DRNL/TM-7207 Conceptual Design Characteristics of 2 Denatured Molten-Salt Heactor with Once-Through Fueling J. B, Engel V. R. Grimes H. F. Bauman . H. McCoy J. E. Dearing W. A, Bhoades LUTION OF THIS BOCUMENT 18 UNLIYTER BISTRIE Printed in the United Stataes of America. Available from [\afl@fififl Technical Information Service .8, Depariment of Commerce 5285 Port Roval Road, Springfiel o’ Virginia 22161 PoF NTIS price codes—FPrinted Copy: ADR Microtiche ADT This report was prepared as an account of work sponsored by an agenay of the Linited States Government, Neither the United States Government nar any agency theraof, nar any of their emplovess, makes any warranty, express or imiplied, or assumes any legal liability or responsibility for the acocuracy, carw:ttenmés or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Referance herein to any specific commercial product, DFGufilbé;QV%efl’!Cét} ttrade name, frademark, manufacturer, or otherwise dces notl necessarly constitute or amply iis endorsement, recommendation, or favoring by the United Siates Governiment or any agency thereof. The views and opinions of authars expressed haerein 4o aot necessarity state or reflect those of the United States Governmeant or any agenay thieraof. ] ~ ORNL/TM~7207 Dist. Category UC-76 Contract No. W-7405-eng-26 Engineering Technology Division CONCEPTUAL DESIGN CHARACTERISTICS OF A DENATURED MOLTEN-SALT REACTOR WITH ONCE-THROUGH FUELING J. R. Engel W. Re Grimes H. F. Bauman H. E. McCoy J. F. Dearing W. A. Rhoades Date Published: July 1980 NOTICE This document contains information of a prefiminary nature. It is subject to revision or correction and therefore does not represent a final report. Prepared by the OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee 37830 operated by UNION CARBIDE CORPORATION for the DEPARTMENT OF ENERGY & iit i CONTE NT S Page ABSTRACT .OCQQ.Q.t..0000.00GOOIOODGOOOO0.0CO..OGOOOQ.O..‘0OOSBGDQOB 1 1. INTRODUCTION .OICGOU.00..000..0.005000.0.0.QQ.OOOO'0.0CGGOOQ00‘ 2 2! GENERA.L DESCRIPTION OF DMSR 000DUGDODODQOQOOOCQOGEQOOOOODOOOOOO 5 2.]— Fuel Circuit Q.0.0.0EO‘S&QOCGO..08‘0OOQOQDOOGOOOOOOOGUGGGO 6 202 CQO].ant Circuit QOOGOE.IOOOCOU.QQOG99.0000fie!".&@.o."“flfl. 7 293 BalanCE“Of‘"Plafit 6§ 00288 @0 0E85680008EEEPPBEOEE0O660QRELERES EHSC 0 7 2.4 Fuel Handling and Processing eseececocsscesscsescsosscoscne 3 3. REFERENCE-CONCEPT DMSR ..IQQOQ.OGQO...OOC!OQOO9.....3000&00000. lO 3!1 Neutronic Properties .00.0.'.3.0000&0.0'fifiOOGQOGO....GOOIG 10 3.1.1 Neutronics core model .cccccceeccscessssccnossscnasss 10 3.,1.2 Core design considerations ecescesevscoccossssossse 14 3.,1.3 Neutronics calculation approach .cscscesscooccoccce 14 3.1.4 Once-through system considerations 50 keV). This reactor could be made critical with about 3450 kg of 207 enriched 235y and op~ erated for 30 years with routine additions of denatured 235y and no chemical processing for removal of fission products. The lifetime requirement of natural U30g for this once-through fuel cycle would be about 1810 Mg (~2000 short tons) for a 1-GWe plant operated at a 75% capacity factor. If the uranium in the fuel at the end of life were recovered (3160 kg fissile uranium at ~10%7 enrichment), the U30g requirement could be further re- duced by nearly a factor of 2. The lifetime net plutenium pro- duction for this fuel cycle would be only 736 kg for all iso- topes {238, 239, 240, 241, and 242), A review of the chemical considerations associated with the conceptual fuel cycle indicates that no substantial difficulties would be expected if the soluble fission products and higher ac- tinides were allowed to remain in the fuel salt for the life of the plant. Some salt treatment to counteract oxide contamina- tion and to maintain the oxidation potential of the melt prob- ably would be necessary, but these would require only well=-known and demonstrated technology. Although substantial technology development would be re- quired, the denatured moltem-salt reactor concept apparently could be made commercial in about 30 years; if the costs of in- termediate developmental reactors are included, the cost for development is estimated to be $3750 million (1978 dollars). The resulting system would be approximately economically com— petitive with current—-technology light-water reactor systems. 1. INTRODUCTION g Molten-salt reactors! (MSRs) have been under study and development in the United States since about 1947, with most of the work since 1956 . directed toward high—-performance breeders for power production in the Th-233y fuel cycle. The most recent development effort in this area was 2 provided by the terminated in September 1976 in response to guidance Energy Research and Development Administration {(now Department of Energy) (ERDA/DOE) in March 1976. A brief study of alternative MSRs® which em- phasized their antiproliferation attributes was carried out in late 1976. This study concluded that MSRs without denatured fuel probably would not be sufficiently proliferation-resistant for unrestricted worldwide deploy- ment. Subsequently, a more extensive study was undertaken at Oak Ridge National Laboratory (ORNL) to identify and characterize denatured molten— salt reactor (DMSR)} concepts for possible application in antiproliferation situations. This work began as part of the effort initiated by ERDA in response to a nuclear policy statement by President Ford on October 28, 1976;4 it was continued under the Nonproliferation Alternative Systems Assessment Program (NASAP),> which was established in response to the Nuclear Power Policy Statement by President Carter on April 7, 1977,6 and The National Energy Plan.’ - The DMSR is only one of a large number of reactors and assocciated fuel cycles selected for study under NASAP. However, it is also a member of a smaller subgroup that would operate primarily on the Th~233U fuel cycle., Molten—salt reactors, in general, are particularly well suited to this fuel cycle because the fluid fuel and the associated core design tend to enhance neutron economy, which is particularly important for effective resource utilization. In addition, the ability of the molten fuel to re- tain plutonium {produced from neutron captures in the 238y denaturant) in a relatively inaccessible form appears to contribute to the proliferation resistance of the system. The MSR concept also offers the possibility of system operation within a sealed containment from which no fissile mate- rial is removed and to which only denatured fuel or fertile material is added during the life of the plant. This combination of properties sug- gests the possibility of a fuel cycle with a low overall cost and signifi- cant resistance to proliferation. S The primary purpose of this study was to identify and characterize one or more DMSR concepts with antiproliferation attributes at least equivalent to those of a “"conventional” light-water reactor (LWR) oper- i ating on a once-through fuel cycle. The systems were also required to show an improvement over the LWR in terms of fissile and fertile resource utilization. Considerable effort was devoted to characterizing features of the concept(s) that would be expected to affect the assessment of their basic technological{feasibility. These features included the estimated costs and time schedule for developing and deploying the reactors and their anticipated safety and environmental features. Although the older MSR studies were directed toward a high-perfor- mance breeder [and a reference molten-salt breeder reactor (MSBR) design8 was developed], the basic concept is adaptable to a broad range of fuel cycles. Aside from the breeder, these fuel cycles range from a plutonium burner for 233y production, through a DMSR with break—-even breeding and complex on-site fission-product processing,9 to a denatured system with a 30-year fuel cycle that is once-through with respect to fission-product cleanup and fissile-material recycle. Of these, the last one currently appears to offer the most advantages for development as a proliferation- resistant power source. Consequently, this report is concentrated on a conceptual DMSR with a 30-year fuel cycle and no special chemical pro- cessing for fission-product removal; other alternatives are considered only briefly. Section 2 contains a general description of the DMSR concept, with emphasis on those features that would be the same for all DMSR fuel cy- cles. Section 3 presents a more detailed treatment of the reference- concept DMSR covering the neutronic and thermal-hydraulic characteristics of the reactor core, fuel-salt chemistry, reactor materials, plant safety considerations, and system~specific environmental comnsideratiomns. A gen- eral treatment of the antiproliferation attributes of the concept is also included. The next section (Sect. 4) addresses potential alternatives to the reference concept and their perceived advantages and disadvantages. Section 5 addresses the commercialization considerations for DMSRs, in- cluding the perceived status, needs, and potential research, development, ] and demonstration (RD&D) program; a possible schedule for major con- struction projects; the estimated performance of commercial units; and any special licensing considerations. Finally, Sect. 6 presents the gen- eral conclusions of the study, alcong with suggestions that weould affect @ any further work on this concept. S | 2. GENERAL DESCRIPTION OF DMSR The plant concept for a DMSR is a direct outgrowth of the ORNL reference-design MSBR, and, therefore, it contains many favorable fea- tures of the breeder design. However, to comply with the antiprolifera- tion goals, it also contains a number of differences, principally in the reactor core design and the fuel cycle. Figure 1 is a simplified sche- matic diagram of the reference~design MSBR. At this level of detail, there is only one difference from the DMSR concept: the on~line chemical processing plant {shown at the left of the core) would not be required for the DMSR, ORNL-DWG 68—-1185ER ™ : SECONDARY NaBF, - NaF | (SALT PUMP COOLANT SALT ‘ PRIMARY ? SALT PUMP — /’fi ‘ PURIFIED SALY P GRAPHITE MODERATOR REACTOR HEAT | EXCHANGER |{{RUTE - 566°C CHEMICAL PROCESSING PLANT TLiF - BeF, - ThF, - UF, FUEL SALT | STEAM GENERATOR Tuéééék'LL;__ © GENERATOR \———/ STEAM Fig; 1. Single-fluid, two-region molten salt breeder reactor. 2& I. Fuel CirCUit ,\_V‘ The fuel circuit for a DMSR would be very similar to that for an MSBR; only the core design would be changed. The primary requirement for this redesign was a reduction in the core neutron flux (and power density) to 1. extend the life expectancy of the graphite moderator to the full 30- year plant lifetime, 2., limit neutron captures in %233Pa which, to enhance proliferation re- sistance, would be retained in the fuel szalt. The lower power density would also tend to reduce the poisoning effects of short—-lived fission preoducts and to simplify the thermal-hydraulic con- straints on the design of the moderator elements. The principal unfavor- able effects would be the increases in inventory of the fuel salt and fis- sile fuel. Reference-design features of the DMSR core are described in greater detail in a later section. . At design power (1000 MWe}, the fuel salt, which would have a liqui- dus temperature* of about 500°C, would enter the core at 566°C and leave at 704°C to tranmsport about 2250 MWt (in four parallel locops) to the sec~ . ondary salt. The flow rate of salt in each of the primary loops {includ- ing the bypass for xenon stripping) would be about 1 m3/s (16,000 gpm). The primary salt weuld contain 0.5 to 1% (by volume) helium bubbles to serve as a stripping agent for xenon and other volatile fission products. Helium would be added to and removed from bypass flows of ~10%Z of each of the primary loop flows. This gas stripping would alsc remove some of the tritium from the primary salt,T partly as 3HF; however, most of the tritium would diffuse through the tube walls of the primary heat exchang- ers into the secondéry salt. Helium removed from the primary circuit would be treated in a series of fission-product trapping and cleanup steps before being recycled for further gas stripping. Provisions would also be *The temperature at which the first crystals appear on equilibrium cooling. YEstimates are that 18 to 19% of the total tritium produced would be removed in this gas. made in the primary circuit to remove and return fuel salt without opening the primary containment and to add fuel=-salt constituents as required to maintain the chemical condition of the salt. 2.2 Coolant Circuit The secondary, or coolant-salt, circuits for the DMSR would be iden- tical to those developed for the reference-design MSBR. The nominal flow rate of the secondary salt (a eutectic mixture of NaBF, and NaF) would be about 1.26 m3/s (20,000 gpm) in each of the four loops, with a temperature rise from 4534 to 621°C in the primary heat exchangers. This salt would be used to generate supercritical steam at about 540°C and 25 MPa to drive the turbine-generator system.” In addition to its primary functions of isolating the highly radio- active primary circuit from the steam system and serving as an interme- diate heat~transfer fluid, the sodium fluoroborate salt mixture would play a major role in limiting the release of tritium from the DMSR sys- tem. Engineering-scale tests in 1976 (Ref. 10) demonstrated that this salt is capable of trapping large quantities of tritium and transforming it to a less mobile, but still volatiie, chemical form that transfers to the cover-gas system rather than diffusing through the steam generators to the water system. Consequently, the majority of the tritium (~807%) would be trapped.or condensed out of the secondary circuit cover gas, and less than 0.2%Z of the total would be released. 2.3 Balance-of-Plant The balance-of-plant for a DMSR primarily would be identical to that for an MSBR. Because the samé salts and basic parameter values are in- volved, there would be no basis for changing the normal auxiliary systems required for normal plant operation. Differences, however, could appear in some of the safety systems. Because of the lower power density in the *The supercritical steam cycle appears to be particularly well suited to this concept because of the relatively high melting temperature (385°C) of the secondary salt and the desire to avoid salt freezing in the steam generators. DMSR, the shutdown residual-heat-removal (RHR)} problem would be less se- vere than in the MSBR. Consequently, a less elaborate RHR system than would be needed for an MSBR might be acceptable for a DMSR. However, for purposes of characterizing the DMSR, the assumption was that the balance- of-plant would be the same as that for an MSBR. 2.4 Fuel Handling and Processing The performance of an MSBR would be strongly dependent on the avail- ability of an on-site continucus chemical-processing facility for removal of fission products and isolation of protactinium on relatively short time cycles. These treatments would make possible the achievement of a posi- tive 233y breeding gain in a system with a low specific fissile inven- tory. Because a DMSR on a 30-year fuel cycle would not require even nomi- nal break-even breeding and because a significantly higher fissile inven-— tory could be tolerated, the processing requirements for a DMSR would be much less stringent than for an MSBR. Isolation of protactinium would be avoided for proliferation reasons, and chemical processing to remove fis- sion products could be avoided without severe performance penalties. Despite these concessions, some fission—-product removal would take place in any MSR. Most of the rare gases (and some other volatile fis- sion products) would be removed by the gas—sparging system in the primary circuit. In addition, a substantial fraction of the noble~metal® fis- sion products would be expected to plate out on metal surfaces where they would not affect the neutronic performance. However, the reference-design reductive—extraction/metal-transfer process would not be involved. Although there would be no chemical processing for fissionm—product removal, the DMSR likely would require a hydrofluorination system for occasional (presumably batchwise) treatment of the salt to remove oxygen contamination. In addition, because a DMSR would require routine addi- tions of fissile fuel, as well as additions of other materials necessary to keep the fuel-salt chemical composition in proper balance, a chemical *Nobility is defined here in relation to the U*Y/U3T redox potential (see Sect. 3.3.2). addition station would be required. The technology for both of these S operations is well established and was extensively demonstrated in the molten—-salt reactor experiment (MSRE). These and other aspects of the DMSR fuel chemistry are treated in greater detail in a later section. 10 3. REFERENCE-CONCEPT DMSR A preliminary conceptual design has been developed for a DMSR oper- ating on a 30-year fuel cycle. The emphases tc date have been on the re- actor core design and fuel cycle, with less attention to other aspects of the system. Although this design establishes the basic concept and char- acterizes its major properties, it is tentative and wculd be subject to ma jor refinement and revision if a substantizl design effort were under- taken. 3.1 Neutronic Properties The basic features of this DMSR concept which distinguish it from other MSRs are established primarily by the reactor core design and its associated neutronic properties. The design described here represents the results of a first-round effort to balance some of the many variables involved in a reactor core, but it is by no means an optimized design. 3.1.1 Neutronics core model From a neutronics point of view, the core is simply designed as fol- lows (Fig. 2}. 1. The core and reflector fill a right circular cylinder that is 10 m in diemeter and 10 m high. The core, which is a cylinder 8.3 m in dizmeter and 8.3 m high and centered inside the larger volume, is filled with cylindrical graphite logs in a triangular array of G.254-m pitch. Approximately 957 of the core (core B) has log diameter of 0.254 m, with the fluid fuel filling the interstitial volume to produce a fuel volume fraction of 9.31%Z. An axial cylindrical hole of 0.051-m diam in the cen- ter of each log admits another 3.63% fuel for a total of 12.94 vol %Z. To achieve flattening of the fast flux and thus maximize the lifetime of the graphite moderator, the remaining 5% of the core (core A), a cylinder 3 m in diameter and 3 m high, has a log diameter of 0.24 m, resulting in a total fuel volume fraction of 20.00% in this zone. 11 s ORNL-DWG 80—-4263 ETD t [ S F 0-f5 ! TOP REFLECTOR - 0.20 TOP PLENUM 3 ? t o O - RADIAL GAP 3 . LL LLt s g . CORE B :é & o CORE A & 81- _ ___MIDPLANE _ 18 S o w m m — I X | l Y & i 0.20 I _BOITTOM PLENUM____#i 0.?5 ! | BOTTOM REFLECTOR l } -——71.50 | 0.05—> -t 4.15 | [-a-(}) 80> - 5.00 - Fig. 2. DMSR core model for neutronic studies — cylindrical geometry (all dimensions in meters). 12 2. The radial reflector is graphite 0.8 m thick and is attached to S the reactor vessel at the 10-m diam. This leaves a gap of 0.05 m filled with fuel salt surrounding the core laterally. 3. The inlet and outlet plena cover both the core and radial gap to their full diameter and are each 0.20 m thick. They consist of 507% struc- tural graphite and 50% fuel. 4. The axial reflectors are each 0.65 m thick and extend to the full 10-m diam. 5, All reflector regions contain a small amount of fuel salt for cooling, which is estimated as 1 vol % at operating temperature. 