = Mg = 249 Then uo 2\ ., S= Mg NIS = 03 \/*bz )\ofb )n. )l% ,T;) ’ “‘!,13 .__.L_..._ — { fi(\m,‘)\m_,)efl) + Fz()‘nu)'fi: A"B) +§()\fi1>—5%>’%)§ >‘8 >31)53 vhere F(a,b,c) = e-at/(b-a)(c—a)_ a S = emission rate for 2, 1=iieV 208'1‘]. S2mmas 208 - . . c and the units of )\lSN 18 & < Tl-decays/s. Substituting in I\a23= 23 233 19 gm 233“U metal cm-3 x 6.02 x 10 U atoms per 2333 233’6 R [)‘;.g , one gets a source rate 23 Se19.x 22922307 036 x 1.9 % 1074 x 1.2 x 107 x3.1 x 10710 { 233 — —--—"""’__— F,; N23 AIB )3.?. )DB 1=t —4 V- o0.049 [ 3 S(scmj) N 22 _____.}___..—-ZF N—2-3- >1e )32 >‘D‘B Note that for at least one Fi the expanential numerator should be near cne and its dencminator very small so thatZFi is large, Section 3.5.11.2 evaluates S for situations of interest. 31.5.5 Pertinent Physics Metrics 3.5.5.1 BG: actual breeding gain. The canventicnal definition of BG is rate of fissile production BG = rate of fissile destructian (345.5-1) The question then arises: which materials are fissile and which fertile? Conventian says 23 L'U is fettile, but in the very hard HCFR neutron spectrum it behaves like a fissile: G exceeds Oy over most of the neutrecn spectrum (Fig. 2.1-5), suggesting a bare 237 2L05 (159 kg), or “*n (114 kg) {1351, 232y and 8U cannot sustain a chain 233:‘—' 23-4-1}', a-n.d 235" sy critical mass smaller than that for Np (69 k&), rzaction, Thus we lacel U as fissiles in an 0932, 232'3?3:1 and 238‘6 2s non-fissilcs, BG can mean the instantaneous BG at any givén point in time (of the burnup cycle) or a time-averaged BG such as over a given LMFBR re- fuelling cycle or reactor life. As the survey nature of this study precludes a detailed fuel management analysis, this study deals anly with instantaneous BG, presumably on equilibrium cycle: BG = Th capture rate rate of 23 3“U ' 23 l‘U, 23 5U fission This certainly differs in <°U treatment from that for a thermal Th/>>-U- MSBR: however each reflects the physical realties of the correspanding neutron spectra, and this definition does agree with equatian (3.5 5.-1) above. 251 3.5.5.2 BGX: BG extended to zero neutron leakage. BGX gauges the BG potential of a reactor design in the event that neutrans which presently escape from the reactor (outer) blanket (CB) were instead captured there in the same proportims as present. / L;total abs rat?J ~ Th capture -] BGX = (1+BG)§1 J-_--—._-— £ Z -1 OB This metric thus indicates what additional BG can be obtained through a larger blanket. 3.5.5.3 BGP: BG potential asscociated with the core spectrum. The theoretical limit to BG forms another valuable metric of an MCFR(Th). Taube calculated previously-unheard—of BG = 0.7-0.8 (BR = 1.7-1.8) for an MCFR(Pu)e Section 3.1.1.1 defined BG potentizl as the nfimsgr;of neutrons available for breeding per 233U fission. That igrpored all actinides 5 232 excep Th .and 233U. The net result depended primarily on 30r for 233U and indicated 2 potential for BGs up to 0.4 on the 232Th/233U fuel cycle, The ideal situatica would be for 232Th to capture neutrons which leave the core: 1i.e. for none to leak out the blanket nor be lost to parasitic capture, Then BG = Vy -2, However, for the MCFR here-we must return from utopia and allow at least for parasitic losses due to necessary core salt components. This then gives 252 E (parasitic captures) BCP = (V) =2) core . B reactor C s Zreac tor (f:r.ssn.ons ) 3.5.5.4 d) « The core—averaged total flux will primarily core indicate the rate of damage to, and thereby replacement of, the graphite tubes. 3e5e5.5 (E c>ore<é « The median core flux energy is found as- suming a constant lethargy flux within a group's boundaries, This metric proves to be a valuable barometer of B&® (Fig. 3.5-0) since V and O increase wi:h fast neutron energy while & decreases. 3.5.5.6 Power densitv. Together wita ( E>core ¢ , this metric determines @ core® 1t also affects heat exchanger design: the higher the power density in the core fluid salt, the faster must heat per unit fluid volume be removed in the exchanger. 3,5.5,7 @ (vessel), total and > 100 keV. Unlike the core tubes, the MCFR design and operation plan does not readily tolerate the re-~ pla.cemént of the reactor vessel, Therefore the neutron damage, especially measured by the flux above 100 keV, should not exceed ].Ol3 -2 ncm sec - or so. 253 Flux- Spectrume Averaged (P =27 0456 0655 Flux~-Spectrum-Lveraged Energy (keV) Figure 3.5=6. Variaticn of Utopian BG (Iznoring Parasitic Losses and Leakage) with Core Flux=Spectrum Gardness. 254 3.5.5.8 Fuel inventory and doubling time. Doubling time (DT) provides a pertinent measure of the economics and physics of a breeder reactor: the time required to breed encugh extra fuel to start up another re- actor. That should really encompass the whole fuel cycle: core + blanket + primary loop + ineplant stockpiles + transport + fuel fab- rication + fuel reprocessing + out=of=plant storage. However, perhaps because solid fuel reactors all require similar ex-reactor fuel cycies, DI conventionally measures just the time for duplicating core and tlanket. To facilitate comparison, this study adopts that definition also except that the MCFR must also include its full primary circuit cantents: Ui + U0 + Ub DT = (3.5-5‘2) BG production rate Here U; = U content inside the reactor vessel; more precisely be- tween the ex—vessel bends of the tubes (Fig. 3.2-9,-10, and-11) U = o = U inventory in primary circuit beyond (outside) the reactor vessel: piping, pumps, and heat ekchangers. Section 3.4 indicates the volume Vo = 23 m3. Ub = Blanket U content This should result in a conservative comparison of the MCFR fuel cycle to the solid fuel cycle of other breeder reactors: solid fuel will require much more inventory in storage, transportation, decladding, and fabrication operations; the MCFR, little. Both cycles should involve similar amounts of chemical processing, but 255 the MCFR goihg from liquid to liquid state, may require fewer steps and therefore less U. The odd geometry of the skewed tube configuration increases the in-vessel U content, Uj} considerably beyond the calculated critical spherical mass, U_. If there was only ane cylindrical tube (N=1) then the volume Vi assoclated with Ui would be Vi(N=l) = 'n‘n‘?L/z+ = VS/SF(L/D) (3e5¢5=3) wiiere v s critical volume in spherical geometry SF = L/D=dependent shape factor (Sece 3.5.1.2) L = critical cylinder length. With a single cylinder, L = H H = distance along tube between bends. Considering the 2 meter blanket surrounding the core, we estimate H = 5 m D = critical cylinder diameter Rearranging Eq. 3.5.53, get L L/D = (TrL3 SF(L/D) / & V) ' (3e545=4) Since SF depends on L/D, Eqe 3.5.5=4 is transcendental: it must be solved iteratively or by graphical intersection technigue. If N> 1 and the tubes lay sufficiently apart from one another (lcosely coupled), then ' (N > 1)=N vy (N=1). However, this is not of interest because we prefer strong intertube neutronic coupling (Section 3.1.3.2). With strong coupling, if N tubes were not skewed but lay ad- 256 jacent to one another, then their combined cross sectional area would appreach that of a single large tube, then Vi(N >l) = Vl (Nzl) Of Eq. 305-5—3. In contrast, when skewed the maximum, the tubes closely approach ane ancther only near their midpoints. Then no tube sustains the neutran chain reaction very well ogtside the close-approach zone, The critical geometry of that zone should resemble a square cylinder (L/D=1), somewhat squeezed at its middle into the shape of an hour glass or a wheat stack (U). The large mean free paths of fast neutrans tend to negate the "tubular" protusions of core salt from this geometry: next colliéions will probably occur in the neutron-absorbing blanket. Then with L/D=1 2 V/SF(L/D=1) = NWD, D, /4 wherse Dt = diameter of a single tube D, = (bV /N T sF(L/p=1) )Y/ Vi(N >1) = NTH th/‘* = H((vs/o.gv)2 N '[T/A)l/B 0.94 H (N vs‘g)l/3 (34545=5) Figure 3.5=7 displays the results of Eq. 3.5.53 and 3.5.55. The di.i‘;'erences in V]-_- correspond to the differences between zero and maximum skew angle respectively, Thus the true results would lie intermediate, For the present study we take Vi at some value of low significant figure slightly below the Eq. 345~7 line (for N=3), ' This should result in a comservative estimate of DT, To get Uy or U we miltiply V. or V_ by U_/7 . 257 A (m®) Figure 3.5=T. 7Tariation of In-Vessel Primary-Circuit Volume with Critical-=Sphere Volume for Maximum~ and Zero-Angle Tube Skew 258 3,5.6 Fuel Cycle Modelling 3.5.6.1 Fuel cycle model. Figure 3.5-8 describes the actinide flow scheme. (Section 3.7 describes the chemical aspects of the fuel cycle.) Th supply to both primary circuit and blanket maintains the Th con=- centrations. U exits with fractiaonal removal rates p and b respect- jvely, passing into the shim tanke b 1is actually a net removal rate: part of the removed blanket U returns. | fi@v ' Primary BlanRet Circutr | (xnel. core) 8 f d}— t bus) & Fig. 3.5-8. Actinide Flow Scheme Fractian £ of the shim tank inflow feeds the primary circuit, simultaneous with pUp and fission product removal. The remaining fraction g = 1=-f is exported as BG. Extending Sec. 345.4e1, U varies in the blanket according to 259 aUy /a6 = < Ome ) Th B Uy B Uy (3.5.6-1) 3b de_b/dt = (Cwbdy Upy = (Fsad » Uy, - BUg, (3.5.6-3) Simjlarly in the primary circuit Wy /dt = (Gor D) Thy = (5,6 > g Uzg = Flgy + 2(pUy, + bUsy) (3e5.6-4) dilbp/dt =S ) Ugg = (5D o Us = g + i'(ppr + bth) (345.6=5) QU /dt = B &) Uy = Eab D g Usg = PUg, . ‘% + i‘(pUsp + bU5b) (3.5 ) Here the subscript s denotes the spherical core region. The in- tegrated reaction rates may differ slightly in few-tube geometry, but they can be adjusted later to reaction rates from assemblies of higher Th concentration. We define here U, = U35 + Ul;s + U55 (3.5.6=7) U-b = UBb + Ul{.b + Usb (3.5.6_8) Usp= s vp/vs, i = 3,4, or 5 for 233y 23“1.!, or %, € =V Vs Vp: Volume in the primary circuit VS= Volume of a critical sphere 260 Then, substituting in Eqe. 3.5.6=4 through =6 for the {Uip} - b T3 T = { Ty ¢>s Ths" <6.34¢>5 UBS- pSUBS + f(pSUBS +D UBb) ¢ Dus at = 5 UBS- <6u¢>s UAS- pSULS + f(pSULS + b Uh.b) 5 5 £ S = (‘Siwq’>s U™ <55Q¢>$ Ugem PSUg, + (g Ugg + b U5b) 3.5.6.2 BEquilibrium fuel cycle. We now study a specific situation of interest - the equilibrium cycle where compositions have stabilized and {cfigj /dtz each = O. The starter fuel does not affect the equilibrium composition; only the feed material and neutron spectrum control it. This section seeks to determine the relations which express 1. The equilibrium proportions of actinides in primary circuit and blanket 2. The equilibrium breeding gain and/or reactor doubling time. ANISN calculations provide the reaction probability rates (per atom) {6 ¢>) « In addition we specify the core uranium concentration, and blanket thiciness. Together with a critical core radius search, this fixes Us ,Ths, and Thb. Nine unknowns remain: UBs' Ubs' U5s' UBb' th, Ugps f, by, and c. There are seven equations, 3.5.6=1 = 3.5.6~7. That suggests that we also choose the reprocessing rates b and c¢; fihese will then determine f, and from it,the BG export and doubling time. Now setting dUij/dt = 0 for each i and j, then from Eqe. 3.5.6=1 through 3.5.6=3, in t.hé blanket: 261 Coqr &) Uy = =e————eee—ee T (3.546=9) 3b b + h h.b Gy O U, = _ S Ugp (3,5.6~10) 4b b + >b < O_“Y¢> b U = e o U (3e5.6=11) sb b+ <651¢>h Lb Normalizing uranium contents to 'I‘hb, get : & Ty = COme 7 (233Ub a.toms/?'3 2Thb atom) b + <63“'¢>b . {63387, h 54 = Lb b + <°-Qc¢>:9 ujb _ {Cuy by 50 L + b “Lb where Uiy = Usy/Thy In the core, by equations 3¢5¢6-4 through 3.5.,6-6, get , 4y Th. + bf U Ugg = (G ¢ 25 ™s b (3.5.6=12) s + P‘S(l—f) g U, +bfU Ul:,s - < ch>5 3s Lb (3.5.6_13) (Ot +p5 (1-f) add Yy tbf U Ugg = (Sud)s ks 20 (3e5.6~14) s + c(1~-f) COsy d’ls%s * Y U= (Shad)s + ¢(1-1) 0 = {Surd)sUss M Usy o8 (56 Vg + c(1~f) Here Uyg = UiV by = Ths/Us o e Thy/Us o Now to calculate these requires knowledge of f» Summing Eq, 345.6=5 through 3.5.6=7 gives a recursion formula suitably weighted over all the actinide cantents: f + °U3+ bUb Normalizing numerator and denominator to U, get f = c+b- u, - t’b Owr iterative procedure will be to (1) Choose b (2) Calculate {uibg and ug (3) Make initial {“is% guess 263 & CUS + ; <5'.f- ¢>5 ’ Uis : + GSQ.