6. All stated dimensions are assumed to apply at nominal operating conditions, During system heatup, the length and diameter of the core vessel are assumed to increase at the rate of expansion of Hastelloy-N. The reflectors are assumed to expand at the expansion rate of graphite but to remain attached to the vessel. Because graphite expansion is less than that of the vessel, this will result in admitting additional salt to the reflector zones. The core and plenum regions are assumed to expand radially only at the expansion rate of graphite, which will establish the thickness of the radial gap. The axial configuration is affected by the logs floating upward in the salt and by the lower plenum being constructed so that it always contains 507 salt. The thicknesses of the core and the upper plenum, then, increase at the graphite expansion rate, but the lower plenum grows at such a rate as to span the gap between the core and the bottom reflector. Mechanical properties used for the principal constituents are sum- marized in Table 1. The salt is taken to have the nominal chemical com— position shown in Table 2. The term "actinides” in this study refers to all elements of atomic numbers > 90 and not just to transplutonium ele- ments. The actinide percentage is subject to small variations depending on the fuel cycle and the history of the fuel. The ifiventory of fuel salt, both in and out of the core, is summa- rized in Table 3. This is believed to be a genercus estimate of the re- quired inventory for a 1-GWe system. The thermal energy yield per fission is assumed to be 190 MeV for translation of absolute fission rates to ef- fective power level. 13 i Table 1. Reference properties of fuel salt and moderator for a DMSR Characteristic | Value Craphite moderator density, Mg/m3 1.84 Fuel-salt density, Mg/m 3.10 Graphite linear thermal expansion, X 10-6 gl 4.1 Vessel linear thermal expansion, x 107 g1 17.1 Fuel volumetric thermal expansion, X 1076 g1 200 Table 2. Nominal chemical composition of DMSR fuel salt Material Molar percentage 7LiF 74,0 XF,” 9.5 Fission products Trace aX refers to all actinides, Table 3., DMSR fuel-salt inventory Location Volume (m3) Core 59.4 Top and bottom plenums 11.1 Radial gap 10.9 Reflectors 3.0 External loop - 20.0 R 14 3.1.2 Core design considerations The size of the core was determined so as to allow a graphite mod- erator lifetime equal to the design lifetime of the plant. As compared with a smaller core, this resulted in lower neutron leakage, higher inven— tory of fissile material, and lower loss of protactinium due to neutron capture. If higher levels of graphite exposure were indicated by future data or decisions, a smaller core would probably be chosen. The circular cylinder moderator shape‘resists binding effects that can cccur with other shapes. The hole in the center is sized to provide desirable resonance self-shielding without undue thermal flux depression. The lattice pitch is simply a convenient one from both thermal and neu- tronic points of view. The reduced diameter of the central section of the logs was adjusted to give the proper degree of neutron flux flattening. There is no doubt that flux flattening results in more core leakage, slightly degrafled breeding, and more flux in the reactor vessel as com— pared with an unflattened core. The unflattened core, however, woculd have a much larger volume and much larger inventory of fissile material for the same maximum neutron damage flux. 7 The thorium concentration of the salt has been adjusted to give near-— optimum long-term conversion and a low requirement for makeup fuel. This appreach leads to a relatively high in-plant fissile inventory, which may have economic disadvantages. Thus, overall optimization might suggest more favorable combinations of inventory and makeup. The other actinide cencentrations are determined by the varicus fueling policies considered and by the operating history of the fuel. 3.1.3 Neutronics calculation apprecach 3.1.3.1 Overall strategy The overall approach was designed to couple numerous computer runs -0of relatively short duration. The objectives were good accuracy, rela- tively quick computer response, and the ability to repeat and revise dif- ferent portions of the procedure as the design evolved. 15 Initial scoping studies showed that the self-shielding of thorium and 238y has a most critical effect on the system neutroenics, while that of the other uranium nuclides was comparatively less. Concentrations of protactinium, neptunium, and plutonium remained small enough to make self- shielding treatment of those nuclides necessary. The effect of resonance overlap between 2327y and 238y was of particular interest and was studied in some depth using the ROLAIDS mddule of the AMPX code system.l1 The conclusions were that this effect could be ignored safely in the present study and that treatment of the effect would have been burdensome had it been required. Statics. A set of cross sections for the more significant nuclides (Table 4) was prepared based on the ENDF/B Version 4 set of standard cross 12 A total of 123 energy groups was used, with boundaries as sections. listed in Table 5. Downscatter from any group to any other was allowed, and fipscatter between all groups below 1.86 eV was allowed. The 123-group set was then reprocessed to enforce strict neutron conservation. This was especially important in the case of graphite. Table 4. Nuclides in library of 123 energy groups used for DMSR study 2327y 238y F 233p, 239p, 714 233y 240p, Be 234y 24lpy 6L 235y 2u2p, 10g 236y Graphite 238py Self-shielding of thorium and uranium nuclides was treated using the NITAWL module of the AMPX code system. The Nordheim integral treatment was selected in each case. The geometric parameter applicable to the tri- cusp fuel area between the logs was determined by a special Monte Carlo computer code devised by J. R. Knight of ORNL.!3 Figure 3 illustrates 16 Table 5. XSDRN 123-group energy structure Boundaries Boundaries Boundaries Group Group Group Energya Lethargy Energy Lethargy Energy Lethargy 1 1.,4918E07 0. 40 43 2.2371E05 3.80 84 2.2603E01 13.00 2 1.3499EC7 —0.30 44 2.0242E05 3.90C 85 1.7603E01 13.25 3 1.2214807 =0, 20 45 1.8316E05 4,00 86 1.3710E01 £3.50 4 1.1052E07 -0.10 46 1.6573E05 4.10 87 1.0670EC!L 13.75 5 1.000Q0EO7 0.0 47 1.4996E05 4,20 88 8.3153E-01 16.3C 6 9.0948E06 0.10 48 1.3569E05 4,30 a9 6.4760E=01 16.55 7 8. 1873E06 0.20 49 1.2277E05 4.40 S0 5.0435E-01 16,80 8 7.4082E06 0.30 50 1.1109E05 4.50 g1 3.9279E~01 i7.50 g 6.7032806 0.40 51 8.6517E04 4,75 92 3.0590E-01 17.30 16 6.0653E06 0.50 52 6.7379E04 5.00 93 2.38248-01 17.55 11 5.4881E06 0.60 53 5.2475E04 5.25 94 1.8554E~-01 17.80 12 & ,9659EC6 0.70 54 4,.0868E04 5.50 25 1.7090E-01 17.88 13 4,4933E06 0.80 55 3. 1828E04 5.75 96 1.5670E-01 17.97 14 4.0657E06 0.90 56 2.4788E04 6.00 a7 1.4320E-01 18.06 15 3.6788E06 1.00 57 1.9305E04 6.25 98 1.2850E-01 18.17 16 3.3287E06 1.10 58 1.5034E04 6.50 99 1. 1340E-01 18,29 17 3,0119806 1.20 59 1.1709E04 6.75 100 3.9920E-02 18.42 18 2,7253E06 1.30 60 9.1188EC3 7.00 101 8.8100E-02 18.55 19 2.4660E06 i.40 61 7.1017E03 7.25 102 7.6840E-02 18.68 20 2.2313EC6 1.50 62 5.5308E03 7.50 103 6.5520E~02 18.84 21 2,0180E06 1.60 63 4,3074E03 7.75 104 5.4880E-C2 19.02 22 1.8268E06 1.70 64 3.3546E03 8.00 105 4,4850E~-02 19,22 23 1.6530E06 1.80 65 2.6126E03 8.25 1086 3,6140E-02 19,44 24 1.4957E06 1.90 66 2.0347E03 8.50 107 2.9940E-02 19.63 = 25 1.3534E06 2.00 o7 1.5846E03 8.75 108 2.4930E-02 19.81 26 1.2246E06 2.10 68 1.2341E03 9.00 109 2.0710E~02 20.00 27 1. 1080E06 2.20 69 9.6112E02 9.25 110 1.7980E-02 20.14 28 1.0026E06 2,30 70 7.4852E02 9,50 111~ 1.5980E-02 20.25 29 9.0718EQGS 2,40 71 5.8295E02 9.75 112 1.3980E-02 20.39 30 - 8.2085EC5 2.50 72 4,5400E02 10.00 113 1. 1980E-02 20.54 33 7.4274E05 2.60 73 3.5357E02 10.25 114 9.9700E-03 20.73 32 6.7206EQS 2,70 74 2.7536E02 16.50 115 8. 2300E-03 20,92 33 6.0810E05 2.80 75 2.1145E02 10.75 116 6.93900E-03 21.08 34 5.5023E05 2.90 76 1.6702E02 11,00 117 5.99%00E-C3 21.24 35 4,9787E05 3.00 77 1.3007E02 11.25 118 4,.9900E-03 21.42 36 4,5049805 3.10 78 1.0130E02 11.50 I1¢ 3,9800E-03 21.64 37 4.0762805 3.20 79 7.8893E01 11.75 120 2.9800E~-03 21.93 38 3.6883E05 3.30 80 6.1442E01 12,00 121 2.1100E-03 22,28 39 3.3373ECS 3,40 81 4.7851E01 12.25 122 1.49C0E-03 22.63 40 3.0197EGS 3.50 82 3.7267E0] 12,50 123 9.8000E-04 23.05 41 2.7324E05 3.60 83 2.9023E01 12.75 124 4,7000E-04 23.78 42 2.4724E05 3.70 Crxx corresponds to 10%X, Lower boundary of group 123. 17 : CRNL-DWG 804264 ETD 0.75F 20/PITCH o g1 S I 0.25 0.80 0.85 0.90 0.95 1.00 ROD DIAMETER/PITCH RATIO Fig. 3. Mean chord length £ of fuel surrounding triangular arrays of moderator rods. Circles illustrate predictions by the 4V/A rule. the results of this treatment., The salt in the plenum and radial gap re- gions was represented as a 0.05-m plane envifonment. The resulting multigroup cross sections were used with the XSDRNPM module of AMPX to accomplish a discrete—ordinates cell calculation in the S-4 approximation and to accomplish group reduction to three energy groups, as shown in Table 6. A separate cell calculation was performed for each of the two log diameters. Plenum and gap cross sections were weighted over the spectrum of the smaller log diameter bécause it lies between the standard diameter and the pure salt region in hardness. The basic concentrations of the nuclides were based on estimated midlife conditions. Additional cases of self-shielding for the thorium 18 Table 6. Few—group energy structure - for DMSR neutronic studies Energy group Energy range Fast 14,918 MeV to 52.475 eV Resonarnce 52,475 eV to 2.3824 eV Thermal 2.3824 eV to 0.00047 ev and uranium nuclides were prepared for use in the depletion and reactiv- ity coefficient studies. These were weighted over the neutron spectra calculated in the cell calculation. The macrospatial effects were treated using the reduced cross sec=- 4 tion set with the APC Il computer code,! Separate axial and radial flux proflies were found with mutually consistent flux and leakage results. Core heterogeneity was treated by transverse flux weighting of the de- tailed geometry. Reaction parameters necessary for burnup were deter- mined from these results, with care taken to combine all reactions rep—- resenting a particular nuclear species regardless of positions in the cell or the identity of the cell involved. This is consistent with an as- sfimption of rapid fuel circulation and mixing. Burnup. A simple burnup ccde, QUAB, was devised to treat the un- usual requirements of this study. Special features include the follow- ing. 1. Sufficient 238y is added at all times to maintain the denatured con- dition. 2., The thorium concentration can be held constant by automatic addition, allowed to decline naturally, or adjusted to maintain constant total actinide concentration. 3. Periodic additions of enriched fissile material can be made. 4. Periodic withdrawals of fuel can be made selectively by nuclide. This fuel can be held until the protactinium decays and then be reinserted selectively by nuclide into the machine. The first removal is re- placed with fuel identical to the initial loading. 5. Enriched material can be added on demand to maintain a specified re- activity margin. 19 S The code calculates nuclide concentrations, total inventories, reactivity, and breeding ratio as a function of time. Treating the lengthy transplutonium and fission-product chains in * QUAB was not practical; multigroup data were not available for many of the required nuclides and were of dubious reliability for others. In- le es— stead, the ORIGEN codel® was used with a library of cross sections pecially devised for its use. The ORIGEN results were then "patched into” the QUAR calculation directly. The burnup calculation allowed the cross sections of thorium and 238y to vary continuously during the calculation; this was accomplished by in- terpolation. 3.1.3.2 Evaluation As desired, the method provided relatively rapid response, detailed treatment of resonances, and a multigroup spectrum and cell treatment. All details of the denatured fuel cycle were treated. The expedient of treating a range of thorium and 2387 densities removed the necessity of imbedding the expensive and tedious resonance treatment inside the loop for varying densities. Deciding on the applicable range was not difficult after a few initial tries. A system coupling the spatial calculation and depletion could be used. Many such systems are available, although all would require exten-— sive modification for MSR use. What of the cell calculation? Table 7 shows the cell faétors from our reference case which have been condensed to three energy groups. This is clearly a heterogeneous core. Further, the actinide densities are continually changing, resulting in time- dependent cell factors. Studies beyond these would be required to prove that a coupled system could be worthwhile without directly coupled cell calculations. The requirement to "ummix” the revised nuclear densities after hav- ing them lumped together during a depletion step represents a complica- tion that would thwart most existing codes. However, this complication must be coupled with logic to provide interpolation between cross—section sets representing various self-shielding situations. With or without an 20 Table 7. Unit cell flux ratios® for reference DMSR Cell material Energy Core group zone Inner Moderator Interstitial salt salt Fast A 1.14 0.96 1.14 Resonance A 0.97 1.01 0.97 Thermal A 0.94 1.03 0.88 Fast B 1.28 0,98 1.12 Resonance B 0.97 1.00 C.98 Thermal B 0.88 1.01 0.93 aAverage flux in material divided by average cell flux. imbedded cell calculation, an unusual code system clearly would be neces- sary to provide a fully satisfying level of detail to this problem. Ob— viously, a true two-dimensional spatial treatment of the flattened core would be appropriate, but imbedding such a calculation inside a depletion loop is expensive. 3.1.4 Once-through system considerations 3.1.4.1 Fueling policy For purpcses of nuclear calculations, the fueling policy for the once—-through DMSR is as fellows. 1. Thorium is added to an initial loading of salt in a specified concen- tration. During operation, the concentration is allowed to decline via burnup. Near the end of plant life, small amounts are removed as required to keep the total actinide content below the startup value. 2. Uranium is added at the maximum allowable enrichment in the amount necessary to maintain criticality. 21 e 3. Additional 2387 is added as required to maintain the denaturing inequality* ° density 2387 > (6 x density 233U) + (4 x density 23%U) ., 4. Removal of certain fission products is accomplished according to Table 8. Table 8. Removal times for fission products in once-through cycles Fission-product Blement Re?oval group time Noble gases Kr, Xe 50 s Seminobie and Zn, Ga, Ge, As, Nb, 2.4 h noble metals Mo, Te, Ru, Rh, Pd, » Ag, Cd, In, Sn, Sb 3.1.4.2 Fission-product buildup A study of 30-year fission-product buildup was made as a function of various continuous removal rates for those products not listed in Table 8. The reactivity effect may be satisfactorily represented by do/dt =Y —~ (A + R)p , where = figsion-product reactivity effect (%), = time (year), | yield (0.93%/year), = burnout rate (6.8 year)_l, o o> 4 ot O il = removal rate (year~!l). *0ther nonfissile uranium nuclides further dilute the 233y dilu- tion to 12% 233U may require additional 238y, 22 The fit to data representing four removal times from five years to infin- S ity had an absclute standard deviation of (.28%, which was considered adeguate. Studies using this model indicated that removal times of a few years but shorter than infinity were not worthwhile, and the results of this section assume R = 0. 3.1.4.3 Transplutonium effects % was made, and the con- A detailed study of transplutonium effects clusion was reached that the resulting fissile production only partially offsets the capture. The balance is less favorable than in reactors of higher power density because of the partial decay of 2%%Cm to 24CPu, which has comparatively less value. The study showed that each atom of 240py produced from 239y is joined by 0.11 additional atoms from the decay of 2%4¢cm, For each neutron absorption in 242py calculated without the trans- plutonium effect, 4.0 additional absorptions and 3.2 additiomnal fission neutrons ultimately result. Although the actual time effects are complicated, the net effect was approximately represented as an additional fictitious nuclide, which was produced by capture in 2%2py and had the absorption cross section of 242py . and no progeny. This would be slightly conservative at equilibrium and probably at earlier times also. 3.1.5 Static neutronic results 3.1.5.1 Inventory and neutron utilization Table 9 indicates the inventory of actinides at the beginning, mid- dle, and end of the 30-year operating period, assuming a 75% capacity fac- £ 235U tor. The high initial loading o is largely replaced by 233y pred by the system in the first half of the lifetime. Toward the end of lifetime, enriched uranium additions required to override fission—product buildup cause a final inerease in both the 235 and 23% content. The plutonium inventory is never large because of its high cross section in this spec- trum. Table 10 shows the midlife neutron utilization information. Note the low capture rate in nonfuel salt constituents (0.0153) and the fission- 23 s Table 9. Actinide inventories in DMSR fuel salt Inventory (kg) goL? Mor? roL’ 232 110,000 103,000 92,900 233p, 0 L5 38 233yd 0 1,970 1,910 234y 0 372 596 235yd 3,450 1,020 1,250 236y 0 661 978 2375p 0 75 136 238y 14,000 19,600 28,600 239p,d 0 179 231 240py 0 102 133 241p,d 0 76 100 242py 0 99 179 238py 0 36 93 Total actinides 127,000 127,000 127,000 Fissile uranium 3,450 2,990 3,160 Total fissile 3,450 3,440 3,490 %Beginning-of-life. bMiddle-of-life. CEnd-of-life. dNuclide treated as fissile in inventory calculation. product capture rate (0.0563). A total of 22.2% of the fission take place in 23873 gnd its progeny, even though they comprise only 9.8% of the fissile inventory. In spite of the high value of v for these nuclides, they would not be a sufficient fuel without the thorium chain. The slight contribution of the transplutonium nuclides to total mass has been ignored, and the absorption value shown for this nuclide group is a net of absorp—- tions less fissions. About 4% of the 2%lPu is lost through decay to zinm, a poor fuel. The capture in 233p, ig particularly expensive be- 233U cause each such atom otherwise would result in a highly profitable fission. 24 Table 10. Nuclide concentrations and neutron utilization after 15 years of DMSR operation a . Concentration Neutron Fission Vvoe /o Nuclide (x 10%%) absorption fraction £7a 232y 2,561 0.2561 0.0017 0.0070 233py 1.13 0.0018 0.0000 0.0033 233, 49.0 0.2483 C. 5480 2.2427 2344y 9.21 6.0120 0.0002 0.0143 235y 25.1 0.1161 0.2272 1.9894 236y 16.2 0.0075 0.0001 0.0168 237, 1.83 0.0047 0.0000 0.0102 238y 476 0.0901 0.0017 0.0194 239p,, &, 34 0.0896 0.1578 1.7905 24 0p., 2.46 0.0324 0.0001 0.0032 2ulp, 1.84 0.0293 0.0628 2.1754 242p,, 2.38 0.0039 0.0001 0.0136 ¢ Transplutonium 0.0014 238p,, 0.882 0.0024 0.0003 0.1245 - Total actinides 0.8956 1.0000 Fluorine 48,000 0.0079 Lithium 24,500 0.0062 Beryllium 5,470 0.0012 0.9109 Graphite 92,270 0.0172 Fission products 0.0563 Total Oo 9844 a Nuclei per cubic meter of salt or moderator. Absorption per neutron bern; leakage is 0.0156. 240 241 e Includes Pu, and 242py produced from a decay of Cm. Pu, 2ik 3.1.5.2 ¥Flux and power distributions and graphite lifetime The relative fast flux (E > 52.4 keV) and power-peaking factors are given in Table 11, These factors include the effects of flattening. For comparison, the overall fast flux peaking in an unflattened core would be ~2.3: the neutron leakage, however, would be only 0.8% vs 1.36% for this COore. 25 S Relative power distributions (Fig. 4) show no serious problems. The peak occurs in the well-cocled inner zone. A power peak per unit of core volume occurs in the gap between the core and the reflector, but the power ' per unit volume of salt is actually relatively low in that regiocn. Table Il. Neutron flux and power-peaking factors Fast flux Power Radial 1.32 1.36 Axial 1.15 1.15 Overall 1.52 1.56 ORNL DWG 80 4265 FTGC 1 E i t I [ I I AX|ALy L_ g 15 RADIAL— 1 >— '— 3 = o — L = C O Lu = '—: =T, 1 L o (50% SALT) PLENUM OR \ GAP CORE A (20% SALT] CORE B (12.9% SALT) _f—qx\%‘ja\‘ 6o | 1 l l | ; 1 l 0 0.5 i 1.5 2 25 3 35 4 25 GISTANCE FROM CENTER GF CORE (m) Fig. 4. DMSR relative power-density distribution. Axial and radial profiles are separately and arbitrarily normalized. 