¢>5 USs - <§1\Y¢>s Th.S 4‘- c +1§’<°"’ QY Uis + (O &) Ugg = {Spr P ts (3e546=15) (L) Choose ¢ (5) Calculate f by Zge 3¢506=15 (6} Calculate corresponding {uisi and ug = Z‘uis (7) If ]us-l |< &, renormalize Zuisz and u_ (8) Reset starter guess and go to (4) until desired range of ¢ i1s covered. (9) Return to (1) until desired range of b is covered. The BG export rate is 'fie = (1-£)(eU, + bUy) (34546-16) The doubling time with respsct to reactor inventory is then Ug T (1) (U, + b (3.5.6-17) b) where Ut = Ul+ UO + Ub (Sec. 3.5.5-8)' ANISN calculates U_ and reaction rates (Table 3.5~V1), U, depends on removal rate b (Eq. 3.5.6-9;10 & 11) Fige 3.5=7 provides the means of getting Ui from Us Uo = 2=3 mj, as discussed in Sece 3+5.5.8. 3.5.6.3 On the choice of U removal rates ¢ and b. This section studies how ¢ and b affect the core = 3U/23 4 ratio and blanket U content (Ub) using reaction rates calculated for an MCFR model with 30/35/35 NaCl/ThC1, AUCL, molar composition and 7/1 2223/%2hy ratio in the core salt. Section 3.5.8 shows the importance of high 2330/231“[} ratio in the core salt. Uy, should be minimized because 264 it degrades BG export (68) and increases doubling time (DT); it also generates unwanted power and radicactivity in the blanket. Figure 3.5-9 shows how increasing b quickly reduces Uy to nil. The total U inventory (U,) then comprises just the primary circuit contents, U; + Uj. Meanwhile, fie rises from oblivion to its max= imum level. Since Ut and UE vary with b in opposite directions, their quectient DT (EqQ. 3.5.6=17) varies even more strangly with b (Fig. 3.510). Core 233U/EBL‘U ratic also varies markedly with b in Fig. 3.510. All these comparisons argue for b2210-7, but not much beyond thate Fige. 3.511 shows thzt high ¢ keeps 233U/231*U high and only slightly affects DT adverse_y. It seems to warrant .c as high as 10_6/5, but not crucially so. Note that ¢ = 1077 implies a primary Up removal rate of p=¢/% = ¢ Vs/tp. Thus the requirement on p will typically fall a factor of 10 or more below (be far less stringent than) that on c. A rate of 10—6/5 implies the removal of 1 ppm of the U per second, or a complete processing of that zcne every 11.6 days. Though that may be doable, a rate of 10-7/5 or full processing in 116 days seems more reasonable while imposing no severe penalty. Detailled analysis in other studies shows the same b and ¢ be- havior patterns there. Thus we adopt b=c=ld-?/s as a reference for this woric, We also conclude that physical reprocessing limitations should not prevent us from achieving near-optimum 233U/ZBLU ratio and DT and small Ub. 265 T 10°° 16" 1o 0 w U, Extraction Rate, b (Fraction/s) Figure 3,5-9, Sensitivity of Blanket Uranium Content (Ub) to Its Extraction Rate (b) Bagis: ANISN calculations of critical size and reaction rates of a core mixture with 30 mole % NaCl, 35 mole % TaCL,, 35 mole % UCL 233y/234y ratio of 7. 3! 266 (yrs) Figure 34,5=10 Sensitivity of DT and 233y/234y patio to U5 Extraction Rate 233, 234, ~ - Based on ANISN calculations. of critical size and reaction rates of a core mixture with 30 mole % NaCl 35 mole % ThCl 35 mole % 0013 233U/234"U ratio of 7T 4 ¢ = core U extraction rate (fraction s ) -4 10 10"11 104 US Extraction Rate, b (fraction 8-1) 267 DT : | z - b = 1078 Fige 3e5=11 _— o Sensitivity of Doubling Time and 60 - 233y/23% Ratio o - | Core U Extraction L Rate 40 < e | — b= 10'% ————-—‘-——/-‘ ; = 233y Based on ANISN calculation of - — critical size and - - - T reaction-rates o? ‘ 10-11 10—9 10—7 10—5 a core nixture with 30 mole 7 NaCl d 35 mole % ThCl4 Core U Extraction Rate, c 35 mole % ucl, 233U/234U ratio of 7 b = blkte U extraction rate (fraction /s) (fraction 5-1) 268 3.5.7 Reactor Design: Configuration Trade Studie 3,5,7.1 Inner blanket study. Taube studied [17] inner blankets up to 1.1 m in radius for spherical MCFR(Pu) and (fluoride) MFFR(Pu). 1In each case the outer blanket remained cne—meter thick. With pure fissile (no fertile) salt in the core, the BG slightly increased ‘for an MCFR(PU) and decreased for an MFFR(Pu). Fig... 3e5~12 shows the changes by zone in the MFFR(Pu study: inner blanket breeding increases with its volume, vat also softens the core neutran spectrum. The result is increased core BG but lower total 3G, more so for F than Cl s2lt. Thus for a 0.207 ' 3¢ inaer bl. 0.10 Helative BG care BG 0.00 010 BG outer bl. R =160 cm R =280 cm . I 1 o 1 0 1 2 2.37 3 Relative Inner Blanket Volume. . Fig. 3.512 Impact of Inner Blanket Volume on BG far a MFFR(Pu) 269 fixed reactor dismeter, which controls the overall plant size, an inner blanket does not appear worthwhile relative to just increasing the out- er blanket thickness. Furthermore even if marginal BG advantages were found, the added engineering complexity wofild probably discourage it. To verify the effecfi of an inner blanket a study was made in spher— ical gecmetry. Figures 3.5-13 and 3.5-14 show the results for‘a censtant total blanket thickness of 200 cm. As the IB diameter increas- es, the annular core moves radiaily cutward. As a result 1. The core volume increases, decreasing the power density 2. The core spectrum softens 3+ Neutron leakage from the reactor increases Le Inner breeding increases and cuter breeding decreases resulting in a strong net decrease The accelerating BG decrease is primarily caused by the increasing neu- troan leakage and secondly by the softened core spectrum effect on BG potential, 3.5.7.2 Cptimum outer blanket thickness. A study was made to determine what constituted an infinite (cuter) blanket thickness to BG. It assumed no inner blanket. The study cansidered ThClL/UClB molar ratios in the core of 0,1, and 3, The NaCl molar content in the fuel salt was 30%. The results (Fig. 3.5-15) indicete that 2 m is near in- finite. Figure 3.5-16 shows that the ocuter blanket also shields the vessel from the high core flux: a two-meter thickness reduces the damage (> 0.1 MeV)} flux by over 3 magnitudes do:n to the level of the delayed neutrons. 270 Median Core Flux Energy (MeV) BG 0 40 80 120 160 Inner Blanket Thickness (cm) Figure 3,5-13, Decline of BG and Median Core Clux Energy with Tncreasing Inner Blanket Size Bases: = 30/35/35 molar proportions of Na/ThCl4/UCl3 in the core « 200 cm total (inner + outer) blanket thickness - Ni reflector 3.5.7.3 Reflector. A pair of ANISN calculations determined the effect of reducing the reflector from 4O cm graphite to 20 cm USinE:a,keff"Precision of 0.001 or 0,1%, P1 cross sections, and 1 S8 quadrature. With a 2 m blanket, the BG dropped from 0.3448 271 Neutron Flux SRR DU r"___ T -2 2 : P N7 total (n em ~ sec ) - | BNYLAN] in Reactor B b \t o s AN NN fofi/ | 7 0,1 Mev Vessel Wall Outer Blanket Right Leakage (neutrons per fission source neutron) 0.01 0 50 100 150 Inner Blanket Thickness (cm) FPigure 3,5-14, Increasing Neutron Leakzge into Reactor Vesgsel Wall with Increasing Inner Blanket Size, Bases: - 30/35/35 molar pronortions of NaCl/ThCl¢/UCI3 in the core -« 200 cm total (inner + outer) blanket thickness - Ni reflector 272 BG Blanket Thickness {(m) Pigure 3¢5~15. Determining the Outer Blanket Thickness Which Is Quasi=-Infinite to the BG, for Two Core Salt Mixtures Bases: blanket salt = 30/70 molar NaCl/ThCl 4 ~ 40 cm graphite reflector <~ no inner blanket - 30% molar NaCl fraction in core salt 273 104 Neutron Flux in Reactor Vessel (n cm-gsec-T) 10'3 1ol2 Total Flux (ThCl¢/UCl3 = 3) 10! 0e? MeV flux ‘ - (Thel,/ucl, = 3) 0 1 2 3 4 > 4 3 Blanket Thickness (m) Figure 3.5~16, Typical Decrease in Vessel Flux and Damage with Increasing Blanket Thickness, Graphite Rellector ThCl4/U01 Molar Legend Symbol Ratio in gore Salt Flux 0 - 7 01 MeV 3 total + 3 > 0e1 MeV A 5 > 0.1 MeV 274 to 0.3442 or 0.2%. The critical radius increased from 26,11 to 26,1, cm or 0.1% (a critical volume increase of 0.3%). Thus cutting the reflector thickness in half barely shows up above the accuracy of the calculaticns. 1his result cerresponds well to that of the previcus secticn: namely that two meters censtitutes a near-infinite olanket thickness. Comparisan of C and i reflecters far a 3/1 Th/U core ratio and 5m thick blanket showed no detectable differences. For a 0/1 Th/U core ratic and 1-m thiek blanket, C performed much better (BG=0.106) than Ni (BG= 0.0L0); the critical core rziius differed little. The BG difference probably stems from the moderating power of C: neutrons reflected by C stand a much better chance of capture by Th than those reflected by Ni. This recommends C for cases where.a non=infinite blanket is used. For > 1.5=2.0 m thick blankets and 65/35 ThClh/NaCl mixture, a neutron re— flector appears unneeded. 3_.5.7.4_ Core tube material, In an MCFR (Pu) study (751, & 20 MO/SO Fe alloy for the core/blanket wall reduced 3G by 0.04 relative " %0 a C wall., As the 20% Ho dominated the neutron absorpfions there (617), 100% Mo tubes might Teduce BG by C.2. The MCFR(Pu) study assumed in~core cooling with the tubes occuvying 5.9% of the core volume, In the present MCFH(Th) concept the 2-cm tubes occupy several times thai proportion of volume but they lie more on the periphery. Thus a 3G effzct similar fo 0.2 might be expected. 2775 3.5.8 Reactor Design: Controlling the Core 233U/231‘U Ratio Je5.8.1 Basis for studying the ratio. 