26 The absolute maximum fast flux is of special interest because of its effect in limiting the graphite lifetime and, thus, in defining the core size. The maximum damage flux calculated in this study occcurs near the edge of the inner core and is, at full power, bpax = 3.9 % 1017 neutrons m~ 2 51 (3.9 x 10'3 neutrons cm™? sy . In 30 years at 75% capacity factor, this leads to a fluence of 2.7 x 102 5=2 (2.7 x 1022 ¢p~?), which is well below the nominal graphite damage limit of 3 x 10%% m™2 (3 x 1022 cm~2), 3.1.5.3 Spectral and cross—section effects A summary of relative absorptions, fission neutron productions, and neutron flux by energy group is shown in Table 12. Many of the captures are in the resconance range, largely in thorium and 238U9 which leads to a larger absorption fraction in that group. sions are caused by thermal neutrons. In contrast, most of the fis- Table 12, Spectral distribution in neutronic effects in a DMSR core Neutron Relative Fraction of Fraction of energy neutron neutron fission neutrons group flux absorptions produced Fast 66 0.007 Resonance 131 0.087 Thermal 134 0. 906 Of special interest are the resonance cross sections of 232y, and 238y, because these largely determine the relative weight of the high- yield 232Th breeding chain vs the lower-yield 238y chain. Table 13 shows the effect of the lumping parameter % on these data at typical densities. This shows that the spectral difference between the cells of core zones A and B (Fig. 2) gives a lower 2387 capture effect in the harder spectrum 27 S Table 13. Effect of lumping on key resonance cross sections _ 238 Corga Salt zone 2 03(232Th) 63(238U) o {770) zone in cell (em} (barns) (barns) o, (23211) A Inner 2.540 2,44 7.86 3.22 A Outer 2,032 2,51 7.96 3.17 B Inner 2,540 2.42 7.96 3.29 B Outer 1.022 3. 14 10.6 3.38 Gap 5.0 2,14 6.6 3,08 %see Fig. 2 for identification of core zones. of zone A, as judged by the two zones with £ = 25.4 mm. To estimate the effect on neutron yield, the 232Th chain has an ultimate yield in a par- ticular situation of 1.06 neutrons per capture in 232Th, while the yield of 238y ig only 0.84 (Ref. 9). With ~40% of the fertile capture in 238y, as it is for the present system, a 10% increase in the 238-t0-232 capture ratio would reduce neutron yvield by ~0.5%. Accordingly, the variation in Table 12 is not a large effect. Even though cell geometry changes the cross sections significantly, the 2329 and 238y changes approximately cancel each other. Another variable of interest is the density of the fuel-salt heavy nuclides. Table 9 shows that 238y density approximately doubles during the life of the system, and this is not accompanied by a corresponding change in 232y, Varying over a range of reasonable interest, resonance data vs density are shown in Table 14 for the case of core zone A with £ = 25.4 mm, While the 232Th density increases by 51%, the product of den- sity and cross section increases by only 28%. For 238U, the density in- creases by 129%, and the product increases by only 577%, thus illustrating that nuclide density and its effect on resonance cross section are both large, but partially cancelling, effects. A similar table for other nu- clides for which resonances were calculated shows relatively less influ- ence of nuclide density on cross section (Table 15), 28 Table 14, Effect of nuclide density on key resonance cross sections Nuelid N Y Na uctlde 1024 puclei/m3) (barns) (m~ 1) 2327y 2200 2.61 0.5742 2400 2.52 0.6048 2582 2.44 0.6300 2800 2.37 0.6636 3318 2.21 0.7332 238y 350 9.03 0.3160 481 7.86 0.3781 650 6.85 0.4452 722 6032 0.4563 800 6022 0.4976 Table 15. Effect of nuclide density on other resonance cross sections - Concentration o AN/N A(No_ )Y/ {(Ns.) a Nuclide (102% nuelei/m3) (barns) (%)Y %%) & 233y 47,2 33.3 233y 56,0 32.7 19 17 234y 8.62 39,9 234y 15.8 35.6 83 64 235y 25.9 27.5 235y 40,8 27.3 58 56 236y 16.4 22.9 236y 26.3 21.1 60 48 Spectral effects are also important, because a more thermal spectrum improves the neutron yield of both 233y and 23U but also results in more parasitic capture in these fissile nuclides. To illustrate this, Table 16 shows the effective neutron yield for the hard spectrum of core A vs the soft spectrum of core B. While vog/0, within each neutron group shows relatively little change, the overall ratio shows a 37 increase in yield because of the scfter spectrum. 29 e Table 16, Effect of neutron spectrum on neutron yield for homogenized cell material Neutron Core energy Neutron yield index zone eroun (vogl/oy) A Resonance 0.338 A Thermal 1.442 A Overall 1.050 B Resonance 0.330 B Thermal 1.432 B Overall 1.080 a - & > o * See Fig. 2 for identification of core zones, 3.1.6 Burnup results 3.1.6.1 Reactor fuel cycle The time history of the fuel cycle in the DMSR provides some insight into the uranium resource utilization in this concept. The available re- activity in the core (Fig. 5) shows an increase during the first year as the inventory of 233U, a more efficient fuel than 23°U, builds. This rise would have to be controlled so that fuel consumption was minimized. Thus, a temporary removal of some denatured fuel or additions of fertile material might be more effective than insertion of simple neutron poisons. After the first year, the reactivity begins to decline as fission-product poisoning increases and overcomes the 233y effect. Reactivity is subse- quently kept above 1.0 by periodic additions of makeup fuel, containing 20% enriched 233U, The net conversion ratio of the system (fissile production divided by fissile consumption), which'is shown in Fig. 6, undergoes a much more persistent rise that lasts about five years before a gradual decline sets in that lasts until the end of the 30-year cycie. Much of this decline is attributable to neutron poisoning by 238y, which is added with the s 30 ORNL-DWG 804266 ETD 5.0 T I | l ! 233, & FIRST FUEL Jr“ U suILoup //_ADDHWON AVAILABLE REACTIVITY (% {Ak/Kk)] OPERATING TIME (yr) Fig. 5. Time variation of core reactivity in a once-through DMSR operating at 75% capacity factor. ORNL-DWG 804267 ETD 0.9 T l CONVERSION RATIO | | G 10 20 30 OPERATING TIME (yr} 0.7 Fig. 6. Conversion ratio vs time. makeup fuel. The lifetime average conversion ratio for the 30-year fuel cycle is close to 0.8. The schedule of fuel additions, including the initial critical lcad- ing for a 1-GWe plant operating at a 757% capacity factor is shown in Table 17. This table also includes the quantities of the U30g and separative work required to supply the fissile material, Thus, the lifetime core re- quirement would be about 2000 tons of U30g if no credit were allowed for the end-of~life fissile inventory. However, uranium is readily recover— able from this fuel in a pure and reusable form as UFg;. The recovered uranium would have to be reenriched, either by isotopic separation or by addition of high-enrichment fuel, before it could be reused in another DMSR, but reuse in some manner might be preferable to discarding the -------------- 31 Table 17. Fuel addition schedule for once-through DMSR 2387 a4ded 235y ,dded U30g requirement Separative work Year a requirement (kg) (kg) (Mg} (103 kg) ob 14,000 3,450 788 789 I 0 0 0 0 2 174 0 0 0 3 105 0 0 0 4 890 - 203 46.4 46.4 5 0 0 o 0 6 822 203 46.4 46 .4 7 0 0 0 0 8 822 203 46, 4 46,4 9 822 203 46.4 46,4 10 0 0 ' 0 0 11 822 203 46.4 46.4 12 822 203 46.4 46,4 13 822 203 46,4 46.4 14 822 203 46.4 46.4 15 822 203 - 46,4 46.4 16 0 0 0 0 17 822 203 46.4 46.4 18 822 203 46.4 46.4 19 822 203 46.4 46.4 20 822 203 46,4 46.4 21 822 203 46.4 L6, 4 22 822 203 46, 4 46,4 23 822 203 46.4 46,4 24 822 203 46.4 46.4 25 822 203 46,4 46.4 26 822 203 46,4 46.4 27 822 203 | 46.4 46,4 28 822 203 . 46.4 46.4 29 822 203 46.4 46.4 Total 32,400 7,920 1,810.0 1,810.0 % Mg = 1.102 short tons. Initial loading. 32 "spent” fuel. If credit were allowed for the residual fissile uranium in the salt {(plutonium presumably would not be recovered), the net UBOS requirement would be reduced by almost one-half. The temporal distribution of fuel requirements in a DMSR is also significant. The data in Table 17 show that only about 36Z of the makeup fuel is required during the first 15 years of the cycle; the major demand occurs toward the end-cf-life. Thus, if the reactor were operated at a lower capacitv factor in later years,* the U305 requirement could be re- duced further or the plant calendar lifetime could be extended. The ad- vantage associated with the time distribution of the makeup fuel require- ment is partly offset by the large initial fuel loading and the high in- plant fissile inventory. Therefore, an optimum fuel cycle might conceiv- ably balance a lower initial loading (and inventory) having a lower net conversion ratio against a higher requirement for makeup fuel. There ap- pears to be some latitude for optimization of the fuel cycle in this area. 3.1.6.2 Potential for improvement While the fuel utilization of this conceptual system comparés favor- ably with that of other reactor systems, some further improvementis may be possible. Only a limited range of fuel volume fractions and core zone sizes has been considered for this core, and other values could lead to higher performance. However, there appears to be little potential benefit in using more than two core zones. | The actinide content of the salt is thought to be near optimum for long~term, high-performance conversion, but, as implied previously, an- other concentration might be better for the 30-year cycle. Certainly, some improvement in fuel utilization would come from relaxing the re- quirement for 238y content either of the system in operation or of the makeup material being added. Table 18 shows the approximate effect of these comstraints on 235U requirements. Removing the requirement to fully denature the makeup feed material has only a small effect on the fuel requirement. Similarly, increasing the allowed enrichment of 233y *This is frequently done in electric power stations as newer and cheaper plants are built, 33 Table 18. Effects of denaturing on 30-year cycle performance 23 23 235y In-core ) aSq U Total enrichment initial feed 536 ?eed loading requirement U enrichment 235y 233 “(kg) (kg) (kg) 0.1 0.2 0.142828 3450 5120 8570 0.2 0.2 0.142828 3450 4470 7920% D4 0.2 0.142828 3450 4260 7710 0.2 0.2 0.071414 3450 4670 8120a 0,2 0.2 0.142828 3450 4470 7920 0,2 0.2 0.285656 3450 4470 7920 0.2 O.l 0.142828 3980 5120 9100 0.2 0.2 0.142828 3450 4470 7920% 0.2 0.4 0.142828 3040 4670 7710 0.2 1 1 2800 4060 6860 1 1 i 2800 2800 5600 aDenotes reference conditions. in the core has little effect. Only complete removal of all enrichment constraints would achieve an important fuel saving of 29%Z. Thus, the re- quirement for denaturing cannot be regarded as an overriding limitation on the potential performance of this fuel cycle. 3.1.7 Dynamic effects The dynamics of the DMSR would be dominated by the following factors: 1. a prompt, negative fuel-temperature coefficient of reactivity; 2. a slow, positive moderator-temperature coefficient of reactivity; 3. a negative fuel-salt density coefficient of reactivity; 4., an interaction between fuel-salt flow rate and the neutronic response of the core caused by the sweeping of delayed neutron emitters out of the core; 5. a long fuel-salt residence time in the core (relative to the average neutron lifetime). 34 3.1.7.1 Material reactivity worths The reactivity worths of the major fuel constituents are shown in Table 19. The total worth is negative because the effects of fertile thorium and 238% overcome the positive effects of the fissile materials. This means that the reactivity could be made significantly higher by re- moving fuel salt, although the breeding performance would be reduced. Table 19. Material concentration coefficient of reactivity c c Specific coefficient Total coefficient OmPOBENL (ax/k)/&N (10724 g3y (8k/k)/(AK/R) Fuel salt —0,14 Uranium 0.00026 0.15 Plutonium —. 00091 0.010 Thorium —£0.00011 .29 33p,4 ~0.0023 —0.,0026 233y 0.0040 0.20 234y —0.0015 —0.014 2335, 0.0025 —0.063 236y —0. 00050 ~0.0079 238y —0.00019 —0.092 Removing 1% of the uranium would have a reactivity effect of —0.0015 Ak/k, and reinserting it would have a comparable positive effect. A com- parable result for plutonium would be only 0.0001 Ak/k. If 1% of the fuel salt could be replaced suddenly by bubbles, the effect would be an increase of 0.0014 in reactivity, which is sufficient to induce a signifi- cant system transient. In practice, no likely mechanism exists that could bring about such an effect suddenly. The specific coefficients show that, atom for atom, 233y is a much more reactive fuel than 23°U or plutonium in the reference isctopic mix and that 2387 ig a greater depressant than thorium. 3.1.7.2 Temperature effects on reactivity Temperature affects the reactivity of the core by (1) broadening narrow cross—-section resonances, thus increasing their capture rate 35 (Doppler effect), (2) changing the energy distribution of the thermal neutron spectrum, and (3) causing expansion of the constituent materials. The expansion changes both the size and density of the core, as discussed earlier. Table 20 shows the various components of the total temperature coefficient. The fuel coefficient is dominated by both a large, negative Doppler component and a similar spectral component. This means, for exam-— ple, that an increase of 100°C in fuel temperature would reduce reactiv- ity by 0.009 essentially instantaneously. Table 20. Temperature coefficients of reactivity for DMSR Component Value Fuel—-salt Doppler 57 Fuel-salt density 30 Fuel-salt thermal spectrum ~60 Total fuel salt ~87 Moderator density —2.2 Moderator expansion 7.2 Moderator thermal spectrum 14 Total moderator 19 Total core —68 Reflector density 0.1 Reflector thermal spectrum 1.2 Reflector and vessel expansion —4.9 Total reflector —3.8 Total reactor —72 The moderator effect is dominated by positive spectral and expansion effects. This effect is relatively slow to appear, however, because the time constant for conduction heating of the graphite is on the order of 140 s, If the temperature change were cause& by a rapid power increase, a small portion (~5%) of the excess power would appear in the mcderator 36 T immediately because of deposition of energy by fast neutrons and prompt e gammas. Because the heat capacity of the moderatcor in a zone is always at least five times that of the fuel, the effect of direct transient mod- erator heating would be negligible. The reflector and vessel coefficients would probably be very slow in taking effect because of large heat capacities and low fuel flow rates. Their total is dominated by a negative expansion term. 3¢1.7.3 Delayed-neutron effects The delayed—neutron fraction of 233y is not much higher than that of plutonium. In the reference cycle, the contribution of 233y ig signifi~ cant {Table 21). Table 2!. Delayed-neutron fraction, B Contributor Fission fraction Contribution to B 233y - 0.55 0.0014 235y 0.23 0.0015 Plutonium D.22 0.00046 Total B 0.0034 An unusual aspect of MSRs is that the fuel circulates fast enough to remove significant numbers of delayed-neutron precursors from the core be- fore the neutrons are emitted. The lumped-parameter kinetics equations are taken as dCi(t) _ PE -A3iT T = 44— A436i(8) — Rleg(e) — gyl — e ], .and dP _ P(k — kB — 1) o 'a?:‘ - '“_EK___+Z)\1C1(t) 5 1 37 C; = relative delayed—-neutron precursor concentration, t = time, P = reactor power, B = delayed-neutron fraction, k = multiplication factor, A = prompt-neutron generation time, A3 = delayed-neutron precursor decay constant, aj; = delayed-neutron fractional yield, R = coolant flow constant, T = mean salt transit time in external loop. These equations then show that, where dollars of reactivity are defined as § = (k — 1}/k&B, the steady balance condition requires a nonzero value of $§. Thus, RE; $(steady state) = Oy e gllifliflREi’ where E; is equal to 1 -e_liT. Defining a new effective reactivity as R A =§ — Zai‘————)\-'l'R’ i 1 we can write the inhour equation relating asymptotic inverse period w to A, the amount of reactivity in excess of that required to maintain steady state under the given flow rate. A= ™| = . z:u,- w + R(Fi — Ei) Aq 1 s I Ai + w + RFy Ai + REj4 where F; is equal to 1 — exp{—(Xj; + w)tl. The prompt-neutron generation time was calculated by the boron-poison method to be 362 us. For now, we will ignore the difference between B and Bgff, which is expected to be small in low-leakage systems. Table 22 was compiled using standard 38 Table 22. Kinetic response of DMSR Flow Flow reactivity Net Reactor constant, R loss, § — & reactivity, & period, l/w (s~ 1y (dollars) (dollars) (s) 0 0 0 o 0 Q 0.11 100 0 0 0.26 30 0 0 0.45 10 0 0 0.69 3 0 0 0.92 I 0 0 1.28 0.3 0 0 2.04 0.1 0.0515 0.23 0 o 0.0515 0.23 0.07 100 0.0515 0.23 0,16 30 0.0515 0.23 0.29 10 0.0515 0.23 0.40 3 0.0515 0.23 0.70 1 0,0515 0,23 1.05 0.3 0.0515 0.23 1.82 0.1 delayed-neutron data.!’ The flow constant of 0.0515 s~! corresponds to full-power operation with a mean fuel residence time of 19.4 s in the core. These data show that less excess reactivity is required for a given small power response when the salt is flowing. This reactivity difference becomes constant at higher reactivities. Because of the very long genera- tion time, the response to net reactivity changes of more than 1 dollar would be much smaller than that of many reactor types, which is character- istic of over-moderated graphite assemblies. The overall result would be a system with a power level that fluctuates more than usual because of in~ herent operating ncise but that would be relatively easy to shut down by control rod action in an unplanned event. Power fluctuations woculd be ex- pected to have little effect on the external system because of the large heat capacity of the core and the low flow rate. 39 3.1.7.4 Contrel requirements for normal operation Assuming a core preheated to 775 K* and near critical, a reactivity increase of 0.07% must be supplied by the control system as fuel flow through the core is started to compensate for the loss of delaved neu- trons. An increase of about 1.17% must be supplied to bring the fuel, moderator, and vessel to the average operating temperature near 925 K. Xenon concentration is kept to a negligible level by the salt cleanup system and, therefore, has little effect on the control requirements. A more serious requirement is the longer—term positive reactivity peaking caused by early production of 233U; this is approximately a 37 effect. In addition, some shutdown margin (perhaps- 2%) would be required for safety. This would be sufficient to overcome a 127 change in salt den— sity, for example, or a comparable loss in actinide content of the salt. The total reactivity control span required with respect to a 775 K, no~-flow, just—critical core thus would be from +1.27% to —5.0%. Of this, only about —2% must be rapid in nature, inserted by an active control device. The remainder could be partially supplied by adjusting the com— position of the fuel salt. 3.1.7.5 Stability and transient safety The core is stable to all frequencies of oscillation because the negative prompt component of the temperature coefficient of reactivity dominates the positive delayed component. At frequencies below that as- sociated with the graphite thermal conduction process, the delayed com- ponent could subtract from the prompt component, but the effective coef- ficient would be no less negative than the total core coefficient. These frequencies would be on the order of inverse minutes and should pose no problem for contrel. The response to sudden changes in the fuel-salt inlet temperature is relatively slow due to a salt residence time of almost 20 s. For example, the reactivity response to an abrupt change in inlet fuel-salt temperature * s e - This would be an absolute minimum temperature because the salt would begin to freeze at lower temperatures. 40 of —50°C would be +0.004 (about 1.7 dollars) if the entire core could be filled. However, the external loops contain enough salt to fill only about one-third of the core; thus, the actual reactivity effect would be much smaller, and it would be inserted over a period of several seconds. The control system ccould readily compensate for such a slow change in re- activity. With no control response, a new power level would be gradually approached to counteract the cooling effect of the inlet condition, and then the continued heating effect of the higher power level would cause a reduction in reactivity and a return to a stable condition. 3.2 Reactor Thermal Hydraulics The purpose of the thermal-hydraulic analysis of the DMSR is to dem- onstrate that the concept is viable and not to provide a detailed design., Neither the funding nor the necessary thermal-hydraulic properties of mol- ten salt flow in a graphite core are presently available to perform the latter. Conservative estimates of important parameters are taken wherever possible; even if some of these should be nonconservative, simple modifi- cations of the core design apparently could lead tc acceptable results, Thermal-hydraulic behavior does not appear to be a limiting design con- straint on the DMSR reference core. Because of the relatively low power density of this concept, simple core coufigurations that were not possible in the MSBR reference designs may be considered. Three simple designs were considered: 1. a core made up of spaced graphite slabs, 2. a core made up of stacked hexagonal graphite blocks with circular coolant channels, 3. a core consisting of a triangular array of graphite cylinders with central coolant channels, Constraints that must be considered in selecting a core design in- clude maximum graphite element temperature, local salt volume fraction, and the desired 238y self-shielding effect, which imposes 2 minimum limitation on the cooclant channel dimensions. The temperature rise be- tween the coolant channel and the hot spot in the graphite moderator ele- ment is especially important because of the strong dependence of graphite ............. 41 dimensional change on temperature. The salt volume fraction and the 238y self-shielding effect strongly couple the thermal-hydraulic and the neu- tronic core designs. These combined constraints appear to rule cut the possibility of a graphite slab core configuration. Mechanical problems, especially the loss of coolant channel geometry causedrby shifting of stacked hexagonal blocks {thus creating stagnant or low flow zones), rule out the second option. In addition, that option would leave an un- desirably large fraction of the fuel salt in narrow passages between the hexagonal elements. The third design seems to fill all the requirements and is also very appealing because of its structural simplicity, which is important in a core expected to last the life of the plant. The outer diameter of the cylindrical graphite elements in the ref- erence DMSR is 254 mm (10 in.), which is machined down to 244 mwm (9.16 in.) in the central region (core zone A, Fig. 2). The diameter of the in- ner coolant channels is constant at 51 mm (2 in.). This design provides a salt fraction of 20.0% in the central region and 12.9% in the remainder of the core. The motivation behind this two-region design is to provide a first estimate of a flux-flattened core. Flux flattening is crucial to the design objective of reactor-lifetime graphite because both the maxi- mum graphite damage and the maximum graphite temperature are reduced. Figure 7 shows the arrangement of the graphite moderator elements in the outer region, zone B, which occupies most of the core velume. Note the 5l-mm-diam (2=in.) interior salt channels and the exterior salt chan- nels formed between the moderator elements. In zone B, the exterior chamn— nels have a uniform cross section along their entire length, except for possible orificing provisions at the ends. The arrangement is the same in the interior of the core (zone A), but the cuter diameter of the mod- erator elements in the interior is reduced. This provides the higher salt fraction and allows the exterior channels to interconnect in that region. Figure 7 alsoc shows the location of a 30° segment of a graphite ele- ment used in the analysis., The film heat-transfer coefficient at the graphite-salt interface is not well known, primarily because a helium film may exist on the graphite surface. This film would increase the thermal resistance but alsc would give a no-drag wall boundary condition to the salt velocity profile. In addition, near the moderator element contact 42 CRNL--DW{G 78 -10335R FUEL SALT MODEL SECTION FOR THERMAL ANALYSIS — GRAPHITE 254 cm Fig. 7. Arrangement of moderator elements in core B. points (which only exist in the outer core region), heat transfer would be greatly diminished. The following simplifying assumption was made (probably conservative) concerning the salt—-film heat-transfer coeffi- cient: within 15° of a contact peint, the héat-transfer coefficient is zero; elsewhere, it is equal to 80% of the value obtained by use of the Dittus-Boelter correlation.'® This assumption increased the calculated total temperature rise in the moderator element by ~307%7 over that ob— tained by using 80%Z of the Dittus—Boelter value for the entire surface. With these boundary conditions, the heat conduction equation in cy- lindrical finite difference forfi was solved in the 30° section at each axial node (40 nodes total) using the method of successive overrelaxation. Starting at the core inlet, the temperatures of an interior and an exterior salt channel are advanced in an axial marching—-type solution through the core, Axial and radial homogeneocus power profiles (see Fig. 4) were used, along with the following assumptions, to give salt—channel axial linear power profiles (Fig. 8) and local moderator vdlumetric power. At a given location, the vclumetric powers in an interior and an exterior channel were assumed to be the same, and the volumetric power within the graphite was assumed to be 1% of that in the salt and constant over the 30° sec- tion. Neutronic analyses of the previous MSBR design& provided the basis 43 ORNL DWG 83 4268 ETO 180 I I — T |_ I ! E weor e, ] TR - ~ —EXTERIOR £ CHANNEL = 140 | - x O L — < 120 - o L < WL 4 © 100 - / INTERIOR - L / CHANNEL wl T o { < 80 . L s v/ T 60 - - 40 L s | | I | i E 0 1 2 3 4 5 6 7 8 AXIAL POSITION (m) Fig. 8. Core hot—-channel linear power profiles., for the latter assumption; the present analysis is not detailed enocugh to yield a better estimate. As noted previously, the hot-channel axial linear power profiles of an interior and exterior channel are shown in Fig. 8. The hot channel oc- curs (radially) at the boundary between zones A and B. The central loca- tion of zone A (2.65 to 5.65 m) can be seen by the discontinuities in the linear power curves. The curves reflect both heat generated within the salt and heat transferred from the graphite to the salt. Figure 9 shows calculated temperatures in the interior and exterior channels and the maximum temperature in the moderator as functions of ax- ial position for the core hot channel. The highest moderator temperature cccurs at a position 3.0 m above the core midplane. Isotherms in the 30° moderator segment at this location are shown in Fig. 10. The assumption of no heat transfer within 15° of the contact point causes substantial distortion of the isotherms near the outer surface. The calculated maxi- mum graphite temperature is 741°C (1366°F), which is close to the maximum allowable temperature (~720°C) for zero positive irradiation growth at a total fluence of 3 x 1026 p—2, L, ' ORNL-DWG 80—-4262 ETD 750 T u u T E | ] T LOCATION OF GRAPHITE MAXIMUM TEMPERATURE (FIG. 10}-—— e 728 700 p— n ~J o ! TEMPERATURE (°C) [6}] S { 625 p— 1. MAXIMUM TEMPERATURE 660 =7 IN MODERATOR - 2. INTERIOR SALT CHANNEL TEMPERATURE 3. EXTERIQR SALT CRANNEL 575 TEMPERATURE - 550 ; | | { | | | ; ¢ 1 2 3 4 5 g 7 8 AXIAL POSITION (m) Fig. 9., Axial temperature profiles: graphite and fuel salt for core hot chamnel. Table 23 gives flow areas, salt velocities, Reynolds numbers, and heat—-transfer coefficients for the interior and exterior channels (in both core zones A and B) for two cases: the core hot channel and the core average channel. The salt velocities are those necessary for an equal 139°C (250°F) temperature rise across the core in the interior and exterior channels. The hydraulic diameters are not equal; thus, orificing of the interior channels would be necessary. This could be accomplished easily by reducing the diameter of the interior channel by ~507Z for a short interval near core inlet and/or outlet. Overall core orificing would alsc be necessary to equalize core exit temperatures. The fric- tional pressure drop across the core [~7 kPa (1 psi)] is insignificant when compared with the pressure drop across the primary heat exchanger 45 ........... CRNL--DWG 803-427C ETD MODERATOR ELEMENT CONTACT POINT ) EXTERIOR \— CHANNEL f N ooy ‘ r INTERIOR \ { CHANNEL i { f | | | ‘\ \ R | I - | 714720728 732 738 738 732 726 720 7147 Fig. 10. Isotherms in graphite at location of maximum temperature (°C). Film heat—-transfer coefficient h equals 0 where shown and equals 80Z of Dittus—Boelter correlation value elsewhere, Table 23. Thermal-hydraulic data for DMSR core Channel Hot Average Flow area, mZ Interior channel 2.025 x 10-3 2.025 x 1¢~3 4,580 x 10=3 4,580 x 1673 Exterior core A - 2.601 x 1073 2,601 x 1073 Exterior core B Salt velocity, m/s Interior channel 0.601 0,441 Exterior core A 0.418 0.307 Exterior core B 0.735 0. 540 Reynolds number Interior channel ' 1.034 x 10° 7.592 = 10 Exterior core A 6.759 x 10% 4.962 x 104 Exterior core B 6,490 x 104 4.765 x 10 Heat—transfer coefficient, W m2 K_l Interior channel 173 78 Exterior core A 130 59 Exterior core B 232 145 a Maximum/average power is 1.362 over ccres A and B. b Obtained by using 80% of Dittus-Boelter correlation value. 46 [~%00 kPa {130 psi)}. Graphite and fuel-salt properties used in the analysis were obtained from Ref. 8. The reference DMSR core design satisfies the two most important thermal-hydraulic considerations: (1) the maximum graphite temperature is low enough to allow it to last the life of the plant (24-full power years) and (2) regions of stagnant or laminar flow are avoided. Many variations on this design will be possible in achieving an optimum core, but the second consideration must always be noted. Because of the low thermal conductivity of the fuel salt, excessive temperatures can occur in very small stagnant or laminar flow régionse The graphite elements must retain their geometric integrity and must mnot create flow blockages. Extensive in-pile testing would be necessary to ensure that both of these considerations were met before construction ¢of a demonstration plant could - be undertaken. 3.3 Fuel Behavior Excellent neutron economy is an absolute requirement for a thermal breeder (such as an MSBR), and fuel components with acceptably low neutron cross sections are few, We recognized very early in the MSBR development effort that (1) only fluorides need to be considered, (2) only LiF and BeF, would prove acceptable as fuel solvents (diluents) for the fissile and fertile fluorides, and (3) the LiF must be highly enriched in ‘Li. For a break—even reactor or for one that, though it retains most of the fission preducts within the fuel, is to be an effective converter, some sacrifice in neutron economy may be permissible. However, no likelihocd for success seems possible with diluents other than BeF; and LiF highly enriched {probably to 99.99%) in “Li. Accordingly, the fuel system for a DMSR necessarily will be very similar to the system that received intensive study for many years in MSBR development.- A considerable fund of information exists about chem- ical properties, physical properties, and expected in-reactor behavior of such materials. 47 3.3.1 Rasic considerations 3.3.1.1 Composition of DMSR fuel Choice of initial composition. A DMSR will derive some of its fis- sion energy from plutonium isctopes, but 233 and 2%°U will be the primary 2383 s the 232Th fissile isotopes, while 232Th, with important assistance from fertile material. Clearly (see previous neutronics discussiocn) the concentration will need to be markedly higher than the total concentration of uranium isotopes. The only stable fluoride of thorium is ThF,; thus, it must be used in such fuels. Pure UF3 is appreciably disproportionated at high tempera- tures by the reaction 4UF 3 = 3UF, + U . Generally in molten fluoride solutiomns, this reaction proceeds appreciably at lower temperatures. A small amount of UF3 will be formed within the fuel by reaction (reduction) of UF, with species within the container metal and, as explained below, a small quantity of UF3 deliberately main- tained in the fuel serves as a very useful reduction~oxidation (redox) buffer in the fuel. Such UFg3 is sufficiently stable in the presence of a large excess of UF,, but UF, must be the major uranium species in the fuel, 19520 Conversely, PuF, is reduced by the metallic container (and also by UF3), and PuF3 is the stable fluoride of this element in DMSR fuels. Phase equilibria among the pertinent fluorides have been defined in detail and are well documented.!972! Because the concentration of ThF, is much higher than that of UF,, the phase behavior of the fuel is dic- tated by that of the LiF-BeF,~ThF, system shown in Fig. 1ll. The compound 3LiF+ThF, can incorporate Be2* ijons in both interstitial and substitu- tional sites to form solid solutions whose compositional extremes are rep- resented by the shaded triangular region near that compound. The maximum ThF, concentration available with the liquidus temperature below 500°C 1is just above 14 mole %. Replacement of a moderate amount of ThF, by UF, scarcely changes the phase behavior. 48 ORNL-LR-DWG 37420AR7 - The, 4414 e TEMPERATURE IN °C COMPOSITION IN mole 7, LIF-4ThE s LIF2ThE P B8S7 4 LiF 7 i< T - Bef, . 848 aJPBefifsoc%so 400 400 450 500 555 P 458 £ 360 Fig. 1l. System LiF-BeF,;-ThF,. The MSBR proposed to use an initial fuel mixture containing 71.7 mole 7 LiF, 16 mole % BeF,, 12 mole % ThF,, and 0.3 mole 7 (highly en- riched) UFy. The optimal initial concentration of fuel for a DMSR is not yet precisely defined. The initial fuel likely will need to contain be- tween 9.5 and 12.5 mole % of heavy metal (uranium plus thorium}, with uranium (enriched to 20% 233y) corresponding to about 12% of the total. The composition range of interest to DMSRs, therefore, is likely to be bounded by (all concentrationms in mole %): 70.8 LiF; 19.7 BeF,; 8.35 ThFy; 1.1 UF, and 71.5 LiF; 16 BeF,; 11 ThF,; and 1.5 UF,. For such compositions, the liquidus would range from about 480 to 500°C. Most chemical and physical properties of the chosen composition can be inferred reasonably well from existing data for the MSBR reference compesition. 49 2 Variation of fuel composition with time. Fission of 35U in the operating reactor will result in a decrease of that isctope and an in- 233 232Th, growth of fission preducts and in the generation of (1) U from (2) 2394 from 238U, and (3) numerous transuranium isotopes., Further, a once-through DMSR will require additions of uranium at intervals during its lifetime. For example, a DMSR with a fuel containing 9.5 mole % heavy metal would contain about 110,000 kg of 232Th, 3,450 kg of 235U’, and 14,000 kg of 238y ¢ startup. During 30 years of operation at 75% plant factor, it would require the addition of 4,470 kg of 235U and 18,400 kg of 238U. If such a reactor received only additions of UFy and UF3 and if ne fuel were removed,* the final quantities and comncentrations of heavy metals in the fuel would be those indicated in Table 24. The end-of- life fuel would also contain about 1.4 mole % of soluble fission-product species and would have a total of nearly 2.4 mole % of uranium isotopes, about 0.053 mole % of plutonium isotopes (about 32% of which is 239%py), and less ThF, than the original fuel. Thus, the concentration of heavy metal in the fuel changes very little although the species do change; total heavy metals in the end-of-life fuel equal about 9.3 mole 7 compared with an initial 9.5 mole %. Therefore, the physical properties of the fuel would not be likely to change appreciably during reactor life al- though a gradual change in some chemical properties would be expected. Additions of uranium can be made conveniently as a8 liquid LiF-UFy mixturel? (liquidus 490°C) while the reactor is operating, as was done many times during operation of the MSRE.22 To keep a proper concentration of UF3 in the fuel and possibly to remove tramp oxide-ion contamination from the fuel, some fuel maintenance operations will be necessary. The combination of these relatively simple operations likely will result in sufficient addition of LiF and BeF2 to require on-site removal and stor- age of a small fraction of the fuel before reactor end-of-life. These operations and the resulting fuel management options are described in a later section after the chemical basis for them has been presented. *This type of operation may be possible although subsequent discus- sion will show other more likely modes of operation. 50 Table 24. Approximate heavy meta. inventory of end-of-life fuel in hypothetical DMSR with no fuel removal Inventory Specie kg Mole 7% 2329y 92,000 6.84 233pg 38 2,8 x 1073 233y 1,910 0.140 234y 596 0.043 235y 1,250 0.091 236y 978 0.071 238y 28,600 2.05 238py 93 6.7 x 1073 239y 231 0.0165 240py 133 9.5 x 1073 2hlpy 100 7.1 x 1073 242py 179 0.0126 237p 136 9,8 x 1073 aOperated at 1 GWe for 30 years at 75% plant factor. 3.3.1.2 Physical properties of DMSR fuels Table 25 shows key physical properties of the compesitions identified previously to represent the likely limits for DMSR use. As described in detail elsewhere,l9:20,23,24 several of these properties — particularly those of the salt with 9.5 mole % heavy metal — are interpolated from measurements on similar salt mixtures. From careful consideration of very similar mixtures for use in the MSBR, the properties clearly are adequate for the proposed service. However, because estimates rather than measured values are presented in several cases, an experimental program would be 51 Table 25. Physical properties for probable range of DMSR fuel compositions Heavy metal content Properties Low High Composition, mole 7% LiF 70.7 LiF 71.5 BeF, 19.8 BeF, 16 m* 9.5 HM 12.5 Liguidus, °C 80 500 Properties at 600°C Density, Mg/m3 3.10 3.35 Heat capacity, kJ/kg*K 1.46 1.36 Viscosity Parvs 0.012 0.012 Centipoise 12 12 Thermal conductivity, 1.2 1.2 W/K*m Vapor pressure Pa <10 <10 Torr 0.1 <0.1 oM = heavy metal fluorides. required to firm up the physical properties of the composition(s) chosen for service. Careful reevaluation of the properties woculd not be likely to disqualify these compositions from DMSR use. 3.3.1.3 Chemical properties of DMSR fuels A molten—salt reactor such as a DMSR makes a number of stringent demands on its circulating fuel. Some of these demands have been im- plicit in the foregoing discussion of fuel behavior; examples include the obvious need to accommodate moderate concéntrations of UF, and large concentrations of ThF, in relatively low-melting mixtures of materials with small cross sections for parasitic neutron capture and the need for adequate heat—transfer capability. The fuel must be capable of convenient 52 preparation in a pure homogeneous form for introduction inte the reactor. g In addition, the fuel must (1)} be compatible with the structural and the moderator materials during normal operation, (2) be stable to intense radiation fields, and (3) tolerate fission of uranium and plutonium and the development of significant concentrations of fission products and plutonium and other actinides. Also, without dangerous consequences, it must be able to withstand a variety of off-design situations such as heat—exchanger leaks or possible ingress of air. Finally, although not presently necessary, it is desirable that at end~of-life the fuel be ame- nable to recovery of fissile, fertile, and other valuable materials. The ability of the fuel to meet or not to meet these diverse and conflicting demands largely depends on the chemical properties of the fuel. Most of the details of fission—product behavior are deferred te a subsequent sec~ tion, but much of the basis for expected fuel performance is presented in the following discussion. Thermodynamics of molten flucride sclutions. The thermodynamic properties of many pertinent species in molten LiF-BeF, solutions and a smaller number in LiF-BeF,-ThF, solutions have been studied in a long- continued experimental program. A variety of experimental techniqués was employed. Much of the data was obtained by direct measurement of equilib- rium concentrations and partial pressures for reactions such as H + FeF = re0, | + 2uF 2e) T O T T (@ (8) and 2HF + BeO = BeF + H»0 (g) 7 ") T T2 T T2(g) {(where g, ¢, 1, and d represent gas, crystalline sclid, molten solvent, and solute, respectively) using the molten fluoride as the reaction me- dium. Many studies of sclubility behavior of sparingly soluble fluoride species have alsoc been made. Baes?