23 I"'U cantent affects the Section 2-adduced goals: hard neutran spectrum, high BG, and mine irum presence of Pu and other heavy actinides. Although 23 L‘U cantent is not a design (independent) variable but rather a physics consequence (dependent variable), still design variables like NaCl/ UClB/‘IhCJ.l+ molar composition can influence it. Therefore we study its effects as if it were an independent variable, loocking at 233 v/ 23y o 3,45y and 7. The MCFR model used as a basis in the study included a core salt molar composition of 30/70/C NaCl/UClj/ThClh and a 200 cm blanket, 3.5,8.2 Effect of the ratic on spectrum, BG, and Pu presence. Fig. 3517 shows that 23l‘U softens the neutron spectrum (( E} core decreases. This is probably due to a higher {G’inel) / ratio for 23l‘U canpared to that for 233 Ue 23 LU also reduces BG and power density. Table 3.5IX shows that the major BG reduction stems directly from the lower (’U(S;-— Oc_> value for 23,4‘[1, not from spectral softening. This same phenomenon decreases the reactivity of high 23 LU systems: that accounts for the larger critical volumes and lower power densities. Intuitively, a high ratio will minimize the presence of Pu and other heavy actinides (the third goal): 234 heads the chain that leads to them. In summary, a high 233U/?'B}‘U ratio promotes all three goals of hard neutron spectrum, high BG, and minimal Pu presence. 276 Average Core Flux Energy (keV) Power Density Breeding Gain Figure 3.5-1T7. Correlation of Decreasing 234U Content With Key Core 880 860 840 820 27 26 25 23 0.34 Q.33 Ce32 0431 233U/234U Ratio Physics Parameiers 271 Table 3.5-7% Calculation of Neutron Procduction by the Two Principal U Isotopes 233U/231»U value for (V) 6c = 6a) Ratio 2y 2y _EEEE_ 3 3.27 1.16 2.74 5 3.26 1.18 2.91 7 3.33 1.02 3.04 3.5.8.3 Zffect on reactor fissile inventory and doubling time. Because 23LU is a fissile MCFR fuel, the reactor U inventory varies little in this study. However, the U export rate (Zq. 3.5.6~16) does increase with decreasing 234y content, reaching 0.25 MT/yr for the ratio 233U/?'BAU = 7. Figure 3.5-18 shows the consequent effect on doubling time. This recommends a high 2'33U/231‘U ratio; however, significant effort to reach higher ratios may not be warranted. 273 DT (years) oL 30 4 =i e 3 5 T Core Sait Z*u [P Ratio Figure 3.5-18. Variation of Doubling Time with Core Salt 233U’/234U Ratio Basis: 30/70/0 NaCLfUCl3/ThCl4 molar szlt composition in the core, 200 cm blanket, C reflector; b = ¢ = 16-7/3 3454804 Deducing the ratio on equilibrium cycle. Given a set of reaction rates and choosing b and c, the equations of Sece 345.8.2 specify the 233U'/234U ratio, However, this ratio for the fuel cycle will depend on the reaction rates used. These, in turn, depend on the ratio assumed apriori in the neutronic analysis of the equil- ibrium fuel cycle. Figure 3.5-19 suggests that the two converge near 233U/234U = 95 for b=c=10"| and near 123=13 for b=c=10‘6. This is for a 30/70/0 IIaCl/UCl3/ThCl , molar compositions Similar behavior is presumed for other MCFR models., The results of later calculations with the refined set of P1 Cl cross sections (Sece 3.5¢342) suggest a slightly lower value, nearer to 233U/234U = 8% and 115 for b = ¢ = ‘IO-7 and 10-6, respectively, 279 -6 Core 233U/234U Ratio =c=10 /s oa Equilibrium Fuel Cycle Deduced from ANISN-Czalculated 10 _“:-“_:5{“:;:::_5;_:7:_'_' Reaction Rates e b = C = 10-7/5 8 e —— 3 p) T Core 233U/234U Ratio Assumed in Calculating Reactiion Rates with ANISY Figure 3.5-19. Calculaiional Approach to the Equilibrium Core 233U’/23‘1U Ratio Bases = 30/70/0 molar p-oportions of Nacl/ThCl4/UCl3 in the core, 200 cm blanket 3.5.9 Reactor Design: Choosing the Core Salt ThClL/UClQ Ratio S - To approach its BG potential (BGX or BGP) the reactor must capture its excess neutrons in fertile material. One means of re- ducing neutron leakage frem the reactor is to thicken the ocuter blanket (Sec. 3.5.6.2) so that the fertile fuel captures or reflects most of the neutrons which escape the core. Another way is to add fertile material to the core fuel thereby reducing the core neutron leakage. However, this also softens the neutron spectrum, which lowers the BG potential (BGP). With an MCFR(U/Pu) Taube reported [17] BG up to 0.7-0.8 with such techniques. The following subsections study the optimum fertile content in an MCFR(Th) core. They assume a salt mixture of 30% 280 NaCl and 70% (UCl3 + ThClh) and a 233U/231‘U ratio of 7/1. 3.5.9.1 Effect on BG. Fige 3.5~20 shows the results of ANISN calculations. Neutrons leak significantly from a one-meter blanket so the choice of reflector affects BG. Then cne prefers ThClk/UCl3 = C. When the blarnket thickmess exceeds 2.0 m however, few neutrons leak and the reflector has little effect. [Th] = O then produces the greatest BG because it engenders the hardestneutron spectrum (Fige 3.5-21) and consequently, the most excess neutrons (Sec. 3.5¢5¢5). Fig. 3520 also shows the BG from the 1956 ORNL study [13] of an MCFR(Pu) which included a Pb core-reflector and a graphite-moderated thin blanket. The larger ¥ value for Pu causes the high BG. Few neutrons leak despite low fertile content, but this is because of high (83%) carrier salt content. BG could actually be higher were it not for the spectrum softening and neutron absorption by the carrier salt. 3.5.9.2 Lffect on reactor fissile inventory and doubling times. Substituting ThClh for UCl3 decreases density, However, the critical core radius increases near proportionally, and volumé increases even faster. Thus, overail, Th dilution increases the spherical critical mass U (Fig. 3.5=22) and the corresponding mass in the tubes Us (from Fig. 3.57). In contrast, the primary circuit volume outside the vessel, Vo’ remains fixed, so U, decreases. While V_ exceeds V., U,=U;+U _+U decreases. (Ub plays little role while b210-7.) Eventually the Th proportion reaches a level where the internal volume Vi exceeds the fixed V,. Then Ut begins 281 L T e . s p— ~__Pb and C,Z0 cm U0, HCFR(Pu) .- Breeding Galn Fertile/Fissile Ratio in the Core Salt Figure 3,5~20, Effect of Fertile Presence in the Core Salt upon BG: Lowering of the BG Potential; Trapping of Leakage Neutrons Legend: C, 200 implies a 200 cm ThCl, blanket and a C reflector(s). The MCFR(Pu) reactor features interlayered blankets and reflector/moderators © carbon reflectors, MCFR(Th) x Ni reflectors, MCFR(Th) A Be and C reflectors/moderators, HCFR(Pu) 282 900 800 - Median 700 ' Core-—- 600 Neutrecn Erdergy 5C0 " (keV) L0OO 300 o 1 2 3 L 5 Th/U Ratio Fig. 3.5-21 Effect of Th Presence in Core Sal;c, Upon Core Neutron Spectrum to increase with increasing Th/U ratio. This U -minimum occurs near ThClh/UClB = 1-2, In the 1956 ORNL study [13] of an NCFR(Pu) the minimum plant inven- tory occurred a.1:: a 238U/Pu ratio of 2 (Fige 345-23), similar to this work. The MCFR(Pu) total fissile inventory however lies roughly a factor of 3 below that for the present MCFR(Th) study: the reason is great dilution of the Pu fuel mixture by carrier salt(83% melar content). In Fig. 3.524 the BG export rate fJe for b=c=10-7/ s (defined “ by Eqe 3.5.6=16) stays fairly constant as the core Th/U ratio changes. By Eqe 3.5.6=17 then, DT behavior must closely follow that of U L Fig. 3.5-2l shows that the lowest DT clearly occurs at a ThCl L/UC]'B ratio £ O: somewhere around 1-2. 283 U Inventory (u) T | . . X ! == primary circult . outside the- ;rfikfij vessel | T, | | Te0 spherical ; core 1 calculation Th/U Ratio in Core Salt Figure 3.5-22, Dependence of the MHCFR(Th) Reactor U Inventory upon the Core Th/U Ratio, 284 Total Plant 2395, Inventory (+T) 23811/‘23 Pu Ratio in Core Salt Figure 3,5~23, MCFR(Pu) Plant Pu Inventory with a 3 NaCl, 2 MgCl,, ! Ucl3 (Puc13) Core Salt, Ixternal holdup volume = 3,5 m3 285 40 3 Doubling Time, DT (years) BG Ixport Rate, Ue (tr7/yr) Th/U Ratio in Core Salt Figure 3,5-24. Variation of BG Export Rate and Consequent Doubling Time with Core Th/U Ratioa 286 3.5.7.3 Effect on power density and flux levels The power density limit of 10 Mi/liter (Section 3.2.2.2) reflects more a premonition cf problems than a recognition of any specific. ones, Still, flux levels in the core tubes and the pressure vessel closely relate to power density (section 3.2.2.3). Figure 3.5-25 shows how ThClA dilution of the care salt markedly reduces power density and fluxes alike. The skewed-tube geometry will further increase the critical mass (lower the power density). A cylinder with length-to-diameter ratio (/D) near 1.0 requires about 10% more critical mass in low density metal sysiems than does a sphere; with L/D~2, about 40% more, The heterogeneity of multiple skewed tubes further increases critical mass as does neutran capture and scatter by the blanket ThClL in between the tubes. Altogether these effects should reduce power density and fluxes by more than a factor of two compared to spheres. Aplied to Fig. 3.5-25, the result is that ThClL/UCl3 = 0.1 now satisfies the power density criteria of < 10 iW/1. At higher ThClh/UClB ratios the > 100 keV flux in the vessel “Lor ¢ 102 1 o2 in 30 years. That falls below 101‘3 n cmm2 S level will allow operation over the reactor life for many metals. Meanwhile the core flux level will be «~--102'3 n cm-z mo—l. Presumably, most materials won't be able to take many months of that. This emphasizes the importance of our design: straight tubes, easily replaced, and not touching one another. 287 Power Density (/1) Flg\ll‘e 30 5-25. 100keV) Power Density Th/U ratio in Core Salt Reduction in Flux and Power Density as Th Presence in the MCFR(Th) Core Salt Softens the Newtron Energy Spectrum Bases: 2 m C reflector, 233U/234U in core 7:1, Na/Actinide in core 30/70. 288 233 10 U Isotopic e e e 100 keV Flux =2 =1 necm S ) in Rezctor Vessel Core Fover Density (mA) UCl y Holar Content (%) in Core Salt Figure 345-31. Effect of Core Salt Dilution on Flux Levels and Power Density: X - NaCl Substitution, O = ‘I‘hCl4 Substitution Basis: 7/1 233U/234U ratio in core salt 200 cm Dblanitet Starting core salt composition:70% U013,30% NaGl,0% ThC1, 295 Table 3.5~X Some Pertinent Microscopic Absorption Cross Sections Cross Section (mb ) Element Core* Blanket* Cl 15 6.5 Na 1.0 1.7 Th 230 11 * As calculated for an MCFR(Th) with core salt mixture at 30/35/35 NaCl/‘UCJ.3 /ThCl , molar composition. changes increase the ZBAU content. 233U capture increase also de- creasesthe 233U content, giving a double effect. All three effects work in coucert to decrease the ratic (Fig. 3.532). Differences between NaCl and ThClh as salt substitutes appear to be small. NaCl mostly scatters elastically while ThCll+ scatters more inelastically. The different shapes of the NaCl and ThClh curves in Fig..3.5-32 may reflect these differences: Th inelastic scatter should contribute more to spectrum softening at low non-fissile con~ tent where the neutrons still exceed the inelastic threshold energy. 