°?2% has reviewed all these studies and has tabulated thermodynamic data for many species in molten Li,BeF,. Table 26 shows values for standard free energy of formation of major constituents and some possible corrosion products. BaesZ%226 hag slso 53 e Table 26, Standard free energies of formation for compounds dissolved in molten LioBeF, b AGE¢900 K) Compound a b (keal/atom F)e LiF(l) —141.79 16.58 126.9 Ber(l) —243,86 30.01 108.4 UFng) —338.04 40. 26 100.6 UFq(d) -4 45,92 57.85 98.5 ThF”(d) —491,19 62.50 108.8 ZrFq(d) —452,96 65.05 98.6 Nin(d) —146,87 36,27 57.1 Fer(d) —154,.69 21.78 | 67.5 Cer(d) -—171u82 21g41 7603 Zrdapted from Ref. 7. bThe standard state for LiF and BeF, is the molten LioBeF, solvent. That for solute species — those with subscript (d) — is the hypothetical solution with the solute at unit mole fraction and the activity co- efficient the solute would have at infinite dilution. ®For conversion to SI, 1 kcal = 4,18 kJ. evaluated the effect of solvent composition in the LiF-BeF; system on activity coefficients of a variety of solutes. Using a sophisticated spectrophotometric analysis for UF, and UFg, more recent study by Gilpatrick and Toth to evaluate the equilibrium i B 'E-Hz(g) +UF;+(d) = UFg(d) + HF(g) in several LiF-BeF; and LiF-BeF,-ThF, solvents essentially confirmed the 54 UF; value of Table 26 (if the UF, value is accepted) and showed that the difference between UF3 and UF, standard free energies in LiF-BeF,-ThF, (72-16~12 mole %) is virtually identical to that of Table 26. Bamberger et al.2’ have shown that AGE for PuF3 in molten LizBqu* ig —1358 * 10.9 kJ/mole (—325.6 * 2.6 kcal/mole) and —1357 * 10.9 kJ/mole (—324.6 £ 2,6 kcal/mole) at 888 and 988 K, respectively. From these data and from the solubility of PuF3, they have estimated for pure crystalline PuF3 the following values: -+ AGE = —1453 + 10.9 kJ/mole (—347.7 + 2.6 kcal/mole) at 888 K , AGE = 1392 + 10.9 kJ/mole (—333.1 * 2.6 kcal/mole) at 988 K . Combining these values with those of Dawson et al.?8 for the reaction 4PuF; + 0, = 3PuF, + Pul, yields the expression AGE = —1611 + 36.4 (T/1000) = 11.7 kJ/mole or 385.4 + 8.7 (T/1000) + 2.8 kcal/mole for crystalline PuF, . No definitive study of AGE for PuF, in molten fluoride solution has been made. Its solubility (by analogy with those of ZrFy, UFy, and ThFy} is relatively high. Also, using the hypothetical standard state of unit mole fraction, it is more stable [perhaps by 63 kJ/mole (15 kecal/mole)] in sclution than as the crystalline solid. If so, the reaction UF3(d) + Pqu(d) = Pqu(d) +, UFL’(d) % The standard state of PuF used for solutes in Table 26. is the hypothetical unit mole fraction 3(d) 55 would be expected to have an equilibrium quotient (Q) of [PuF3] [UFu] = . =2 X 6 Pury ] [0F3] 1.23 107 where the brackets indicate mole fractions of the dissolved species. If only 2% of the uranium were present as UF3, the ratio of PuF3/PuFy would "be near 2.3 % 104, and, to sustain a PuF3/PuFy ratio of 1, only about 1} part UF3 per million parts UFy could be tolerated. Clearly, unless very oxidizing conditions are maintained in the melt, the plutonium is essen- tially all Pu’t, The solubility of PuF3 in LiF-BeF3-ThFy (72-16-12 mole 7) has been 29,30 measured in two laboratories. Bamberger et al.?? show the soclubility of PuF3 in mole 7% (SPuF3) to be given by log SpyrFsy = (3.01 * 0.06) — (2.41 £ 0,05) x (1000/T) , with a heat of solution (8Hg) of 46.0 £ 1.0 kJ/mole (11.008 £ 0.237 kcal/ nole). If so, the sclubility of PuF3 at 565°C, the likely minimum tem-— perature within the DMSR circuit, should be near 1.36 mole %Z. This value is nearly identical {as is the heat of solution) to that obtained by Barton et al.’! for CeF3 in the same solvent. However, the Indian study30 gave significantly lower values; for example, at 565°C that study would suggest that Spyp; should be near 1.1 mole %Z. The solubility of PuFj alone is undoubtedly much higher than is required for its use in a DMSR. However, as described in a later section, difficulties possibly could ultimately result from the combined solubilities‘of a number of trifluo- rides that form solid solutions. Given reasonable UF,/UF, ratios, americium, curium, californium, and probably neptunium also exist as trifluorides in the melt. No definitive studies of their solubilities in LiF-BeF,~ThF, melts have been made. Such studies are needed, but their individual soclubilities certainly will prove to be far higher than their concentrations in DMSR fuel. Our knowledge of the thermodynamics of molten LiF-BeF3-ThFy solu- tions appears adequate to guide the necessary development studies, but considerable research and development (as well as data analysis) remain 56 to bring that understanding to the level that exists for LiF-BeFy sclu- tions. Oxide~fluoride behavior. The behavior of molten fluoride systems such as the DMSR fuel mixture can be affected markedly by addition of significant concentrations of oxide ion. For example, we know that crys— tals of U0, precipitate when melts of LiF,-UF, are treated with a reac- tive oxide such as water vapor,19932933 The solubilities of the actinide dioxides in LiF-BeF,-ThF,-UF, mix- tures are low, and they decrease in the order ThO;, Pal,, UO0,, and Pu0,,251672% Moreover, these dioxides all possess the same (fluorite) crystal structure and can all form solid sclutions with one another. Solubility products and their temperature dependence have been mea- sured; 34743 their behavior is generally well understood. Trivalent plutonium shows little or no tendency to precipitate as oxide from LiF-BeF,-ThF,~-UF, mixtures.?’ Because relatively large solu- bility seems to be general for trivalent oxides, it is highly likely that precipitation of Am,0; and other trivalent actinides would be dif- ficult to achieve. 1f Pat is oxidized to Pa®" (which can be done readily in LiF-BeF,- ThF,-UF, by treatment with anhydrous and hydrogen-free HF gas), then Pa,0g (or an addition compound of it) can be precipitated selectively.®i,4" Such oxidation to Pa®*t can be avoided by maintaining a small fraction of the uranjum as UF3 in the fuel mixture. The relafively low oxide tolerance of DMSR fuel will require reason- able care to avoid inadvertent precipitation of actinide oxides within the reactor system. However, treatment of melts with anhydrous HF (even when substantially diluted with H,) serves to lower the oxide concentration to tolerable levels, 18,43 Compatibility of fuel with reactor materials. Molten fluorides are excellent fiuxes for many materials. Though some oxides are relatively insoluble, most are readily dissolved, and all are rapidly recrystal- lized; consequently, protective coatings are not useful, and the bare clean metal must withstand corrosive attack. The reactor metal (Hastel- loy-N, described in detail in Sect. 3.4) was chosen and tailored to be thermodynamically stable to the fuel components, as much as possible. 57 s Corrosion of Hastelloy-N by MSRE and MSBR fuel mixtures without ir- radiation and without the consequences of fission has been studied in sophisticated equipment for many years. It has been thoroughly de- 41,2,19,20,28-38,46-56 ,,4 can be said to be well understood. . scribe Table 26 clearly indicates that chromium is the most easily oxidized of the major Hastelloy-N components.* Corrosion of the alloy, therefore, is essentially by selective leaching of chromium from the alloy. A rapid initial attack can result from reactions such as FeFy + Cr » CrF2 + Fe , 2HF + Cr > CrF2 + Hz , 2Ni0 + ThFy * ThO2 + 2NiFg , and NiF2 + Cr * CrFz + Ni if the fuel salt is impure or if the metal system is poorly cleaned. These reactions proceed to completion at all temperatures within the re- actor circuit and do not afford a basis for continued attack. The most oxidizing of the major constituents of the fuel is UF,, and the reaction 1 — L UFq(d) +-§ Cr(c) = 3 Cer(d) + UFg(d) has an equilibrium constant with a small temperature dependence. When the salt is forced to circulate very rapidly through a large (140°C) tem- perature gradient, as is the case within the reactor circuit, a mechanism exists for mass transfer of chromium and for continued attack. The re- sult is that chromium is selectively removed from the alloy in high- | temperature regions and deposited on the alloy in low-temperature regions *Molybdenum fluorides are somewhat less stable than NiF;. MoFg(g) at 900 K has a standard free energy of formation25 of about 215 kJ (-51.4 kcal) per gram—-atom of F~. 58 of the reactor. The rate of transfer of chromium is limited by the rate at which the transferred chromium can diffuse intc the alloy in the low- temperature regionsc19’20’56 The results of two decades of sophisticated corrosion testing have demonstrated the validity of this mechanism and have shown that such corrosion™ will prove to be only a trifling problem for MSBR. Appreciable chromium depletion would be expected tc a depth of less than 0.13 mm/year (0.5 mil/year) in metal at 704°C, 19,147 The initial attack, which is not serious if proper purification of the salt and cleaning of the system have occurred, can be mitigated by the presence of a small quantity of UF3 along with UFy in the salt. Con- trol of the oxidation states of plutonium and protactinium and of certain fission products, along with control of the oxidative effects of the fis- sion process, furnish more cogent reasons for maintaining UF3 in the fuel mixture. Slight continuing corrosion is affected very little (if at all) by the presence of small quantities of UF3. The unclad moderator graphite is not wetted by or chemically réactive to the standard MSRE or MSBR fuel compositions, and these facts appear un— changed by intense irradiation and the consequences of fission, 9229537, 58 Estimates2? are that the MSRE graphite moderator stack (3700 kg) acquired less than 2 g of uranium during operation of the reactor. Obviously, no appreciable interaction of graphite with the fuel {(whose UFg/UF4 ratio was never above 0.02) occurred in that reactor. However, given a sufficiently high UF3/UF% ratio, formation of uranium carbides must be expected, and this should be avoided. Toth and Gilpatrick, who used spectrophotometry in a graphite cell with diamond windows to assay equilibrium UF; and UF, concentrations, have carefully studied uranium carbide formation using Li,BeF, (Refs. 59 and 60), other LiF-BeF, mixtures,®9,6l and LiF-BeF,-ThF, (72-16-12 mole Z) (Ref. 61) as solvents. A surprising finding of these studies is that, contrary to generally accepted thermodynamic data,62 UCs is the stable carbide phase over the temperature interval 550 to 700°C. el Figure 12 shows the results” of equilibration experiments in MSBR fuel solvent. Apparently, at the lowest temperature (565°C) within a DMSR, *Corrosion in the presence of fission and fission products is more complex. See Sects. 3.3.2 and 3.4 for additional details. 59 g ORNL-DWG 72-42321 TEMPERATURE (°C} 500 550 c00 650 700 _i —I—A] - -Ir‘ -,..l , | b 0.5 _3 . 0.4 L = + Q3 ) > ~ ‘ u” 0.2 & — 04 S/ . | | 1.30 1.25 1.20 1.45 110 1.05 1.00 1000/, oy Fig. 12, Equilibrium quotients, Q = (UF3)%/(UF,)3 vs temperature for UCy + 3UFy(gq) = 4UF3(4) * 2C in the solvent LiF-BeF;-ThFy (72-16-12 mole 7). _ formation of UC2 should not cccur unless the UF3/UFy ratio is above about 0.17. Further experimentation with larger systems is desirable, but UF3/UFy ratios of at least 0.1 apparently can be accommodated, if desir- able. ......... 60 3.3.1.4 Operational constraints and uncertainties s Most of the fuel behavior described or implied above may be consid- ered well authenticated. Several constraints and at least minor uncer- tainties are obviocus. 1. Fuel for a DMSR must be prepared from LiF containing a very high percentage of 711, 2. The fuel mixture must be managed and maintained so that an appre- ciable fraction of the uranium is present as UFj. 3. Additional experiments are necessary to establish exactly what fraction of the uranium may be present as UF3; without deleterious chemical reactions of the UF3 with graphite or possibly with other materials within the primary reactor system. 4, Direct measurement of the physical and heat—transfer properties of the DMSR fuel mixture must be made. 5. Further study of the fundamental thermodynamic properties of so- lutes in the LiF-BeF;-ThF,~UF, mixture are needed to ensure that basic understanding of the chemical behavior is accurate. 3.3.2 Fission—product behavior 3.3.2,1 Fission and its consequences Fragments produced on fission of a heavy atom originate in energy states and with ionization levels far from those normally considered in chemical reactions. When the fission occurs in a well-mixed molten-salt liquid medium, these fragments must come tc a steady state* as commonly encountered chemical entities because they quickly lose energy through collisions with the medium. The valence states that these chemical spe- cies assume are presumably defined by the requirements that (1) cation- anion equivalence be maintained in the molten-salt medium and (2) redox equilibria be established between the melt and the surface layers of the container metal.l9,63,8% The fission—product cations must satisfy the fission—-product anions plus the flucride ions released by disappearance % , \ . . . The rapid radioactive decay of many species further complicates an already complex situation. - PR 61 t63 of the fissioned atom. Early assessmen indicated that the cations would prove adequate only if some of them assumed oxidation states cor- 25 strongly supported rosive to Hastelloy-N. A more recent examination this view; these studies indicated that the summation of the products of fission yield and stable valence for each species might be as low as three per figsion event. Accordingly, fission of UFy [releasing 4 F~ + (¢.015 (Br— + I~) per fission] would be intrinsically oxidizing to Hastelloy-N.* Maintenance of a small fraction of the uranium in the fuel aé UF3 was successfully adopted to preclude corrosion from fission of 23SUFu, in the MSRE.1% A properly maintained redox potential in the fuel salt apparently will prevent any untoward immediate consequences of the fission event and will permit grow—in of the fission products in valence states defined by the redox potential. 3.3.2.2 Effects of radiation When fission occurs in a molten flucride solution, both electro- magnetic radiations and particles of very high energy and intensity originate within the fluid. Local overheating is almost certainly not important in a DMSR where turbulent flow causes rapid intimate mixing. Morecover, the bonding in molten fluorides is completely icnic. Such a mixture, with neither covalent bonds to rupture nor a lattice to disrupt, should be quite resistant to radiation. Nevertheless, because there plau- sibly exists a radiation levél sufficiently high to disscciate a molten fluoride into metal and fluorine,T were made.19,20,63,65 Many irradiation tests were conducted prior to 1959 with NaF-ZrF, -UF, a number of tests of the possibility mixtures in Inconel at temperatures at or above 815°C (Refs. 24 and 25) and at quite high fission power densities from 80 to 1000 MW/m3 of fuel. No instability of the fuel system was apparent, and the corrosion did not exceed the considerable amount expected from laboratory-scale tests, *Fission of PuF ; probably would be nearly neutral in this regard, and fission of a mixture of UF, and PuF,, as in the DMSR, would be inter- mediate between these extremes. ' TThis would occur even though the rate of recombination of Li0 and FOC should be extremely rapid. 62 These early tests seemed to show that radiation posed no threat even @ at very high power levels, but further studies®6-63 were conducted primar- ily to test the wetting of graphite by LiF-BeF,-ZrF, -ThF, -UF, mixtures under irradiation. Examination of these capsules after storage at ambi- ent temperatures for many weeks revealed appreciable quantities of CF, and, in most cases, considerable quantities of fluorine in the cover gas. Careful examination20,70,71 strongly suggested that the F, generation had not occurred at the high temperature but had occurred by radioclysis of the mixture in the solid state. This suggestion was confirmed by irradiationlof two arrays of Hastelloy-N capsules, all containing graphite and LiF-BeF,-ZrF,~UF, mix- tures. Two of the capsules in each array had gas inlet and exit lines to permit sampling of the cover gas as desired. Gas samples drawn from the test capsules at operating temperatures and at various power levels up to 80 MW/m3 showed no F, (though an occasional sample from the first array showed detectable traces of CF,). However, during reactor shutdowns with the capsules at about 35°C, pressure rises were observed (usually after an indfiction pericd of a few hours), and F, was evolved. 1In the second array, the capsules were kept hot during reactor shutdown as well as dur- ing operation; no evidence of F, or CF, was observed. Such F, generation at ambient temperatures was subsequently followed for several months in ORNL hot cells. The generation diminished with time in a manner corres— ponding closely with decay of fission-product activity; F, evolution at 35°C corresponded tc about 0.02 molecule per 100 eV absorbed, could be completely stopped by heating to 100°C or above, and could be reduced markedly by chilling to —70°C. The F, evolution resumed, usually after a few hours, when temperature was returned to 35 to 50°C. These and subsequent experiences, including operation of the MSRE, strongly indicate that radiolysis of the molten fuel at reasonable power densities is not a problem. It seems unlikely, though it is possible, that DMSR fuels will evolve F, on cooling. If they do, arrangements must be made for their storage at elevated temperature until a fraction of the decay energy is dissipated. The results of a program of solutien thermodynamics,zs’z6 a long—term 57 and a number of special experiments program of in—-pile irradiations, permitted generally accurate predictions,63“65 but much of our detailed - and still incomplete — understanding of fission-product behavior comes from operation of the MSRE. 19:22,72 he ability of the fission products to form stable compounds and to dissolve in the molten fuel serves to divide them into the three distinct groups described in the following discussion. Noble-gas fission products and tritium. Krypton and Xenon (which is an important neutron absorber) form no compounds under conditions ex- isting in a DMSR or other molten—salt reactor.! 973 Moreover, these gases are only very sparingly soluble in molten fluoride mixtures, /478 As with all noble gases (see Fig. 13),7® their solubility increases with temperature and with diminishing size of the gaseous atom, while the heat of solution increases with increasing atomic size. This low solubility is a distinct advantage because it enables the ready removal of krypton and xenon from the reactor by sparging with helium. The relatively sim— ple sparging system of the MSRE served to remove more than 80% of the 135%a, and far more efficient sparging was proposed for the MSBR.* Stripping of the noble gases from the reactor after a short residence time avoids the presence of their radioactive daughters in the fuel. Tritium qualifies as a fission product beczuse small quantities of it are produced in ternary fissions. However, essentially all of the tritium anticipated in MSBR1%577 results from other sources, as shown in Table 27. A DMSR at similar power level and with a generally similar fuel must be expected to generate tritium at approximately this same rate. This tritium will originate in principle as 3HF; however, with appreciable concentrations of UF3 present, this 3HF will be reduced largely to 3H2. *However, note that the pores in the moderator graphite can offer a haven for these gases. 64 ORNL-LR~DWG 41908 E TEMPERATURE (°C) 800 700 650 600 500 b K, c/p [moles gas/{cc meit) atm] 8 9 10 14 12 £3 14 45 £0,000/7 (°K} Fig. 13. Solubilities of four noble gases as function of temperature in LiF-BeF; (64~36 mole %),76 Solubility of 3H2 in molten salts has not been measured, but the solubil- ity of Hy; in molten LijBeFy, is known’® to be very small.* Some of this 3H2 would be removed, along with krypton and xenon, by sparging with he— lium. However, the extraordinary ability of hydrogen isctopes to diffuse *Solubility of Hy increases with temperature and with pressure of Hy and is near 6 X 1072 moles Hy/m® salt at 700°C. 65 g Table 27. Sources and rates of production of tritium in a 1000-MWe MSBRA Production rate Source MBq/s Ci/d Ternary fission 13 31 6Li(n,o)3H 518 1210 "Li(n,na) 34 501 1170 19%(n,70)3y 4 9 Total 1036 2420 2From Ref. 77. through hot metals will permit a large fraction of the 3H2 to penetrate the primary heat exchanger to enter the secondary cooliant. This phenome- non and its consequences are described briefly in Sect. 3.3.3.2. Fission preducts with soluble stable compounds. Rubidium, cesium, strontium, barium, yttrium, the lanthanides, and zirconium all form quite stable fluorides that are relatively soluble in molten fluoride mixtures such as MSBR and DMSR fuels. Iscotopes of these elements that have no noble—gas precursors, as expected, appeared almost entirely in the circu- lating fuel of the MSRE,19,20,22,72 Very small quantities appeared at or near the surface of exposed graphite specimens; most of this deposition evidently resulted from fission recoil. Isotopes such as 8395r and lqOBa,, whose volatile precursors have appreciable half-lives and which were par- tially stripped from the reactor, were found in samples of the cover gas and within specimens of moderator graphite as well as in the fuel of the MSRE. Along with behavior of other isotopes, Fig. l4 shows the profiles observed for 137Cs and !“OBa in graphite specimens through diffusion of their respective (137Xe,.3.9 ming 1“0Xe, 16 s) precursors. Bromine and iodine would be expected to appear in the fuel as soluble Br and I_, particularly in the case where the fuel contains an appre- ciable concentration of UF3. No analyses for Br were performed during 66 ORNL-DWG 868-1G755R 10 LT ‘ Lo T —— FREE-FLOWING SALT . SURFACE 4017 1046 O-- o 3, H atoms of nuclide / em® of graphite {oiZ o . b . : 1 : ' STAGNANT SALT — ‘ 1 = SURFACE o 0 i0 20 3¢ 40 50 80 70 80 90 100 {40 142G 430 440 {B5C 460 470 180 490 200 20 DEPTH IN GRAPHITE (rmils) Fig. l4. Fission product distribution in CGB (855) graphite specimen exposed in MSRE core during 32,000 MWh of power operatiomn. 67 e operation of the MSRE. Analyses for !3!T ghowed that a large fractiom of the iocdine was present in the fuel20:72 apnd thar 131 deposited on metal or graphite surfaces in the core region. However, material balances for 1317 yere generally low. It is possible that some of the-precursor, 131pg (25 min), was volatilized and sparged with the krypton and xenon. Fur- ther, 1317 produced by decay of 13lra in complex metallic deposits (as in the heat exchanger) may not have been able to return to the salt. Noble and seminoble fission products. Some fission—-product metals (Ge, As, Nb, Mo, Ru, Rh, Pd, Ag, Cd, Sn, and Sb) have fluorides that are unstable toward reduction by fuel mixtures with appreciable concentra- tions of UF3; thus, they must be expected to exist entirely in the ele- mental state in the reactor. Selenium and tellurium were also expected toc be present as elements within the reactor circuit, and this behavior was generally confirmed during operation of the MSRE, 19520,72 The MSRE fuel samples usually contained far less than the generated quantities of these elements. Portions of the MSRE samples were found (probably as metallic particulates) in the helium sparge gas,* deposited on metal sur- faces, and (a reasonably small fraction) deposited on graphite specimens. However, the distribution and especially the inventory in the fuel at the sampling point in the pump bowl showed major variations. Further study will be necessary before details of their behavior can be predicted with confidence for a DMSR. In general, the results from MSRE operations suggest the followingo20 1. The bulk of the noble metals remain accessible in the circulating loop but with widely varying amounts in circulation at any particular time. 2. In spite of this wide variation in the total amount found in a par- ticular sample, the proportional composition is relatively comstant, indicating that the entire inventory is in substantial equilibrium with the new material being produced. *Much-improved gas-sampling techniques used in later stages of MSRE operation showed the fraction carried in the gas to be less than 2% of the quantity produced?? (with the possible exception of 11 * 3% of 103Rru), 68 3. The mobility of the pool of noble-metal material suggests that de- i posits occur as an accumulation of finely divided well-mixed material rather than as a “plate.” Such precipitation within the reactor, though expected, is a dis~- advantage. Precipitation on the metal surface (most of which is in the heat exchanger) will be quite insufficient to impede fuel flow, but radio- active decay of the deposited material contributes tc heat generation dur-— ing reactor shutdown. Precipitation on the moderator graphite, which ap- peared to be considerably smaller than on the metal, would maximize their opportunities to absorb valuable neutrons. Operation of the MSRE did produce one untoward effect of fission products,” Metal sfirfaces exposed to the fuel in the MSRE showed grain boundaries that were embrittled to depths of Q.1 to 0.3 mm {5 to 10 mils), In the heat exchanger, the embrittled boundaries copened to form metallo- graphically visible cracks; in other regions such cracks formed only when the specimens were deliberately strained. Early studies8Y implicated fission-product tellurium as responsible for this embrittlement, and sub- sequent work has confirmed this."®>4® However, more recent studiesh6,81,82 strongly suggest that (1) if the molten fuel is made to contain as much as 5% of the uranium as UF3, the tellurium would be present as TeZ™ and (2} in that form, tellurium is much less aggressive. Much further study will be necessary, but use of this higher but still moderate UF3/UF, ratio apparently will markedly alleviate, and probably control, the tellurium embrittlement problems. 3.3.2.4 Operational constraints Avoiding the detrimental effects of fission-product tellurium (de- scribed immediately preceding) may make necessary the operation of the DMSR with as much as 5% of the uranium fluoride present as UF;. Consid- eration of possible reactions of UFy to produce uranium carbides (de~ scribed previously) suggests that operation with a considerably higher *This subject 1is addressed further in Sect. 3.4. 69 UF3/UF4 ratio would be possible. However, a more subtle comnstraint on DMSR operation possibly may result. The lanthanide trifluorides are only moderately soluble in molten LiF-BeF,-ThF, -UF, mixtures; if more than one such trifluoride is present, they crystallize as a solid solution of all the trifluorides on cooling of a saturated melt. Though not definitive, there is evidence that the actinide trifluorides (including UF3) might also join in such solid solu- tions. If so, the total {(lanthanide plus actinide) trifluorides in the end~of—-1ife reactor might possibly exceed their combined solubility. The solubilities of PuF3 (Ref. 29) and CeF; (Ref. 31) have been care- fully determined in LiF-BeF,-ThF, (72-16-12 mole %) and can be considered to be moderately well known in DMSR fuel. According to Baes et al.,29 the solubility of PuF; in the LiF-BeF,~ThF, melt at 565°C (the minimum temperature anticipated within the DMSR fuel circuit) is 1.35 mole %. The solubility of CeF; under the same conditions appears to be very slightly smaller (1.3 mole %).29,31 Solubilities of the other pertinent fluorides are not well known. In Li,BeF,, the solubilities of several lanthanide trifluorides (including CeF3) have been shown®3 to be considerably smaller and to vary with some more and some less soluble than CeF3. As a reason- able approximation (obvicusly, many additional data are needed), the solubility of the lanthanide-actinide trifluoride solid solution may be assumed to be near 1.3 mole %. From Table 9, the DMSR fuel at end-of-life will contain some 1.404 x 10° moles of uranium isotopes and 3.64 x 103 moles of transuranium iso- topes. The end-of-life inventory of lanthanide plus yttrium isotopes will be near 4.7 x 10% moles. If 5% of the uranium is present as UF3 and if all transuranic and lanthanide species are assumed to be trivalent, the end-of-1ife reactor fuel will contain about 5.77 x 10% moles of trifluo- rides. The DMSR system (with about 5.3 x 106 moles of fluorides), there- fore, would contain about 1.1 mole % of trifluorides. The solubility of the combined trifluorides likely woul& not be exceeded within the reactor circuit, but additional solubility data are needed to make this point certaine. 70 3.3.2.5 Uncertain features From the foregoing discussion, several uncertainties are apparent. Details of behavior of the noble and seminoble fission products are still poorly known. The fractions of each isotope that will appear in the off- gas, deposit on the moderator graphite, and deposit on the heat—exchanger metal of the DMSR can only be crudely estimated. That fraction which ap- pears in the reactor off—-gas would seem to cause no insurmountable prob— lems though that system — at our preéent state of knowledge — would need to be overdesigned substantially. Their deposition in the heat exchanger is a recognized disadvantage and will be quite insufficient to impede fuel flow, but radicactive decay of the depcosited material contributes heat that must be removed during reactor shutdown. Precipitation on the graphite, which appears to be smaller than on heat—exchanger metal, maxi- mizes their opportunities to absorb neutrons. Clearly, a better knowl- edge of this situation is needed. While the results obtained in the recent past are highly encouraging, additional data — especially on a larger scale — are needed to establish the redox potential (UFS/UFQ ratic) required to keep the tellurium crack- ing problem to tolerable levels. Should the required UF3/UF, ratic rise substantially above §.05, the probability of precipitation of trifluoride solid sclutions would be increased. Finally, additional information about the collective solubility be- bavior of the lanthanide-actinide trifluorides is required. Should they prove appreciably less soluble than now believed likely, some replacement of fuel might be required late in the 1ife of the essentially unprocessed DMSR. 3.3.3 TFuel maintenance To achieve fuel maintenance, (1) the fuel must be delivered to and into the reactor in a proper state of purity and homogeneity, {(2) the fuel must be sufficiently protected from extraneous impurities, and (3) sound procedures must exist for addition of the required uranium and provision of the required UF3/UF, ratio. 71 3.3.3.1 Preparation of initial fuel Initial purification procedures for the DMSR present no formidable problems. Nuclear poisons (e.g., boron, cadmium, or lanthanides) are not common contaminants of the constituent raw materials. All the pertinent compoun&s contain at least small amounts of water, and all are readily hydrolyzed to oxides and oxyfluorides at elevated temperatures. The com- pounds LiF and BeF; generally contain a small quantity of sulfur as sul- fate ion. Uranium tetrafluoride commonly contains small amonts of U0,, UFg, and UO,F 5. Purification procedures®3,84,85 ysed to prepare materials for the aircraft reactor experiment (ARE), the MSRE, and many laboratery and en— gineering experiments have treated the mixed materials at high tempera- ture {usually at 600°C) with gaseous H,o-HF mixtures and then with pure Hjp in equipment of nickel or copper. The HF-H, treatment serves to (1) re- duce the U°T and U8t to U“+, (2) reduce sulfate to sulfide and remove it as HyS, (3) remove C1 as HCl, and (4) convert the oxides and oxyfluorides to fluorides. Final treatment with H, serves to reduce FeF3 and FeF,; to insoluble iron and to remove NiF, that may have been produced during hy- drofluorination. To date, all preparations have been performed in batch equipment, but continucus equipment has been partially developed.86987 For a DMSR, as for the MSRE, 85 purification of the bulk of the fuel would presumably be conducted on LiF-BeF;-ThF,-UF, mixtures containing perhaps 85 to 90% of the required UFy and on molten LijUF; to provide the additional uranium necessary to bring the fuel to the critical and oper- ating concentration. Such a purification procedure can provide a sufficiently pure and completely homogeneous fuel material for initial operation of the reactor. 3.3.3.2 Contamination possibilities Though the fuel material can be supplied and introduced into the reactor in sufficiently pure form, contamination of the fuel is possible from several sourcese. Other reactor materials. The moderator graphite can contain a large quantity of COp, €O, and Ho0 by virtue of its porosity and intermal 72 surface. Outgassing of the moderator by pumping at reduced pressure and elevated temperature is necessary and sufficient to prevent contamination of the fuel by oxide ion from reactive gases from this scurce. Oxide films on the structural metal can also contaminate the fuel by oxide ion, and, as described previously, the dissolved Fe3*, Fe2t, and Ni2* can be responsible for subsequent metal corrosion. In operation of the MSRE, the system was flushed with an LiF-BeF, mixture for cleaning at start-up and after each shutdown before introduction of the fuel mixture. This precaution might be unnecessary, but it did suffice to keep oxide contamination caused by surface oxidation of the metal to a minimum. A small (~100-ppm) concentration of Cr2% in the fuel as a consequence of reaction of the metal with the fuel cannot be avoided. However, in the absence of extraneous oxidants, the reaction is very slight, and the pres- ence of Cr2% is completely innocuous. Grow—in of the fission preducts is also unavoidable, as is the pres- ence of a relatively small steady-state concentration of 3H,. Atmospheric contamination. Reaction of the DMSR fuel mixture with oxygen. is relatively slow,* but reaction with water vapor is more rapid. Further, contamination of the fuel with 40 to 50 ppm (by weight) of oxide ion could result in precipitation of a uranium-rich (UTh)C, solid sclu~ tion. A large ingress of contaminant air would be required te produce 40 ppm of 027 in the fuel, and the DMSR would be designed and coperated so as to minimize the chances of such contamination. Operation of MSRE during much of a four-year period with many shutdowns and several minor repair operations showed no evidence of an increase in oxide contamina— tion level.22 Treatment of the initial fuel charge with anhydrous HF-H4 mixture during its preparation reduces the 02” concentration to innocucus lévels, and similar treatment of contaminated fuel would serve tc remove the 027, Such treatment might never be required, but in the DMSR, simple equipment should be included that is capable of treatment to remove oxide ion should inadvertent contamination occur,. *However, the reactor metal at high temperature can readily react with oxygen, and the fuel can react with the oxides formed in this manner. 73 Contamination of fuel by secondary coolant. The only secondary ccol- ant that has been demonstrated in a molten-salt reactor is Li,BeF,, which is prepared from ’LiF and purified through procedures described previously for the fuel used in the MSRE.22 This material is not cofisidered suitable for an MSBR or a DMSR because it is.expensive and its liquidus is too high, Many substitutes have been considered, but none have properties that are all near the ideal. Omn balance, the best choice appears to be a mixture of 8 mole % NaF and 92 mole % NaBF, (Refs. 19, 20, and 88), These compounds are readily available at low cost. The liquidus®? (see Fig. 15), stability toward gamma radiation in the primary heat exchanger,%0 heat—~transfer properties,23’2“991 and compatibility”7»“8 with modified Hastelloy=-N all appear adequate. Intermixing of the fuel and the secondary coolant salts, as caused by leaks in the primary heat exchanger, would be an important considera- tion. The MSBR design8 and presumably the DMSR design assured a slightly higher pressure on the ccolant side so that most leaks would be of coocl- ant into fuel. Such a leak, however small, should be recognized at once ORNL-DWG 67-8423AR iOOO T T T ’ \\,L\\I. i | e LIQUIDUS | s SOLIDUS < L "995°%C [~ 800 4o CRYSTAL INVERSION 1 — 800 f — | — 7\'# | LIQ‘UED = ] | o o | NoF + LIQUID . | i i y 700 | e e NG 5 600 N S S S SR S o ‘ NOBE‘ \\b . &' (HIGH-TEMPERATURE FORM) yoq c & 500 | — ¥ l‘_lQU|D| —————\ == \*— . !- : i 384°C ' - 2\ 400 - LA S ] i i 300 _NaF + NoBF, (HIGH-TEMPERATURE FORM) | | gazc | L L4 L 200 ____NaF + NoBF, (LOW-TEMPERATURE FORM) | NaF 20 40 60 80 NoBF, NaBF, (mole %) Fig. 15. System NaF-NaBFy. 74 because of the marked reactivity loss caused by admission of boron into the fuel. A small quantity of NaF-NaBF, added to the DMSR fuel would allow dissociation of the NaBF, into NaF and BF3. The NaF would dissolve in the fuel and remain as a minor parasitic neutron absorber. The BF3 1is relatively insoluble in the fue192:93 znd would be readily sparged with the krypton and xenon intoc the off—-gas system. A sufficiently small con- tinuing leak could possibly be tolerated with some impairment in system performance,* Given that the leaking tube could be plugged, infrequent small leaks almost certainly would not pose safety problems. Addition of a sufficiently large quantity of NaF-NaBF, could lead to formation of two immiscible liquid phases.93 Such a leak (one capable of adding a2 few tens of percents of coolant to fuel) seems incredible; the presence of the large quantity of boron should certainly preclude reactivity accidents, but the fuel would be ruined. Returning the fuel mixture to some secure site for recovery would be necessary, and a most difficult cleanup and repair of the reactor would be necessary, if possible. Small leaks of coelant inte the fuel system probably pose no safety problems. However, additional study of the mixing of these fluids in realistic geometries and in flowing systems is needed before we can be certain that no potentially damaging situation could arise as a conse- quence of a sudden major failure of the heat exchangerozg The fluoroborate seccondary coolant apparently will contain small quantities of oxygenated species and some species containing hydroxyl 20 These would be capable of precipitating oxides from the fuel if ions, the coolant were mixed with fuel in large amounts, but the effects would be trivial compared with other effects noted previcusly. These substances in the secondary coolant, however, appear to have a nontrivial and bene- ficial effect on DMSR performanceoz This beneficial effect is the appar- ent ability of the secondarv coolant to scavenge tritium and convert it to a recoverable water—-soluble form. As noted earlier, a DMSR must be expected to generate about 1 GBq/s (2500 Ci/d) of tritium, and most of this must be expected to diffuse *Hewever, the off-gas system would have tc be designed to accommo~— date the consequences of BFy admission. 75 through the walls of the primary heat exchanger into the coolant., Early 34 suggested that, unless other mechanisms for tritium retention estimates were provided, as much as 60% of the tritium generated would be lost through the coclant piping to the steam system, from which it would be presumed to escape to the enviromment. Such a loss rate to the environ- ment would be intolerably high. Small-scale studies?®:96 suggested that oxide-bearing and protonated (e.g., BF30H”) species were present in the molten NaF-NaBF, mixture pro- posed as the secondary coolant; the hypothesis was that exchange reactions might offer a mechanism for holdup of tritium in this mixture.