296 T e e Go T e e 2T Bquilibrium e e ey T 8. o Ratio in R Ll TR T i S P Q TTTTLVTL o = o Core Salt T L - S Te x - - -’ - 6. | o Tor 0 20 40 60 80 ucl, Molar Content (%) in Core Salt Figure 3,5=32., Effect of Core Salt Dilution on the Equilibrium 233y/23%y Ratio in the Core Salt: X - NaCl Substitution, 0 - ThCl4 Substitution Basis: U reprocessing rates b = ¢ -7/ =10 S reaction rates based on 233U/234U ratio = 7 in core salt 200 cm blanket starting core salt composition T0% UCl,, 30% Nacl, o ThCl, 297 3.5.11 Mutation Effects J.5.11.1 Significance of actinides which emit alphas. Reactor irradiation transmutes actinides into long-lived alpha-~emitters: they increase in atomic number (2Z) and weight (A) through (n,Y¥ ) reactions and /fi —decay; they then generally decrease slowly in 2 and A through alpha decay, finally becoming a stable nucleus. Alpha emitfers are undesirable because 1. 2. 3. Le They pose a long-lived radiobiological hazard They produce neutrons through ( & ,n) reactions Some produce neutrons also through spontaneous fission They destroy reprocessing chemicals The radiobioclogical hazard wi-l depend upon 1. The probabili‘h;y cf the isotope reaching the p;lblic en- vironment in an ingestible form Its radioactive half-life: very short-and very long-lived isctopes present little danger The biological~eiiminaticn half-life; . how leong before the body removes it naturally The chemical affirity of the isctope for spaéific crgans or other members of the bcdy. Thosz which seek out bones are labelled carcincgenic (cancer causing) The available energy per decay The cccurrence of secondary radiations by reactions such as (oL,n) ffir high energy alphas The quantity of isctope present. 298 Combination or these faciors defines the toxicity of an isotope - just how much of a practical threat it paces, Table 3,5XI includes a tox~ icity classification along with other pertinent physics information, The moderate helf-lives of Pu plus their early preeminence in the act- inide burnup chains warrant their highly toxic classification, Carcin- ogenicity further emphasizes avoiding their.production, The problem with (o(,n) reactions as well as spontaneous fission neutrons is that neutrons penetrate gamma shielding: in addition +o harming workers directly, they can activate the surroundings, presenting longer-term hazards. Yumerous light elements (up through Al and Si) underzo these re- actions with alphas, Fige 3e5=33 shovws some relative yields, F, though not included there reacts almost as strongly as Be., F, Li, and Be are candidate materials for (blanket) salt mix; this phenom- enon argues against their use unless the neutron sigmal and hazerd are desired for non-proliferation pur-oses, Section 3.7 mentions the use of Iy O, ligy 2nd a with various revrocessing schemes; = 10 mos. estimated in Seces 3¢%5¢4e2. The several-magnitude higher 232U concentration in the core compared to blanket reflectc the 1000-higher production coefficient CX23N in that herder neutron spectrum, This means that fresh blanket U will contein not only far less rfission product hazard but also far less 232U hazard., Thus it constitutes the biggest threat to prolifer-~ oo ation., It is therefore very important to mix it with irradiated core U so as to "spike it" into a2 less proliferation-hazardous form, 301 Core Blanket Time in Reactor (Months) 232 Figure 3.5-34. Buildup of U in Core a2nd Blanket Salts 302 e would like to ectimate what level of radiation hazard this 2321} dauzhter activity actuzlly poses to thefi, diversion, and weapoas manufacture, The minimum critical mass of 233{1 for weapons use will nominally be 6 kg [139]. With a U metal density of 19,2 g/cm3 (50], that implies a radius of 4.2 cme The corresvonding flux from a sphere of such materizl is given [139] by S $=,-[2Ra - (@ -RY) n[(g-2)fa-R)] for a2R where S = uniform source rate = >\18 le (see Eg, 3e5¢4-8) 2 = radial distance from sphere center R = sphere radius For a = R ¢- = 2 At a distance of 1 meter from the sphere d) = ;j:m [ZR(loo-t-R)— C;Z,"f-ZDcR + lo“'_fiz) In (zr-z +1oo)] |00 For 2.6-MeV photons the dose conversion factor is 3.6 x 10—6 P.em/";fi' per unit Y flux [81-, Then the dose _ -7 [ Na { _ 2 = 37 <10 [;}-;aj{xs)u)m ;FEE 303 ' 3 N } z | - -0 Nz ——— F. . = 40xI0 [ T { )‘\3>32A08 for a = 1 ‘From Figure 3.5-34 [N22/N23] is typicelly 10-3. Fize 3.5~35 presents the results for De One sees that after 10 years the dose from 208Tl aprracaes D{surface) = 45 Rem/nour I D( 1 meter) = 45 miem/hour fl To remove most of this zad other radiation one could chemicaliy separate the uranium in the core sali{ I{rom the 208T1 and all other actinides and fission productis, However, this would not separate 233U from 232U. Then, in only fwo weeks the surface dose would again reach 1 Rem/hour. Purthermore, if only 99% removal efficiency was achieved, 2 3 Rem/hour surface dose from 208T1 alone would be present in the interim, From this one concludes that a highly efficient scientific group with laboratory facilities (such 25 a national laboratory} might be able to make a bomb without getting a deadly radiation dose, but all others would encounter difficulty, In addition to these complications, the tell-tale 208Tl signals would continue to emanate from the separated waste as well as from the new regenerating source in the separated U, All these froubles must discourage subnational groups which contemplate taking and using 233U for weapons purposes. 304 S0¢€ Dose at Surface (Rem/hr) or Dose at Distance of 1 Meter (mRem/hr) Figure 3.5-35. Variation of Radiation Hazard from 0.1% 100 0. 01 = 1 hr 10 "mo 10 mo 10-3mo 1 day 1 vk, 232 Time Since Chemical Separation out of U Daughters from Fuel Salt 2325 paughter Content in 6 kg of 25°U I'ollowing Reprocessing The effect of such activity on a2 conventional solid Fuel cycle, would probably be to require costly shielding and remote handling fo protect the workers, This means in fuel fabrication vlants and, pos- ~ioly, 2t the end steges of drocessing dlanis. 4 molten salt fuel cycle autoustes the reprocessing far more; the shorter time out-of- core should also reduce the activiiy due to neutron absorption re- actions with each member of the decay chain. Meanwhile, in any breedins zain set aside, %the raediation hazard will stert adding to that from fission products a2ad actinides and in- crezse Tor several years, Thus the oiggest problem which 232U ore- zents in a2 molien salt thoriun cycle should ve transport of oreeding gain: radiction will make th2 fuel unwieldy (reavy shielding and re- mote handling) and hazardous, thereby somewhat unattractive to diver- sion and to uveczpons manufacture by subnatioral groups. Since 232U and 233U are chemically inseparable this radiation hazard will contiaue to manifest itself for many multiples of the 232U T2—year halflife, i.es for centuries. 3y leaving i ing - : y le€e « 3y leaving the breeding gein in chloride form it is ready Vo stariup another plant, without intermediiate handling, 3.5.11.3 Impact of fission preduct concentration upon neutron physics performance. ANISN calculations showed that a FP concentration of b 1.-1 5.8 x 10 aem b =0.10 [Ufissile] in an MCFR(Th) reduces the 3R by 0.007 cut of 1i.34 or 0.5%. Within the accuracy of the cal- culation, this amount of FP did not aff:ct the critical dimensions. Figure 3.5-36 shows the FP effect on BG calculated by Taube et al for an MCFR(Pu): [FP] = 2 x 1074 - 0.10[Pu] appears to cause a similar BR loss of 0.01/1.45 = 0.7%. [FP] = 1.0 [Pu] seems to produce a (proporiicnally) larger loss of about 10% in ER. 1e5 S 1.4 le3 - l.2 4 T ¥ 1 l T o2 ks L2 L) Fission Product Concentration (a cm—lb-l) Fig. 3.5=36 Impact of Fission Products upon the ER of an MCFR(Pu)[17] l1.-1 [Pu]core = 0.0021 a em ™ b 307 3.5.12 Summary Section 3.5 attempted a first cut at optimizing the reactor des ign and compesition and also analyzed some of the consequences there- from. The first six subsections defined methods, data, and models. Sec. 3.5.7 discouraged the use of an inner blanket and showed that two meters constitutes an infinite outer blanket, In that event a reactor reflector contributes little., In addition to achieving maximum BG, the 2 m=blanket greatly reduces the flux dose to the reactor vessel., Sec. 3.5.7 also favored graphite as the core tube wall material because Mo strongly absorbs neutrons. Section 3.5.8 showed tnat design decisions which favor a high 233U/23£‘U ratio in the core salt produce 1. A hard neutron spectrum there 2. Higher BG and lower DT 3. Minimal presence of Pu and other high-A actinides Fortunately, within the ranges of interest, all the trade studies indicate that the 233U/ZBI*U ratio on equilibrium fuel cycle will exceed 7. That is sufficiently high to achieve the above goals; additional major efforts are unwarranted. .Sections 3459 and 3.5.10 revealed that both ThClh and NaCl, acting as diluents in the core salt, greatly reduce flux levels and the amount of out~of-core fissile inventory. They also reduce BG. The combined effects of inventory and BG produce an optimum DT near a U molar composition of about 30% and a Th/U ratio of 1=2. This and the lower flux levels explain the preference of other MCFR 308 studies for a fissile molar content of 15=30 %. However, low fissile content will decrease ther233U/23AU ratio, an indication of greater transplutenium buildup. Thus the true optimum remains to be found and will depend on final assignment of priorities. DT of 20-30 years appear feasible, This is high compared to quoted DT 13-20 years for an MSBR on the Th fuel cycle. However, an MCFR(Th) should build up less transplutonium. Efforts to lower DT should focus on the out=of-core inventary. Note also, that, compared teo conventional reactors, the absence of fuel fabrication and decladding operations and the minimal trans- portation and storage needs provide additional fuel inventory ad- vantages which are not taken into account here. Section 3.5.11.1 quantified somewhat the basis for avoiding transplutonium buildup. Sece 3.5.11.2 pinpointed fresh blanket U as the greatest proliferation hazard. By immediately mixing it with core salt, one adds not only fission product activity, which is chemically removable, but also 232U—daughter radiocactivity which is not. The radioactivity will discourage U handling, but not make it impossible. Sec. 3.5.11.3 showed that the presence of too much FP can sericusly redace BG. 309 3.6 Safety & Kinetics 3e6.1 Physics of Reactor Safety and Kinetics 3.6.1.1 Effective delayed neutrcn fraction. Delayed neutrons frem fission cf 2331;, 235U, and 229Py amount to 0.0027, 0.00635, and 0.0022 of the respective total neutron yields., At first glance this suggests that the effective delayed neutron fractio for 7 B/vers 233, . 239 . : a U=-fueled }CFR exceeds that fer a Pu~fueled [MFBR. However, most cf the MCFR fuel lies cut-=of-core in the primary circuit. This makes .3 .. much smzller. / eff Passing through a core less than 2 meters high, with a ve- locity of 9 meters per second, the fuel is present for only 0.2 seconcs. crcach decay half-i:fe of thne six delayed=-neutron groups (Table 3.6=I) far exceeds that. Therefore the delayed neutrons originate near uniformly throughout the primary circuit and /éhff diminishes by the ratio of fuel in-core to that in the total cir- cuit: by 0.2-0.3 to 5eff‘= 0o 0005~0,0008, 3,6.1.2 Prompt neutron lifetime,l . 