®’/ Small- scale experiments®®:%% geemed to show that deuterium diffused through a thin metal tube into such mixtures was retained by the melt but that ex- change with OH was not the responsible mechanism. Though the trapping mechanism remains obscure, more recent testsl0 have confirmed the ability of NaF-NaBF, mixtures to hold up the tritium. An engineering-scale loop, through which the salt could be pumped at 0.05 m3/s (850 gpm), was used. This loop was arranged so that tritium could be introduced by diffusion through thin-walled tubes within the salt; also, the quantities of tritium within the salt, the quantities removed in the gas flow above the free salt surface within the pump bowl, and the quantity diffusing through the loop walls into the cocling air could be determined. During steady-state operation of this device in two tests, each lasting about 60 days, material balance accounted for about 997 of the added tritium.!® About 98% of the added tritium appeared in the ef- fluent gas system of the pump, with more than 907 of this in a chemically combined (water—-soluble) form. The tritium within the salt was essen— tially all chemically combined; the ratio of free tritium to combined tritium was less than 1:4000. Extrapolation of these data to the MSBR coolant system suggests that tritium losses toc the MSBR steam generator could be kept to less than ~4 MBq/s (10 Ci/d).!0 Further studies are clearly necessary; once the mechanism is estab— lished, the performance of the system might be improved. Means for re- plenishment of the active agent must be established, and improved means for recovery and ultimate disposal of the tritium must be developed. 76 3.3.3.3 Fuel maintenance options and methods The initial fuel charge for a DMSR can be prepared in a high state of purity and introduced into the reactor by minor variants of the methods® used for the MSRE® and proposed for the MSBR.® For a once- through DMSR that proposes no chemical reprocessing to remove fission products, the required fuel maintenance operations are relatively few. They include (1) continucus removal (by the sparging and stripping sec- tion of the reactor) of fission-product krypton and xenon, (2) addition of 235y and 2387 to replace that lost by burnup and to keep the fuel suf- ficiently denatured, and (3) in situ production of UF3 to keep the redox potential of the fuel at the desired level; they probably also include (4) removal of inadvertent oxide contaminants from the fuel; in addition, they may include (5) addition of ThF, to replace that lost by transmuta- tion or stored with fuel removed from the operating circuit and (6} re- moval of a portion of the insoluble noble and semincble fissiorn products, Each of these is discussed briefly in the following sections. Continuous removal of fission—-product krypton and xenon. Stripping of krypton and xenon makes possible their continuous removal from the re- gctor circuit by the purely physical means of stripping with helium. For the reference-design MSBR,8 helium flowing at 0.005 m3/s (10 cfm) was to be injected continuously into and withdrawn from fuel-salt side streams carrying a total eof 0.35 m3/s or about 107 of the total fuel flow rate. Some generally similar operation should prove optimal for the DMSR. Such a stripping circuit would remove an appreciable (but not a major) fraction of the tritium and a small (perhaps very small) fraction of the nocble and semincble fission products as gas—borne particulates. In addition, the stripper would remove BFy if leaks of secondary coolant inte the fuel were to occur. None of these removals (except possibly the last) appre- ciably affect the chemical behavior of the fuel system. Addition of fissionable and fertile uranium. Adding ~4,470 kg of 233y and ~18,400 kg of 238y during the lifetime cof the omce-through DMSR *Fuel for the MSRE was prepared in relatively small batches.22,85 If DMSRs were to be of commercial consequence, continuocus purification systems would certainly be devised for initial fuel preparation. 77 will apparently be necessary (see Table 17),* assuming the fuel volume changes from these additions or other causes do not require removal of any fuel to storage. Such additions, which are made over the 30-vear lifetime, would comprise some 30,190 kg (96,332 moles) of UF, added at an average of 1,040 kg (3,320 moles)/year. During operation,?2 many on-stream additions of fissionable material as molten 7Li3UF7 were made to the MSRE, and this method of addition can obviously be used as a clean, convenient way to add the uranium to a DMSR. Using this method of addition would require use of 2.89 x 10° moles (7514 kg) of 7LiF containing 2013 kg of 7Li. This represents about 6.8% of the /Li in the original fuel inventory and would result in appreciable volume increase (especially if BeF, were added proportionally} in the fuel.t During the course of reactor operation, removing some fuel to storage within the reactor complex would probably be necessary if this addition procedure were used. Developing and demonstrating methods of addition of selid UF, (or proper mixtures of UF, plus UF3) should be possible. These will be in- herently more complex (and radiocactively dirty), and stating which of the options would be preferred is not presently possible. Maintaining the desired UF3/UFy ratio. Operation of the MSREL 9,22 demonstrated that in situ production of UF3 could be accomplished readily and conveniently by permitting the circulating fuel to react in the pump bowl with a rod of metallic beryllium suspended in a cage of Hastelloy—N. This technique could be adapted for use in a DMSR; beryllium reduction would be desirable if the fissionable and fertile uranium additions are to be made as 7L13UF7,* 5 _ Additions would be made as depleted material or as material contain- ing not more than 20% of 235U in 238U; the materials would be added as mixed fluorides or possibly as UF, plus UFj. TDevelOping and demonstrating a method of addition of a high-uranium- content liquid with considerably less LiF than Li3UF7 are almost certainly possible. *In that event, adding BeFp will be necessary to preserve the LiF/ BeF, ratio in the fuel at approximately its initial value. 78 ‘The original charge of fuel contains some 7.35 x 10% moles of ura- L niwm. If all this were supplied as UF,, about 1840 moles (16.54 kg) of Be0 would be required to reduce 5% of it to UF3.* While the initial preparation procedure could be modified so that some of the UF; would be present as the fuel was delivered, in situ production likely would be more convenient. The‘delivered fuel could be made slightly deficient in BeF, to accommodate that generated by UF3 production. - As indicated previously, the fission process occurring with UF, is significantly oxidizing. During the 30-year reactor lifetime (22.5 full- power years assumed) with 70% of the fissions occurring in uranium iso- topes, nearly 6.12 x 10% moles of uranium will have been fissioned. If the fissioning uranium is 95% UF,, as much as 5.8 x 10" moles of UF 3 might be oxidized [at a generally uniform rate of 7 moles (~2.1 kg) per full- power day] during the reactor lifetime. Its reduction would require some | 2.9 x 10" moles (261 kg) of metallic beryllium. The BeF, produced in that manner represents about 3.0% of that present in the original fuel charge. Some additional reduction of UF, to UF3 will be required if the fuel must be treated to remove oxide ion. Accomplishing the reduction of UF, to UF3 in situ would certainly seem feasible by using metallic uranium in place of berylliium. Should the decision be made to add the fissionable and fertile uranium as UF,, reduction performance by use of uranium would have the advantage of not appreciably diluting the fuel. Removal of inadvertent oxide contamination. Treatment of complex molten fluorides with anhydrous HF-H; mixtures has been used commonly to reduce the oxide concentration to completely innocuous levels.8%: 8% No real doubt exists that such treatment could be used if required for puri- fication of DMSR fuel mixtures. However, there is little basis to assess the necessity of such purification. Operation of the MSRE during a four- year period with many shutdowns and several minor repair operations showed no evidence of oxide contamination. 1In early versions, equipment should be included in which HF-H» mixtures and then Hy could be used tc remove *aAn additional 9.63 x 10% moles of uranium tetrafluoride will be added during the reactor lifetime. Reduction of 5% of this will require an additional 2.41 x 10% moles (21.7 kg) of berylliium. 79 S such contamination. For a demonstration reactor, this relatively simple equipment probably should be sized to permit treatment of the fuel on a 300-d cycle; if it were pessimistically assumed that 3.3 d would be re- quired to process a batch, the equipment should be sized to accommodate 1% of the fuel charge (~1 mg)a Some fission products would be affected by this treatment; iodine, in particular, would be evolved and would have tc be managed in the off- gas, Selenium and tellurium (if they are scluble as Se?” and Te?™ in the molten fuel) might alsc evolve. Oxidation of Patt to Pa®t could be avoided by inclusion of a few percent of Hy with the HF. However, oxidation of a large fraction of the UF3 to UF, would re- sult unless the HF-Ho mixture contained so large a fraction of H; that it would be relatively inefficient at oxide removal. Accordingly, to allow for additional (beryllium or uranium) reduction of this UF, would be nec- essary to maintain the desired UF3/UF, ratio. For example, in the unlikely event that the fuel must be treated for oxide removal each 1000 full-power days, the inventory would require treatment 8.2 times during the reactor lifetime, The inventory of ura- nium isotopes (see Table 9) increases regularly during the reactor life- time and may average 1.07 x 10° moles during the 30 years. If (as is not true) all the UF3 were oxidized each time and if 5% of the uranium inventory were to be reduced, some 2.2 x 10% moles of Be® would be re- quired during the reactor lifetime. This, when added to the 2.9 x 10% moles of beryllium estimated previculsy to be required to overcome the oxidative effect of uranium fission, would total some 5.1 x 10" moles of BeF, generated or near 5.4%7 of the BeF, in the original feed. This added BeF,, though added at a slowly increasing rate during reactor life, is a good match for the 6.8% of 7LiF needed to add the uranium as LiUF;. A perfect match of LiF and BeF, additions is certainly not required; the maintenance processes briefly indicated sbove might provide a sufficiently good addition rate for LiF and BeF,. Possible addition of thorium, If making a few additions of thorium te the reactor fuel during its lifetime is necessary, then adding it as a liquid containing 7LiF and ThF, should be pessible. A possibility would be a melt containing about 70 mole % LiF and 30 mole % ThF, melting near 80 600°C (see Fig, 11). Alternatively, a procedure presumably could be de- veloped for addition of solid ThF,. Partial removal of noble and seminoble metals. The behavior of these insoluble fission-product species, as indicated previously, is not under- stood in detail. If they precipitate as adherent deposits on the DMSR heat exchanger, they would cause no particularly difficult problems. How- ever, should they form only loosely adherent deposits that break away and circulate with the fuel, they would be responsible for appreciable para- sitic neutron captures. If these species were to deposit on the moderator graphite, they would constitute an even worse neutronic situation. To the extent that they circulate as particulate material in the fuel, insocluble fission—-product species could probably be usefully re- moved by a small bypass flow through a relatively simple Hastelloy-wool filter system. Presumably, such a system would need to have a reasonably low pressure drop and probably would need to consist of sections in paral~- lel so that units whose capacity was exhausted could be reasonably re- placed. 3.3.3.4 Summary, constraints, and uncertainties Very likely, a number of options for fuel maintenance are available. Some of these have been demonstrated and others could be made available if tfiere were good reasons why they were needed. Several uncertainties also exist. Presently, we do not know whether (1) treatment to remove inadvertent contamination by oxide will be neces- sary, (2) addition of uranium to the DMSR fuel will be done by use of 7Li3UF7, (3) the oxidative effect of fission is near 1 oxidative equiva- lent per mole of uranium fissioned, or (4) the removal of nobie and semi- noble metals from the DMSR fuel is necessary or desirable. Should they prove desirable, a relatively large number of options could be made available. A great amount of further optimization of the fuel cycle for DMSR will be required before we know which, if any, of these options are necessary or desirable. S 81 i 3.4 Reactor Materials Although special, high~quality materials probably would be used . throughout in the comnstruction of a DMSR, most of them could be cobtained from commerical sources that routinely supply such materials using cur- rently available technology. Two notable exceptions to this generaliza- tion are the structural alley that would have to be used for components normally exposed to molten salt and the graphite for the reactor core moderator and reflector. Both of these materials would require specifi- cations peculiar to the MSR system. 34401 Structural alloy 3.4.1.,1 Requirements The metallic structural material used in constructing the primary circuit of a molten—-salt reactor will operate at temperatures up to about 700°C., The inside of the circuit will be exposed to salt that contains fission products and will receive a maximum thermal fluence of about 1 X 1025 neutrons/m? over the operating lifetime of about 30 years. This fluence will cause some embrittlement because of helium formed by trans- mutation but will not cause swelling such as is noted at higher fast flu- ences. The outside of the primary circuit will be exposed to nitrogen that contains sufficient air from inleakage to make it oxidizing to the metal. Thus, the metal must (1) have moderate oxidation resistance, (2) resist corrosion by the salt, and (3) resist severe embrittlement by thermal neutrons. In the secondary circuit, the metal will be exposed to the coolant salt under much the same conditions described for the primary circuit. The main differences will be the lack of fission products and uranium in the coolant salt and much lower neutron fluences. This material must have moderate oxidation resistance and must resist corrosion by a salt not con- taining fission products or uranium, The primary and secondary circuits involve numerous structural shapes ranging from several centimeters thick to tubing having wall thicknesses of only a millimeter or so. These shapes must be fabricated and joined 82 (primarily by welding) into an integral engineering structure. The struc- ture must be designed and built by techniques approved by the American Society of Mechanical Engineers {(ASME) Boiler and Pressure Vessel Code. 3.4.1.2 Status of development Early materials studies led to the development of a nickel-base al- loy, Hastelloy-N, for use with fluoride salts. As shown in Table 28, the alloy contained 167 molybdenum for strengthening and chromium sufficient to impart moderate oxidation resistance in air but not enough to lead to high corrosion rates in salt. This alloy was the sole structural material Table 28. Chemical composition of Hastelloy-N Content of alloy o\ Element (wt Z) Standard Modified Nickel Base Base Molybdenum 15~18 11—13 Chromium -8 6—8 b Iron 5 0.1 b Manganese 1 0.16-0.25 Silicon 1 0.1 Phosphoxrus 0.0L5 0.01 Sulfur 0,020 0.01 Boron 0.01%.- 0.001 Titanium Niobium 1-2 aSingle values are maximum amounts allowed. The actual con- centrations of these elements in an alloy can be much lower. bThese elements are not felt to be very important. Alloys are now being purchased with the small concentrations specified, but the specification may be changed in the future to allow a higher con— centration. 83 used in the MSRE and contributed significantly to the success of the ex- periment. However, two problems were noted with Hastelloy-N which needed further attention before more advanced reactors could be built. First, Hastelloy-N was found tc be embrittled by helium produced directly from traces of 198 and indirectly from nickel by a two-step reaction. This type of radiation embrittlement is common to most iron— and nickel-base alloys. The second problem arose from the fission-product tellurium d4if- fusing a short distance into the metal along the grain boundaries and em- brittling the boundaries. Considerable success was encountered in modifying the composition of Hastelloy=N to obtain better resistance to embrittlement by irradiation. The key factor was to modify the carbide precipitate from the coarse type found in standard Hastelloy-N to a very fine type. The presence of 16% melybdenum and CG,57% silicon led to the formation of a coarse carbide that had little benefit. Reduction of the molybdenum concentration to 12% and the silicon content to 0.17 and addition of a reactive carbide former such as titanium or niobium led to the formation of a fine carbide precipitate and an alloy with good resistance to embrittlement by helium. Consider- able progress was made in the scale-up of an alloy containing 27 titanium, but this alloy does not have sufficient resistance to intergranular crack- ing by tellurium. An alloy containing ! to 2% niobium was noted to be very resistant to cracking by tellurium and was produced in small commer- cial melts. The composition of the niobium—modified alloy is shown in Table 28. This alloy maintains good ductility up to the 40-ppm maximum helium content anticipated in the wall of a molten—salt reactor vessel. In studying the tellurium embrittlement problem, considerable effort was spent in seeking better methods of exposing test specimens to tellu- rium. In the MSRE, the flux of the tellurium atoms reaching the metal was about 1013 atoms m~—? s"l, and this value would be 10l% atoms m™2 s~! for a high-performance breeder. Even the value for a high-performance breeder is very small from the experimental standpoint. For example, this flux would require that a total of 7.6 x 107® g of tellurium be transferred to a sample having a surface area of 10 ecm? in 1000 h. Electrochemical probes were immersed directly in salt melts known to contain tellurium, 84 gt and there was never any evidence of a soluble telluride species. However, e considerable evidence existed that tellurium "moved” through salt from one point to another in a salit system. The hypothesis was that the tellurium actually moved as a low-pressure pure metal vapor and not as a reacted species, The most representative experimental system developed for ex-— posing metal specimens to telluriuwm involved suspending the specimens in a stirred vessel of salt with granules of Cr3Te, and CrgTeg lying at the bottom of the salt. A very low partial pressure of tellurium was in equi- i1ibrium with the CryTe, and CrgTeg, which resulted in Hastelloy-N speci- mens with crack severities similar to those noted in samples from the MSRE. Numerous samples were exposed to salt that contained tellurium, and the most important finding was that modified Hastelloy-N containing ! to 2% niobium had good resistance to embrittlement by tellurium {Fig. 16). One series of experiments was run to investigate the effects of the oxidation state of tellurium—containing salt on the tendency for cracks to be formed. The supposition being examined was that the salt might be made reducing enocugh to tie up the tellurium in some innocuous metal com— plex. The salt was made more oxidizing by adding NiF, and more reducing by adding elemental beryllium. The experiment had electrochémical probes for determining the ratio of uranium in the +4 state (UF,) to that in the +3 state (UF3) as an indicator of the oxidation state of the salt, Ten— sile specimens of standard Hastelloy-N were suspended in the salt for about 260 h at 700°C. The oxidation state of the salt was stabilized, and the specimens were inserted so that each set of specimens was exposed to one condition. After exposure, the specimens were strained to failure and were examined metallographically to determine the extent of cracking. The results of measurements at several oxidation states are shown in Fig. 17. At U**/U3T ratios of 60 or less, very little cracking occurred, and at ratios above 80, the cracking was extensive. These observations offer encouragement that a reactor could be operated in a chemical regime where the tellurium would not be embrittling even to standard Hastelloy-N. At least 1.6%Z of the uranium would need to be in the +3 oxidation state {(UF3}, and this condition seems quite reasonable from chemical and prac- tical considerations. 