1 increases with the median fission energy Z_, ranging from 5x1071 for an LMFER (190 keV) to 3:{10_'9 fcr Jezebel and 233U Godiva (near 1.6 MeV). For the inter— meciate MCFR(Th) with B = 400-700 keV, 1 should = 10‘8 st 3.6.1.3 Temperature ccefficient of reactivity. Reactor stability depends on several coefficients of reactivity. In a molten salt reactor, the temperature coefficient due tc expansion dominates. In an MCFR the very hard neutrcn spectrum nullifies any Doppler ccefficient of reactivity. Groug 1 o W - W n 2 ) Table 3.6-1 33U Delayed Neutron Characteristics Haif-life (Sec) Relative Abundance (%) 55. 9 21. 30, Do 25. 2. 28. 0.5 5e 0.3 3. Section 3.3.5.6 indicatzs a typical temperature coefficient of salt density of 0.0004 %4/ £ /O C. ANISN calculations (Section 3.5.8.3) reveal that 0.050 4, lowers reactivity by 0.034 Ak/ Xk, a coefficient of 0.68 AL'/ k / Af’//: . Applying the above gives -27 x 10 5 Ak o -1 . — € . Using /Beff = 0.00068 one gets a strong negative temperature coefficient of =40 ¢ /°C. For comparison, 2 Pu=fuelied LMFBR typically exhibits ~0.4 to =0.5 ¢ /°C [140], assuming Bopr = 0,0034 €alculation for a 1000 Mie MSBR gave an isothermal temperature coefficient of «0.9 x 1077 Ak fx /°c [131]. However this comprised a =4.4 x 107 Lk/k/°C for the negative Doppler coefficient of thorium and positive coefficients for the thermal spectrum graphite moderator and fuel salt density. The latter stems from a neg- ative salt density coefficient of -=0.03 Ak/k/ A/a//o for their low-actinide fluoride salt mix. 3t 3.6.2 inalysis of Normal Cperations 3.6.2.1 Reactor startup. MCFR coolant circuits and salt supply will need preheating just like an lSBR does, A salt of subcritical fuel conicentration or even a low-melting non-fuel salt could {iow test the circuit. Cne might approach zero power criticality and sub- sequent rise to power by gradually adding fissile fuel through the clean-up loop to a suberiticai fuel salt. 3.6.2.2 Reactor control, A iimited study [29] indicated feasible control of reactor and power level by menipulating-the secondary circuit conditicns: wmainly varying the flow rate of salt through- the ccre and heat exchangers. The negative temperature ccefficient will hold the salt temperatire nearly constant at all power levels. The absence of tke heat transfer and fuel pin temperature differentials of a solid-fuel reactor allow this direct respcnse. 3.6.2.3 Reactor stability and inventory ratio. The amount of fuel in the primary circuit outside the core fixes the effective delayed neutran fraction and affects the reactor stability. The British [29] showed that an HCFR(U/Pu) should be stable to small reactivity perturbations if the fraction of primary circuit salt in the core ex- ceeded 25he Engineering studies of their MCFR designs met this re- quiroment with a margin, The primary circuit flow rate seemed to. have little effect upon the delived neutron -concentration in the core, 312 Table 3.6-II Fuel Inventory in a 11 GHth MCFR(Pu) Plant [23] Specific heat exchanger power (canservative data) Total volume of heat exchanger for 11G4(th) Puel frazction -of heat exchznger volume Fuel volume in heat exchanger Fuel 1n the ductrwork Total fuel out of cors Fuel in core Total fuel in system Mean specific power of fuel in the whole system 11GW th - 5.3 Plutonium content of fuel Power rating of whole system lKZ-I/cm3 il m3 0.3 3.3 1.0 L3 o 1.0 0 5.3 ° 2.1 Gi/r 0.8 gPu/cm3 0.385 kgPu/MW th Taube anaiyzed a 11 G{th MCFE(Pu) and inventoried (Table 3.6-II) of the fissile fuel salt is out of the core. 313 the fissile fuel salt in the primary circuit. Assuming a power density of 11 Mith/liter and conservative heat exchanger pericrmance, 81% For a given heat ex- changer rating, as the total plant power decreases the inventory in the heat exchanger will also by about 0.3 m3 fuel per GWth. Littie change occurs in the core or piping. Therefore according to ths British analysis, this plant should operate at lower power. Table 3.0~I11 compares Taube's results to those of 3ritish de- signs and a 1956 ORWL study. The 3ritish designs contain only 56-645% of the primery circuit fuel salt out~of-core, revortedly mostly in the duct work, Unlike Tzube but in agreement with the British designs, the OREL/1956 study also aftributed 2 large portion of the inventory to piping between core and heat exchanger. Table 3.6-II1 Volume Distributions in Some MCFR(Pu) Designs crRL/56 [13] Taube 23] British [29] Vol Distribution Volume Distribution OLUMe (ut~of-core um out=of=core Quaiitative (m3) (%) (m%l_ (%) Proportions Core 3.30 1.0 HX plena 0.35 10.0 HX tubes 1.62 L6.2 HX subtotal 1.97 56.1 3.3 76.7 Piping 1.54 L3.9 1.0 23 .3 dominant Qut~of=core 3.51 100 L3 100 Subtotal Total 6081 5.3 % Out=of= . 51.5 81.1 56=64 core 314 Fige 3e6-1 shous a typical British design [29] with large aeat exchangers to compensate for lower thermal conductivity values for the fuel salt, They have given considerable thought to reducing duct iength but find that unconventional methods tend to be less relizble or less easily maintained, Increasing fuel velocity can reduce duct cross sectionzl areas, but limits due fo vibration and puaping pover costs may occur, The choice of intermediate coolant will also affect oui-~of-core inventory by its influence on heat exchanger arrangement and temper- ature conditions, Je5e2s4 Reactor Shutdowne Basically the ICI'R requires no scram device because the strong temperature coefficient holds the system steady, Adding blanket salt or other absorber should quickly reduce power and temperature, Hovirng part or all of the molten core fuel fo geometrically-safe, cooled holding tanis will accomplish full shutdovm,. Criticality safety calculations must be sure to consider criticality 233 23 3 increase as “Ue. There decay heat will Pa decays (t1 = 27.04d) to <2 help keep the salt molten, while natural circulation cooling holds dovm the fuel vapeor pressure, Je0e245 Chanze in physical pronerties due to Iransmutations, As 232 233 Th and U transmute to actinides higher in Z, the individual salt melting points (excluding eutectic effects) lower (Fige 346-2). How~- ever, since uranium isotopes predominate in the burnup chain, very little higher-Z material appears, Similarly, boiling point and vapor pressure should also change little, 315 91t Figur@ 3. 6—1 ° I CoRg I mAMyt 3 ST IRaTRATE HMEAY [NCHANCAR - B ot 4 RAL AT A -t 3 MR XA - den b RAMRLT COOLAR MMP- 4ot T el IRATL (OO ANT HEADER 8 T Dxart CO AN 1M dT 1 TIREDATE COOy ANT OUTUEY O Gak v ( wuvy 1 el . Yeur ! L_i ‘! Typical MCFR{Pu) Design with Large Heat lLxchangers to Compensate for Low Thermal Conductivity of Salt [29] Melfing Point (°c) Figure 3. 6=24 700 ‘ - . - [ I oo N +3 chemcal stote-—- Koo —- - - - _—— L — —— L e — [ e e 90 91 92 93 9 (fn) (P2) (U) (o) (Pu) Atomic Number Helting Point of Some Actinide Chlorides 317 - 3.6,3 Analysis of Accident Situations Cantinual removal of velatile fission products frem the salts during cperation eliminates several obnoxicus species fram being present in an accident. Yany of the remaining hazards would stay in the salt. Thus, no accident can occur which corresvonds in se- verity of radiocactivity to the meltdowm of solid fuel systems. The major concern then becomes that no primary circuit failure lead to critical masses formirg. 3.6.3.1 Small leakages. Leaks bstween core snd blanket pose ne sericus chemical threat as the salts are compatible. However, other factors influence whether ti2 pressure difference should force core salt leaks into the blenket or vice versa, Higher temperatures in the core salt may produce higher wvapor pressures there then in the blanket. The core salt also circulates under pump pressure vwhile the blanket salt need not. If a reserve tank autcmatically replaced fuel salt leaking into the blanket, the system could simultanecusly approach supercriticality and high temperature, a strong temperature coefficient not withstanding. The high temperature could further abet the detericration. 1In contrast, blanket salt leaking into the core would anly dampen the criticalitys: hovever, this could result in degraded system performance. An anomalous rise in concentration of 233U or fission product in the blanket, or a rise in temperature there (due to increased fissions), ar a drop in core salt pressure would sigmal core-to- blanket leakoge. Leakage of blanket salt into the core would decrease the core mean temperature and blanket salt pressure. 318 The secondary coolant circuit should be pressurized slightly higher above the primary one, and the tertiary even nigher. This will cause less active coolants like helium or lead te leak into the molten salt. He 1s inert and the salt cleanup system, which aiready separates the fission product gases, would remove helium as well. Lead interacts with nickel-bearing allcys, but not with i{o cr graphite. Traps at certain points in the cirecuit wyculd iocate the positicn of .ead .eaxage. The ccntainment building will catch leakage cf volatile fission products from the primary cirauit or btlarnket tc the air, simiiar to MSRZ, with zppropriate detection and leaktight barriers, | 3.6.3.2 Loss of flow. Leaks, pipe break, heat exchanger plugging, and pump failure can all reduce flow in the primary circuit. Figure 3.6=3 analyzes the consequences. In most events the reactar shuts down and the fuel drains to a safety tank; th=re fission product decay and-delayed neutron-induced fission continue to generale heat, but the high capacity of the salt restrains the temperature rise, - The effect of a single pump failure depends upen the primary circuit arrangement. With single punps per channel (Pian A in Fig. 3.2~8) failure of cne pump would shut down one whole core channel. The fuel sait in that core chanrel must then drain out to stop full pover production im it. Subsequent replacement by void, blanket; or carrier salt would cause a large loss of reactivity thereby shutting down the reactor unless (1) The separation between chamnels was such that they were 319 Flow Reduction 02t or Total Loss One Multiple Pump Failure or Heat Pipe Minor Pump Total Power Failure Exchanger Break Leaks Failure (Electric and Diesel Backup) Plugging Suitch to Dumj» Fuel to Safety Tank Provided with Natural| |Spillage into Backup Pump Circulatory Emergency Cooling Epson Salt Bed Fige 3.6-3 Analysis of Primary Circult Failures individually near crifiical, and (2) Missing reactivity could be added through enrichment enhancement Then one could continue operation on a reduced scale, e.z. 6/7 for six out of seven channels still operating, until a more opportune time occurred in which to drain the whole core and replace mal- functianing equipnment. With multiple subchannels per core channel, each with pumps and exchangers as in Fig. 3.2-8-B, C and D, failure of one pump still allows partial reactor operation until a better time for shut-— down and repair. If cne operated the Plan B or C subchannels or | the plan D heat exchanger at less than full efficiency, then their full use is available in the event of one subchannel failure. This would however, increase out=cf=core inventory and capital costs though., Should the core tubes be connected in series as in Secticn 3.2.6.4 then any in~core malfunction (leak or pipe break) would require system shutdown. Should both regular and emergency power supply fail, all pumps would fail unless they are steam turbine driven. However, electrically driven pumps are easier to include in the containment and also easier to supply for pre-testing, etce Inertia would help electric pump run—down rates, or same short~term auxiliary supply might have to be provided until full dump of fuel has takenr place. The fluid could also drain through a "freeze wvalve" by gravity into a natural—convection cooled tank which has a non- critical geomeiry. In the event that flow ceases and core salt remains in place, 321 the strong negative temperature coefficient will control the temp- erature and power: initial temperature increase will decrease density, which decreases reactivity and thereby power, until some equilibrium state is reached. HCFR(Pu) studies [29] with a simpli= fied reactor model showed that the temperature rise of the salt for a reactivity step of up to $1 should be less than BOOOC; for 7 pumps failing out of 8, less than 230°C. In summary, it appears that reactor stability concern over out-of—core inventory may oreclude the use of cut~of-core subchannels as well as Aictate series ccnnecticn of the core chamnels. This means designing the system fer high reliability to minimize shut- dosm and for rapid repair of heat exchangers, piping, and pumps wnen shutdown does occur. Rapid repair will entail expedient removal of all the core salt followed by a salt flushing of all radiocactivity from the system. The strang temperature coefficient should disspell any criticality accident concerns. 322 3.6.3.3 Structural failure. Failure of the tubes separating core and blanket would lead to no chemical or cempatibility problems. Dilution of the core salt will reduce reactivity; anly if core salt were replenished as it entered and displaced the blanket, could a reactivity increase occur. Even then the system could accomodate temperature rise from moderate salt additions, as ample margin exists abecve cperating temperature on a short term basis. The limit to the permissible rate of such salt addition needs to be established. Vessel failure would reguire rapid dumping from both the circuit and the catchpots. Dump tanks should probably be sized to centain the cantents of cne secondary coolant circuit as well as all core and blanket salt. Relief valves on the secondary system will protect the primary circuit from pressurizaticn result- ing from a major steam generator failure into the secandary system. In the event of a major circuit failure, operating pressure would play an important role in the subsequent fate of the fissicn products and the containment, The molten salts themselves ex- hibit low vapor pressures. Although high pumping losses can cause primary circuit pressures up to 460 psi, lead secondary coolant acts essentially like a hydraulic system with little stcred energy. Missile formation is therefore unlikely and it should be possible to demonstrate a containment which will not be breached from this Cause . 323 With a high-pressure helium—cooled system one must ask whether an accident might aerosolize the core salt. However, a properly=- designed reactor vessel could withstand the full helium pressure from a severe rupture between the coolant and primary circuits. Thus release of activity te the reactor containment cell would not occur except under a simultonecus double failure. The cell itself is small and can economically be made in the form of a prestressed vessel to withstand missile damage and to act as an additional barrier to fission product release. A final low pressure building cantainment would prevent release from smell leaks in the earlier centainment stages. 3.6.3.L Emergency cooling. Unlike solid fuel, molten salt can quickly transfer into a geometrically-safe tank. Cravity can provide a failsafe transfer method from reactor to holding tank (Figure 3.2-11): a pipe wherein the salt is normally kept frozen weould open naturally upcn loss of electric power. In the tank, the salt will circulate naturally as heat from fission product decay transfers through tube walls to air convected by natural draft towers. Ctherwise pressure would build up from salt vaporization. The absence of mechanical moving parts will make this whole system hizhly reliable; multiplicity could add even further in- tegrity. A forced draft system would probably layout more compactly 324 A second cooling option would circulzte naturally a low melting- point salt or NaK through U-tuves in the tank to boiling water heat excnangers, Air-cooled condensers situated in a2 normel or forced draft stack would condense *he steam formed, Alternatively, the heat could transfer directly to a large boiling pool, thus accomodating decay heat for a protracted period without makeupe. Condensers or make-up water would be provided for coniinuous operation, A catchall salt bed below this apparatus would serve as a backup for any leaks or breaks, The heat of salt{ formation would greatly 2id in absorbing decay energy. An independent cooling system might remove decay heat. The bed wowld alsec dilute the fuel salt., Proper choice of tank diameter would insure subcriticalitye. Recovery from emergency tanks or beds would require heaters, drains, and pumps. When carrier or blanket salt is added for dilu~ tion and/or heat absorption by latent heat of formation, a method tc separate out the dump salt would z2lso be needed, 325 3.6.3.5 Comparison of MSRs to others. WMSRs should be biologically safer because l. The plant continuously removes velatile fission products 2. The fuel is already molten and in contact with materials designed for that candition. The removal of volatiles fram the primary circuit still requires attention to their presence elsewhere in the plant. However, it should not be difficult to insure the integrity of a storage medium below ground. After a suitable decay period scme of the gases may be releasable to the atmosphere. In the event a pump fails or other flow loss occurs, ths enly concern is that decay heat right build up vaporization pressure, However, the same colls which preheat the fluid could also cool the fluid. 3.6.3.6. Precipitaticn out of eutectic mixtures. After the temperature cof a salt mixture falls to the solidification point, the compositiom of liquid changes, sliding alang the liquidus curve as crystals separate out (Figure 3.6-4). With low initial UCly molar centent, cooling precipitates out NaCl crystals, thereby enriching the fluid in UClB. With high initial UCl3 molar content, cooling produces UCl3 crystals and NaCl-enriched fluid. In either case the liquid migrates to the nearest eutectic point (nadir) composition and causes a concenw tration of UClB. Thus one must consider possible criticality situ- ations and design geometries to prevent them, Probably precipitation of UCl3 crystals from a UClB-rich fluid is the less-seriofis case, 326 Initial Initial Composition Composition A 3 I Temperature All ATl yaCl UCl Molar Composition Fig. 3.6=4 Migration of the Liquid Composition Along the Eutectic Path as the Salt Mixture Cools 3,6.,3.,7 BRBoiling off cf salt mixtures. In the event that the carrier salt had a boiling point much lower than that of the actinide chlorides, one could postulate a positive temperature coefficient contribution in that temperature region. Fortunately evidence from section 53.3.3.5 contradicts this: carrier salts appear to be less volatile than actinide chlorides. 327 3,6,3,8 Containment. Solid fuel plants generally feature triple containment: fuel element, reactor vessel, and tha:n the contain- ment building. The MCFR will feature reduced FP radiocactivity due to continuous gas cleanup but it will inherently have only double containment of piping and 2 building. A second radiation hazard arises from activation of the secondary coolant and nearby equipment by delayed nevirons in the primary circuit. To further ceontain these harzards, a low-pressure leak-tight membrane might be formed arcund the walls, floors and ceiling of the reactor system. This membrane can also form part of the duct- ing for the inert gas circuiaticn required to cool the concrete structure and shielding and inhibit vessel oxidation. Heat losses with insulation restricting the concrete temperatures to below 7OGC would be about 200 watts/m2 with 40 ¢m of insulation. This would equate to about 3 M{ total heat removal by water or air cooling. The main building would constitute the tertiary containment. Its volume must be sufficient to contain the stored emergy of any gases present. In the lead-cooled design these are cover gas vol- umes al low to moderate pressure. The helium—cooled version may require a larger outer containment volume and/or intermediate pre- stressed concrete containment(Sec. 3.6.3.3) 3,6.3.9 Resistance to external threat. Cne can postulate a number of scenarios wherein external explosion or impact threatens th= reactor integrity directly or through a loss—of-coolant accident. These could involve ground or aerial bombs, planes, and space re- entry projectiles (e.g. meteors, missiles, space laboratory), by either accident or intention. An individual, a subnaticnal group, 328 or another country might intentionally initiate, Nature threatens with earthquakes, tornadoes, and dam breaks. In many or most of thase situations one could anticipate the danger. Most reactors already lie below ground level. Still the vessels generally sit high enough to be rupture~prone to strong explosions cr impacts. Such an incident could release volatized fission products and actinides of_consequence greatly exceeding a simple bomb cr other initiating event. To protect against these threats an MCFR could uniquely trans- fer its fuel to a storage tank (section 3.6.3.3), located remotely under additicnal earth (Figure 3.6-~5) or in an otherwise hardened "sunker". containment ground level core | storage tank Figure 3.,5~5 Hardened Starage Tank 3.6.3.10. Molten salt combustion support. Molten salt itself does not burn, but will support combustion with solids such as wood, coke, paper, plasties, cyanides, chlorates, and ammonium salts and with active metals such as aluminum, sodium, and magnesium. Water from spray sprinklers or low=velocity fog nozzles provides good fire protection. 329 347 Fuel Processing [17, 81, 141-146 ] This section analyzes the chemical aspects of {the reactor fuel cycle: principelly vreparation of the ThCl4 blanket fuel and repro- cessing of irradiated blanket and core salts, Some processing should occur continously, some batchwise, Ratch- wise will generally be easier, more efficient in separation, and more economical, This study does not deal witk the near-identical concerns associated with starting early cores up with PuCl3 or 235UCl instead 3 233 of ucl . 3 ThO2 is converted to ThClh, mixed with reprocessed blanket salt, and fed into the blanket simultaneous with blanket salt removal. Blanket reprocessing separates out bred'UClh(pre- dominantly ;233U01h},reduces it to UCL,, and admixes that to reprocessed core salt, forming the core shim material., This constitutes a spiked fuel, suitable for BG export as well as for positive reactivity shim to the MIFR(Th) core (simultaneous with irradiated core salt removal). Core salt reprocessing removes fission products, oxides, corrosion products, and chlorine iransmutation productsj part continuously and part batchwise. It alsoc readjusts the chlorine stoichiometry. 3.7.1 Principal Salt Reprocessing Methods The MSR ccnicept inherently aveids many of the conventicnal reprocessing stages, especially fuel element dissolution and refabrication. The absence of these stages removes the need for high decontamination stages in reprocessing. Proliferation ccncerns 330 also discourage high decontamination: residual radioactivity makes weapcn construction hazardous. Eliminating ail these stages racucss costs., This leaves just the separation stages: core sali cleanup and bred-U extraction frem the blanket. The most promising methods are by l. Solvent extracticn from aqueous soluticn 2. Volatility 3. Pyrcmetallurgy L. Pyrochemistry (molien salt electrolysis) e first review the chemistry of the heavy elements. 3.7.1.1 Chemistry of the heivv elements[31], The separation processes, especially those based on solvent extiraction, take ad- vantage of the scmewhat unusual chemical behavier of the actinides. Elements in the analcgous lanthanide series all exhibit similar chemistry: +the presence of three, relatively loosely-bound, outer electrons causes each atom to exhibit 2 positive valence of 3, The actinides also all form a tripositive (III) valence state. However, some also evidence loosely-bound inner elec- trens. This leads to tetrapositive (IV}, pentapositive (V), and hexapositive (VI) states. Tnese higher oxidation states evidence different stabilities (table 3.7-I1) which facilitates extraction and separation of the heavy elements. Th and U differ proncuncedly, Th evidencing mostly just the IV state. 3 Table 3.7-I RELATIVE STABILITIES COF OXIDATICN STATES CF THE ACTDNIDZ TIZMENTS [81] Atomic NOueases 89 90 91 92 93 9L 95 96 97 Element... Ac Th Pa U MNp P Am Cm Bk T tereces e * *? * HH M O SRR MR TV eennnnese HERE H AR R R X ** “]..'l'll.. **M* * H* H * V-i..-..... X Ak L * Legend - most stable state decreasingly stable The IIT and IV states easily preeipitate from aquecus soluticn as flucrides; the V or VI states do not. The fluorides of the IIT and IV states do not volatilize; the VI-state fluorides do at fairly low temperatures. The IV and VI states appreciably dissolve in certain organic liquids; the III-state nitrates remain virtually insoluble in these liquids. 3.7.1.2 Solvent extracticn from agueous solution. The extractiaon o; actinides from aquecus sclution by an organic solvent is the most advanced process: it has been widely used since about 1951. The Thorex version extracts 233U from aqueous soluticn of irradiated 1n fuel elements [72}. It relies on the stability differences 332 between Th and U in higher valence states (Table 3.7-1). The facility of solvent extracticon for multi-stage operaticn without consuming additional heat or chemicals particularly benefits 1. Situations requiring extreme purification, Use of enough stages can lower the gamma activity from fission products in the extracted urznium tc belcw that of naturai uranium. Thus cne might not want to furnish this capapility tc a foreign naticon. 2. Situaticns where th:s propertiies of two metals parallel one anicther sc clcsely that a single precipitaticn or crystallizaticn can not separate to the degree required. Thus this method may not be needed for Th-U separation. The dissoluticen step for a chloride salt has to be the simplest of all: Just add water. Different organic sclvenis separately extract the U, Th, and wastes. Addition of 012 and CCLL subsequentliy rechloridizes U and Th. Disadvantages associated with this method include l. Extra criticality precautions fer H-mederated fusl soluticns 2. Multipliciiy and complexity of steps 3. Large waste volumes I, Large shielded space required 5. Additional steps needed to produce solid wastes. 3.7.1.3 Velatility orocesses. Volatile UF (b.p. 56.AOC) easily separates from fission product and Th fluorides. The‘ability to de- contaminate to low activiiy levels parallels that for solvent evirzc—. tion, However, the volatility method requires fewer steps and therefore 333 smaller volumes of highly-radicactive wastes compared with aquecus pProcessinge. Volatilization should work especially well for processing a ThFh/%aF (cr other fluoride carrier) blanket salt mix. Cne would cxidize +he bred UFL to UFy, separate it by distillation, chloridize 4 it, and reduce it to UClL,. The remaining ThF 1 would return to the blanket. Distilling UClh, UCls, or UCl, from a mix of ThClb and chlorides of the structural materials and fissicn products would be more diffiw cult: more volatile chloridss compounds exist and their vapor-pres- sure ranges overlap, Howev:zr, considerzble less develooment effor< has been expended on method:s for separating volatile chlorides; the problems may be solvable, 3:7.1,5 Pyrometallurgical processing. In the 1950's ANL developed pyrometallurgical processes to recover and purify fissile and fertile material from breeder reactors. Although demcnstrated cn a pilot- plant scale, much engineering development remained to evolve a workable and reliable process, especially in fiiew of criticality restrictions. A typical process would use a molten chloride flux to contact the oxide fuel and a molten metal to extract the actinide. Hence the need for high temperature (pyro). With molten chloride salt fuels anly the extractant is needed: for chloride fuels, Dillon [28] proposed a Mg-Zn alloy. A similar metal may work with Th/U cycle chlorides. Pyrometallurgical processes decontaminate by factors of only 334 about one hundred. With them one must handle fuel remotely. This poses no problem for molten salt processing: it requires no fuel fabrication or other handling; everything is dcne remotely anyhow. Leaving in racicactivity also inhibits use of the material for Weapans. Because of its compactness, pyrometallurgical process can operate ciose-coupled to the reacter and cn a muich shorter cycle than the acuecus rcute. If econcmical, a pyrochemical processing plant could easily be acccmodated within the reactor building due to its small size., Preliminary work indicates that capital and operating costs may be high because of the small batch type operaticns needed. Detracticns include lcw recovery rates and the development needed to cope with temperatures zbove TOOOOC. High temperature €quipment of great reliability that can be operated and replaced remotely 1s hard to design and expensive to test. However, work already done indicates that the difficulties may be overcome. 3.7.1.5 Molten salt electrolysis. Techniques exist to deposit actinide oxides and carbides from molten =alt sclution through electrolytic raduction [143] e.g. U6,*% (seln) + 267 == 1UC, (solid) at the cathode. At ths anode: 2C1" (soln) === Cl;(gas) + 2e” Similarly Taube [147] mentioned reducing actinides directly in their melten chloride mixture. The Cl, released could be used to chloridize ThC, for blanket feed (section 3.7.7). 335 3.7.2 Core Salt Processing Cn equilibrium cycle, the core fuel will include carrier alkali salt (e.g. NaCl), fissile UC13, diluent ThClh, actinide transmutants (e.g. PaClL, NpClB, and PuC;B), impurity oxides, fission products in various forms and states (section 3.4.2.1), structural cerrosion products, and sfifn { from chlorine transmutation)}. Some of the mutants may fcrm éomplex chlorides like 052U016 angd compounds like UI and US, which precipitate out at sundry temperatures. Core shim material rmust replace irradiated core salt near- ccntinuously so as to maintain criticality. The fuel burnup rate’ (1.6 gm/min for a 2250 }fth plant) fixes the rate of shim replacement. Actual core salt reprocessing can still occur batchwise or continuously. The choice will depend in part on the allowable inventory of core salt in reprocessing and the allowable level of FP concentrations in the core. The latter depends an 1. The reactivity worth of FP 2. The effect of lower Th/U ratio or increased reactor size an reactor performance. 3« The effect of ¥P on BG L« Acceptable radicactivity levels in the primary circuit (high levels necessitate remote reprocessing and extra plant shielding). 336 3.7.2.1 BRecovery options. The first option is whether to clean up the salt (remove the bad part) or just recover the 233UCl3 and scrap the rest since lNaCl is so cheap. Solid fuel cycles conventionally take the latter approach. Here we choose to clean up the salt because 1. NaCl radiocactivity precludes easy disposal 2. Znrichment in 37Cl(if chosen) would forfeit cheap cost arguments 3. Discarding non-uranium actinides would mean poor fuel utilization since every actinide atom can eventually fission in this spectrum L. Continuous gas removal gives a gocd start on salt cleanup. It might be feasible to just remove the parasitic neutron ab— sorbers and corrcsion agents: let the salt accumulate the rest of the mostly-radioactive non-volatile mutants,some as substitute carrier salt. The acdvantages are 1. A suitable place to store non-volatile radioactive wastes 2. The reactor will transmute many of the wastes into a less hazardous form. 3. The radiocactivity will add to the heat source. L4, The radicactivity discourages diversion of the core salt for weapans purposes. Pctential disadvantages to watch for and control are 1. Change in viscosity and other thermophysical properties 2, DlMutant plate-ocut causing flow blockage, radiation sources, or other problems. The enumerated advantages seem real enough to warrant this basic approach; practicalities may require some modifications, 337 3.,7.2.2 Continuous removal of mutant gases. Fission produces Se, Xe, Kr, I,, and Bry, which are all gases at reactor temperatures (section 3.4.2.1). Threshold reacticons élso produce He and HT gas. Cne toa of fuel from the core of a fast breeder contains 2 x IOA Ci ot 8%z 100 0i of Be (after cooling 4 gonths), 130 and 0.7 Ci of LBl (after cooling 4 and 6 months, respectively), 0.13 Ci 1291, and 2200 Ci of T. In an MSR circuit rupture, these weculd of all present a radiation hazard. Gas accumulation will also build up pressure, affecting circulation, In remedy for the MSBR,CENL proposed a cleanup system which recirculates gas in lcops across each main salt pump. Injected helium nucleates bubbles which absorb gases, the small-particle fog of inert metals (Pd, Tc,Ru,Rh, and Te) and some of the volatile’ chlorides (Fig. 3.7~1). “hile in a hold-up tank to reduce decay heat, some of the metals and chlorides deposi£ out. Passage through traps and beds removes Kr, Xe, water, etc., before returning to the helium injector. About 20% of the bypass flow undergoes a long delay in which all isotopes except the 1(-year 85Kr decay to an insignificant level. Presently 85Kr levels from reactors are small encugh to discharge directly to the atmospherc. Should nuclear power abound in 30-40 years it may be necessary to separate out 85K; low=temperature fractionation looks like a promising method. Eveporation and fractionation can also concentrate tritium for lengthy storage; this is presently done in solid fuel reprocessing. 336 6 Lr 50 X {ast alow Jas 39 ar 153 1 PrE Estraction vary slow 3 3 52 Te 33 As 51 Sb PPS Volatile chlarides 32 de 5¢ Sn A N s 3 In Low volatile chlorides FPs 30 In "0 4 Nan volatile 15 Rh PP v Ry Bethls 1y T 2 Mo Low £l W Fes volatila cnloriles W Ir 64 04 61 Fu €2 sa Porlodlic Table &L Pm Non 60 Na PPA volatile enlerides 5§ o= 58 Ce 39 Y |57 la 38 3r [S6 Ba Low volatile chlorides 37 R |ss ca | FA Fig. 2.7=1 TFiszicn Producis in Molten Chlorides edia —A——L—-——n An intermediate half-life makes 137Cs ane of the more troublesome fission prcducts (Fige 2.1=2). Figure 3.7=2 shows that only a small portion of the A=137 yield comes directly to 13705. Though the gaseous precursors decay re- latively quick, continuous gas extraction could remove a lot of them before non-gaseous 13705 formeds This would 1. Lower the activity of the circuit 2. Avoid the more difficult removal of CsCl, or other compound later 3. Help isclate Cs for individualized volume reducticn. A 2 g J2 33 5 55 sg il s Tote [ lp w—Cyp —wgy AsL37 from — ) tisaion {@ j 1oa 10 Fi . . — - ependant lg 3 7 Tiels 10 Feprocessing Characteristics L 0.1 - of the A = 137 Fission s 1’ , Tavmdier anad salr oir Ca-13 a-1y7 rreduct cnain. ¢ T now Sacaonas 2 1o 10 1 v —— £xcract fan ~ Rate a7k L The main design parameters for gaseous fission product removal 1. The absorption required in the delay bteds 2. The heat removal needed to avoid excessive temperature rise in the charcoal traps. Cne form of delay bed im a trough of swimming pool size. Low pressure steam forms above the pool and passes to condenser qnits. Such a system would dwarf the reactor in size. A4lternatively, cne might store the gases safely and reliably at high pressure. 3.7.2.3 ‘Removal of non-rasecus fission products. The inert metals (section 3.4.2.1) not escaping as fog will likely deposit in var— ious parts of the primary or FP removal circuits. 3Ixternal processing will remove the remaining 50-60% of the fission firoducts as nec-— essary unless on—line treatment processes can be developed. 340 Section 3.7.1 mentions three possible rrocessing methods, Cf these solvent extraction frcm aqueous sclution is the best established. Illcwever, ecanomics may require the reprocessing plant to serve nultiple reactor facilities. Reprocessing at a different location, would mean 1. Increased out-of-plant inventory 2. Increased nazard of radiocactivity, sabotage, and pro= liferation Ccmbined with the disadvantages menticned in Secticn 3.7.1.2 this argues for consideraticn of the less—developed processes of sectiecn 3.7.1. _ 3.7.2.4 Contrcl of the oxy-en levels, Cxyzen and oxygen—cantaining compounds react with UCl; to precipitate uranium o des and oxy- chlorides. Oxygen can enter the salt through air, water vapor cr transmutation of F. The vagaricus nature of air and water vapor entry necessitates keeping the oxygen content well below saturation. A continuous gas bubbling system (with chemical reducing agent) should help: so should the appreciable capacity of the salt for OXygen. Experiment indicates difficulty with simple methods of salt cleanup such as the small change in sclubility by temperature adjustment.. An alternative effective methed routes the bypass gas flow through a bed, vhere the gas mixes continucusly with injected liquid NaAlClh. Greater stapility of A1203 causes it to form over AlCl3 (Fig. 3.7-3). The .solid alumina then separates out by filtraticn or cyclaone. 341 -!.00-1 | w o o i cHiories [KT-mol™' ] & T 1000°% form AG ~100- —100 L —200 ~3d0 , " 1 - AG_ OXIDES [KI-Jamel 0] Figure 3.7-3 Competition Between Formatimn of Cxides and Chlorides at 1000 K. 342 3e7e2sH Removal of sulfur impurities, Several nuclear reactions convert chlorine to sulfur (Figure 3.7;4). Mass balance and Coulonmb barriers combine to inhibit most of these reactions; but high flux levels bolster production, 3501 converts to 325, 335, and 365. 3731 should produce only small amounts of 348 (favoring 2 3701 enrichment 2gain), A natural chlorine-fuelled system will produce more sulfur than the salt solubility can handle, Phosghorus, though present is only transitory, decaying quicily to suliur, The presence of sulfur, or of any other element capable of compounding with uranium, need not adversely affect the feasibil- ity of the system: simple zdjustment of temperature and UCl]+ | content can induce precipitztion in a clean-up circuii. Because one can reliably predict the production rate of these mutants, the concentrations in the core could be safely maintained close to saturation: then even an inefficient removal process would suffice. The effect of the sulfur presence will, in part, depend on its oxidation state. For molten sulfur, the oxidation-state equi- librium is fairly well fixed; the predominant state tends to be‘ positive (Fig. 3.7-5). The "positivity" also increases with irrediation (Fig. 3.7-6). 3,7.2.6 Maintaining the chlcrine stoichiometry. Zach actinide atom initially binds three or four CL atoms. Although fission splits each actinide atom into two product atcms, the net valency reduces (in part due to inert gases)} and an excess of chlorine occurs. This can lead to several corrosicn agents, especially UC14.- (The 343 Reacticn Reaction Q-value (MeV) 1. %1 (n,p) BET (%) 35 - 0.615, 9.89 7,28 35Cl 2. 2%1 (n,x) 3% ll*-jf-_d-». 325 | 0.935 3. =6 36 L 3533_ (2,Y) 2°c1 (n,p) 778 8458,1.92 (n, ) 2p —2-,2-,2&-335 2.46 (Y1245 3Ly 6,28 5, 3761 (n,p) s 219'-“_;—_-'1‘-;-3701 -3.98 / 6. Tgi (n,o0) Shp L2:bs | 3bg =1.29 5 Figure 3.7-4 Nuclear Reactions Which Produce Sulfur in an MCFR(Th) intense fissian fragment irradiation of the salt produces short- lived ions which quickiy oxidize UCl3 to UClh.) To hold down the concentration of UClh, and excess Cl in general, one can react the fuel salt at a modest rate with metal of naturel uranium or thorium or with other reducinz agents. 3eTe2:s7 Storing troublesome fission products by usinz them as carrier salis. Section 2.,1,2 points out that, of all the fission 90 137 products, “ Sr and Cs should present the most trouble, However, Sr and Cs belong to the alkaline earth and alkali classes of ele- ments examined in section 3.lede Thus a good way to manzge these 344 100 - tulphur species (¥s) ¢ T =T Y 0 00 00 trradiation temperature (°C) Fig. 3.7=-5 Irradiaticn-tempe-ature dependence of the oxidaticn- state distribution of 275 species [148] Fig. 3.7-6 Effect of length of irradiaticn time on the --S- species distributian, 150°C [148] 345 long=lived salts might simply be to-use them as carrier salts SrCl2 and CsCl. Unlike natural Sr and Cs, they would not increzse in radiological hazard, but rather decrease, This would also apply to some of the other fission products, especially RbCl znd BaClZ. 2.7.2.8 Comparison to MSSRX reprocessing. In FSBR reprocassing cf the fuel salt (Fig. 3.7-7) a volatility prcecess (3ection 3.7.1.3) first extracts uranium and immediately returns it to the primary circuit. Subsequent steps prceceed more slowly and sometimes tor— 233 tuously. They especially include Pa extraction by a ligquid bismuth contact procsss and -are earth extraction. After 233Pa decays to 233U, it returns t> the reactcr. }SBrs must carefully hendle tritium gas produced by 6Li (n, ® ) reacticns: releases to the environment must be strictly centrolled. Reprocessing for an HCFR(Th) should be much simpler: enhanced 233Pa and 23&U fissionability makes it unnecessary to separate and hold up 233Pa. Also an MCFR produces magnitudes—less tritium, 3.7.2.9 liaterials reouirements. The processing plant may require special materiais. The transfer lines will probably be molybdenum tubing; scme of the. largze vessels may be graphite. For an HSBR a frozen layer of sali protects the wall of the fluorinator from corrosion, 346 Zower Razetor -3233 AWz uczive txtracti Proces3ing cine Separa - e I tidn 9t by s 84520 = I 3=z iT. Fa frradiated Fuel ELflozaney Fmoval Fluorinatior itnout Jpgree—= Therma = e u-233 for sale Figure 3.7-7. SBR Reprocessing Scheme 347 Ex::;- ian Fl3t Lan 4 _— . 3.7.3 Blanket Salt Processing Blanket salt processing must extract UClh and convert it to UClB. Some degree of separation may be initially accomplish- ed by slowly cooling the salt mix to precipitate out ThC'Ll.l+ until a eutectic composition is reached (Fig. 3.6-1). A low need for high purity plus a desire for compactness favor pyroprocessing with chlorides. Should the blanket salt be a fluoride, oxidation to UFy and subsequent distillation (section 3.7.3.8) appeals. UClh reduces to UCl3 reilatively easy using HZ' alkali (Na,K,etc.), U, or The For metallic reduction, rod, gauze or turnings are popular. Using alkali (ik = Na or K), the chemical reaction is UClh + Ak ——i U013 + AkCL This results in a 50/50 molar mix; lower in uranium content than desired. Any AkCl initially present would even further reduce the UCl3 molar content. This leaves culy Th or H2 as alternatives. Th metal is toxic and flammable. H, produces HCl, which in turn might be used to chlorinate the thorium oxide feed (section 3.7+7)s A possible direct route to reducing UClk.may;be by electrolyéis, releasing 012 which could, in turn, chlorinate ThO 2 feed. 343 Blanket salt should well accord batch reprocessing. The frequency of blanket reprocessing will depend on 1. The tolerable 233U coancentration in the blanket which de~ pends on da« Allowable power production in blanket which depends on (1) Maximum desirable heat removal capability from blanket salt (2) Neutron damage to reactar vessel b. The amount of 233U inver:tory to be tolerated in the blanket, which depends on (1) Proliferation hazard (2) Financial worth of