85 ORNL-0OWG 77-322¢ 8000 70C0 4000 O] Q o O CRACK FREQUENCY x CRACK DEPTH {number/cm) 2000 1000 LX) o) Nb CONTENT (%) Fig. 16, Variatiocns of severity of cracking with Nb content. Sam- ples were exposed for indicated times to salt containing Cr3Te, and CrgTeg at 700°C. 86 ORNL--DWG 77—-4680A | I | 1 I 9500 |- REDUCING OXIDIZING A e & CRACKING 600 PARAME TER [FREQUENCY {cm ") X AVG. DEPTH {um)] — 300 — o o Legm———t——t—7" | | Lo 10 20 40 70 100 200 400 SALT OXIDATION POTENTIAL [U{IV)/U(IIL)] Fig. 17. Cracking behavior of Hastelloy-N exposed 260 h at 700°C to MSBR fuel salt containing Cr3Te, and CrgTege. Presently, the modified alloy composition shown in Table 28 is fa- vored. Considerable progress had been made in establishing test methods for evaluating a material's resistance to embrittlement by tellurium. Modified Hastelloy-N containing from 1 to 2% niobium was found to offer improved resistance to embrittlement by tellurium, but the test condi- tions were not sufficiently long or diversified to show that the alloy totally resisted embrittlement. One irradiation experiment showed that the niobiummodified alloy offered adequate resistance to irradiation embrittlement, but more detailed tests are needed. Several small melts containing up to 4.47% niobium were found to fabricate and weld well, so products containing 1 to 2% niobium likely can be produced with a minimum of scale-up difficulties, 3.4.1.3 Uncertainties Although no basic scale-up problems are anticipated with the niobium— modified alloy, several large heats must be melted and processed into structural shapes to show that the alloy can be produced commercially. A further need exists for longer expeosure of this alloy to irradiation e 87 and to salt containing tellurium to show that it will resist embrittle- ment by these two processes over long periods of time. Numerous mechan— ical property tests must be run on the new alloy to develop the data needed for ASME Boiler and Pressure Vessel Code approval of the alloy and to establish adequate design methods. 3.4.2 Moderator 3.4.2.1 Requirements The graphite in a single-fluid molten—salt reactor serves no struc- tural purpose other than to define the flow patterns of the salt and, of course, to support its own weight. The requirements on the material are dictated most strongly by nuclear considerations, that is, stability of the material against radiation—induced distortion and nomnpenetrability by the fuel-~bearing molten salt. Practical limitations of meeting these re-— quirements impose conditions on the core design — specifically the neces- sity to limit the cross—sectional area of the graphite prisms. The re- quirements of purity and impermeability to salt are easily met by several high-guality fine-grained graphites, and the main problems arise from the requirement of stability against radiation—-induced distortion. 3.4.2.2 Status of development The dimensional changes of graphite during irradiation have been studied for a number of years. The dimensional changes largely depend on the degree of crystalline isotropy, but the volume changes fall into a rather consistent pattern. As shown in Fig. 18, a period of densifica- tion occurs first during which the volume decreases, and a period of swell- ing then occurs in which the volume increases. The first perioed is of concern only because of the dimensional changes that take place, and the second period is of concern because of the dimensional change and the formation of cracks. The formation of cracks would eventually allow salt to penetrate the graphite., Data shown in Fig. 18 are for 715°C, and the damage rate increases with increasing temperature. Thus, the graphite section size should be kept small enough to prevent temperatures in the graphite from greatly exceeding those in the salt. 88 8 ABS RY-12-29 | - YMI3 4 - ! ROB-C / [yMz50-10 O AXM-ITX~ j VOLUME CHANGE, 100 In (H—AE//E/O) f IS Fig. 18, 715°C., 10 20 (x102Y) FLUENCE [neutrons/cm? (£ >50 keV)] Volume changes for conventional graphites irradiated at 89 g | With the different objectives of nonproliferating MSRs, the regquire- ments for the graphite have diminished from those of the high-performance breeder., First, the peak neutron flux in the core can be reduced to levels such that the graphite will last for the lifetime of the reactor plant. Second, both the low power density and the low rate of xenon mass transfer to the graphite tend to limit the xenon poisoning effect in this reactor so that sealing the graphite may not be necessary. The lessened gas permeability requirements also mean that the graphite can be irradi- ated to higher fluences (Figs. 18 and 19). The lifetime criterion adopted for the breeder was a damage fluence of about 3 x 102% peutrons/m?. This ORNL-DWG 71-69§{5R2 | / / l/ 8 , | | / £ /e » eal 1 Ly ) 3 | //,H~364i/ * o o AXF | | éé;:* o ////E////k/ 5 ,&3\ 4 ] \ ™M a AXF~UFG o AXM apP-03 -4 o HL 18 ® H-385 -8 | | 2 0 10 20 30 (x40} FLUENCE [neutrons/cm?2 (£ >50 keV)] Fig. 19. Volume changes for monolithic graphites irradiated at 715°C, 90 was estimated to be the fluence at which the graphite structure would con- tain sufficient cracks to be permeable to xenon., Experience has shown that, even at volume changes of about 10%, the graphite is not cracked but is uniformly dilated. For nonproliferating devices, xenon perme- ability will not be of as much concern, and the limit probably will be established by the formation of cracks sufficiently large for salt in- trusion. The GLCC* H~364 graphite likely could be used to 3 X 1026 neutrons/m?, and improved graphites with a limit of 4 x 102° peutrons/m? could be developed. The specific performance requirements for graphite suitable for the reactor design presented in this report are a lifetime fluence capability of 2,7 x 1026 peutrons/m? (E > 50 keV) at a peak temperature of 750°C. Most probably, existing commercial graphites will satisfy this need. 3:4.2.3 Uncertainties Although existing commercial graphites likely will meet the needs of the present design, graphite samples having the same cross section as the reference—design moderator elements need to be irradiated. These tests need to be run to the destruction of the graphite to determine the . point at which the graphite actually heals. This will define failure in -the present concept. Physical properties, particularly thermal conduc- tivity, need to be measured as a function of fluence. A longer-range effort to develop improved graphite for future re- actors should be initiated. Early efforts show promise that graphites with improved dimensional stabiliity can be developed. 3.5 8Safety Considerations The main feature of the DMSR which sets it apart from solid-fuel re- actor types is that the nuclear fuel is in fluid form (moltem fluoride salt) and is circulated throughout the primary coolant system, becoming critical only in the graphite-moderated core. Possible problems and en- gineered safety features asscociated with this type of reactor will be *Great Lakes Carbon Company. ...... 91 quite different from those of the present LWR and liquid-metal fast breeder reactor (LMFBR) designs. A detailed safety analysis of the DMSR must await the results of a research and development (R&D) program; how- ever, identifying possible generic problem areas and some of the advan— tages and disadvantages of this concept is already possible. In the DMSR, the primary sysfiem fluid serves the dual role of being the medium in which heat is generated within the reactor core and the me- dium that transfers heat from the core to the primary heat exchangerse Thus, the entire primary system will be subject to both high temperatures (700°C at the core exit) and high levels of radiation by a fluid contain- ing most of the daughter products of the fission process. Because of the jow fuel=-salt vapor pressure, however, the primary system design pressure will be low, as in an LMFBR. 1In terms of level of confinement, the entire reactor primary system is analogous to the fuel cladding in a solid—-fuel reactor. Although much larger, it will not be subject to the rapid ther- mal transients (with melting) associated with LWR and LMFBR accident sce=- narios. Two additional levels of confinement will be provided in the DMSR, in accord with present practice. Note that the cnce-through BMSR concept has safety advantages over the break—even DMSR because a large and complex part of the primary contaimment — the chemical reprocessing plant — is substantially reduced and because less radicactive material is routinely removed from containment. The problem of developing a reactor primary system that will be reliable, maintainable (under remote condi- tions), inspectable, and structurally sound over the plant’s 30-year life- time will probably be the key factor in demonstrating ultimate safety and licensability. The breach of the reactor primary system boundary, resulting in a spill of highly radiocactive salt into the primary containment, will prob- ably provide the design—-basis accident. The analogous event in a soclid- fuel reactor would be major cladding failure. Possible initiators of this accident include pipe failure, missiles, and pressure or temperature tran— sients in the primary salt system. Failure of the boundary between the primary and secondary salt in the primary heat exchangers could be espe- cially damaging. In the event of salt spill, a possibly redundant sys— tem of drains would be activated to channel the salt to the continuously 92 cooled drain tank. The system primary containment, which is defined as e the set of hermetically sealed concrete-shielded equipment cells, would probably not be threatened by such a spill, but cleanup operations would be difficult. : A unique safety feature of the DMSR is that, under accident shutdown conditions; the fuel material would be led to the emergency core cooling system (ECCS) {represented by drain tank cooling) rather than vice versa. The reactor and containment must be designed so that the decay-heated fuel salt reaches the drain tank under any credible accident conditions. In any case, the decay heat is associated with a very large mass of fuel salt so that melt~through (or "China Syndrome”) is apparently not a problem. The safety philosophy for accidents invelving the reactor core is very different for fluid-fueled than for solid-fueled reactors because the hedt source is mainly in the liquid-fuel salt and not in a solid, which requires continuous cooling to avoid melting. An LMFBR, for exam- ple, has a large amount of stored energy (which must be removed under any accident conditions) in the fuel pins. Dryout, which means immediate meltdown in an LMFBR, would not be nearly as severe in the DMSR because the heat source is removed along with the cooling capability. First—order analysis has shown that a fiow blockage of a central coolant channel of the reference DMSR which reduces the flow to less than ~20% of nominal will probably result in local voiding of that channel. This was not true of the old MSBR désign8 because the channels were more strongly thermally coupled. Whether the safety implications of this will lead to modifica- tions of the DMSR reference design must be shown by future safety analy- sis studies. Under any off-normal conditions, the fuel salt will be chan- neled to the drain tank, which must have reliable systems for decay heat removal. No credible means exists for achieving recriticality once the fuel salt has left the graphite-moderated core. 3.6 Environmental Considerations There are no significant differences in the environmental effects of routine operations between an MSR and reactors presently in commercial operation. Neo gaseous or liquid radioactive effluent discharge occurs 93 during normal operation. Minor amounts of such effluents may result from maintenance operations involving opening the primary system. The MSR {(along with the HTGR an& the LMFBR) is in the class of re- actors which operates at high temperatures and high thermal efficiencies — about 40% compared with about 32% for LWRs. For the same electrical ca- pacity, these more efficient reactors reject about 40%Z less heat to the enviromment. This can reduce impacts such as consumptive use of water re- sources, atmospheric effects, and effects on aquatic life., In the reference DMSR concept, neither the nuclear fuel nor the fis- sion products {except for the volatiles, including xenon) are removed from the primary system during the reactor lifetime. This eliminates a major envirommental problem of present day LWRs: frequent transportation of highly radicactive spent fuel from the reactor site to the reprocessing/ storage facility. Most radioactive material remains within the DMSR pri- mary contaimment for the 30~year reactor lifetime but must be dealt with at end~of-life. Uranium, lithium, and possibly other valuable elements will probably be recovered for reuse, but the remainder, which contains the actinides americium and curium (not found in significant amcunts in spent LWR fuel), will have to be disposed of. Decommissioning the plant may be more difficult than for an LWR because the entire primary circuit will be intensely radioactive. A large amcunt of tritium is generated in MSRs as a result of neutron reactions with the lithium in the fuel salt. Tritium is known to diffuse through metal walls such as heat—exchanger tubes, thus providing a poten— tial route for transport of gaseous tritium through the secondary cocolant loop to the steam generators. Recent experiments have shown that tritium is oxidized in the secondary coolant (sodium fluoroborate), which blocks further transport of tritium. The release of tritium from MSRs to the en- viromment is estimated to be no greater than from LWRs and is well within NRC guidelines. A power economy in which the MSR plays an important role would re- quire large quantities of lithium, beryllium, fluorine (for the fuel-salt mixzture), nickel (which comprises 78% of the Hastelloy-N), and graphite (moderator elements), The environmental effects of obtaining, using, and disposing of these materials would certainly have to be evaluated. 94 3.7 Antiproliferation Features A major feature of the DMSR is the relative unavailability of special nuclear material (SNM) that might be diverted and converted int¢ strategic special nuclear material (SSNM) for use in the production of nuclear ex- plosive devices. Because all the fuel would be in a homogeneous fluid, there would never be any subunits (e.g., fuel elements) that would be par- ticularly enriched in a given "desirable” material or depleted with re— spect to specific contaminants. In addition, because the initial fuel charge as well as all makeup fuel would be denatured 235U and because "spent” fuel would not be removed from the primary containment except dur- ing decommissioning at the end of reactor life, the accessibility of even the mixed fuel would be severely restricted. Postulating ways of obtain— ing SSNM from any mixture containing fissile nuclides is always possible, but, in the case of the DMSR, these appear to involve special difficulties as well as low productivity. 3.7.1 Potential sources of SSNM After the first few years of power Operation,* the principal fissile nuclide in a DMSR would be 233y with a substantial amount of 235y, How- ever, both nuclides would remain fully denatured during the entire opera- tion. Thus, after diversion and separation from other chemical species (many of which would be highly radicactive), the fissile uranium would still have to be subjected to an isotope enrichment process to produce SSNM. Other isotopic contaminants in the uranium, notably 232U, would tend to make this a difficult approach. | The next most abundant fissile material in DMSR fuel salt would be plutonium, with a maximum total-plant inventory (at end of plant life) of 334 kg of 3%y + 241 242 of Pu. However, this material would also contain 182 kg Pu and 139 kg of Zquu, which would tend to detract from its value *Initially, the dominant fissile nuclel would be 2350 denatured with 238U, but because this mixture presumably would be an item of inter- national commerce, the DMSR would not represent a particularly attractive source of supply. 95 E as SSNM. A more attractive isotopic mixture would exist early in the plant lifetime {(e.g., after one vear of operation), but the total inven- tory would be much smaller — only 86 kg of 23%py + 24lpy with 13 kg of ZHOPu + 242Pu. Another potential source of SSNM in a DMSR would be 233p3, This nu- clide would have its maximum inventory of ~63 kg early in the reactor life and slowly decline to about 41 kg at the end of life. In principle, this nuclide, if it could be cleanly and rapidly separated from the rest 233y of the fuel salt, could provide an equivalent amount of high-purity through simple radiocactive decay. 3.7.2 Accessibility of SSNM A major consideration regarding the accessibility of various forms of potential SSNM in a DMSR is that all the materials are intimately mixed with ~350 Mg of highly radiocactive fuel salt with no known method for simple physical separation. Thus, diversion of only a modest amount (a few kilograms of SSNM without plans for isotopic enrichment would require the removal of a number of tons of fuel salt from the reactor system. The need for such large‘(and otherwise unjustifiable) salt remcvals,* which without replacement would shut down the reactor, coupled with the need for an elaborate chemical treatment facility to isolate the product, appears to make this approach relatively impractical. In principal, pure 233y could be diverted via the 233Pa route by modifying the in-plant hydrofluorinator to permit its use as a fluori- nator. This would require two fluorinations of each batech of salt, with one occurring immediately after removal of the salt from the reactor to strip cut the denatured uranium and a second about two months later to recover the 233y produced by 233py decay. However, if the system were originally designed to handle batches of salt no larger than ~1 m3, the *Presumably, this approach would be used only to divert plutonium because uranium diversion would require isotopic enrichment and 233py diversion would encounter serious timing problems, as well as requiring the handling of more salt. 96 233y production capability would be less than 3 kg/year, which seems im- practically low. Although the removal of fissile material from a DMSR may be awkward, if 1t could be accomplished without removing large quantities of salt, then the removal could be easily concealed by additions of denatured 235y to the fuel salt. The change in total uranium concentration would not become significant until after the exchange of a few tens of kilograms of fissile fuel, Although a much more detailed, quantitative analysis that considered the relative values of various forms of SSNM would be required to permit a comprehensive assessment of the proliferation sensitivity of the once- through DMSR, this general treatment suggests that this concept may com— pare favorably with other altermatives in terms of resistance to prolif- eration of nuclear explosives. 97 4. ALTERNATIVE DMSR CONCEPTS Of the several MSR concepts that have been considered, the DMSR de- scribed in the preceding section was judged to be the one most firmly based on currently available technology. However, it is not the only proliferation~-resistant MSR concept that could be considered. However, because a high level of proliferation resistance in an MSR apparently re- guires denatured fuel, which imposes some design restrictions, the major differences among the alternate concepts involve the fuel cycle. 4,1 Fuel Cycle Choices Possibly the most favorable fuel cycle for any DMSR, at least from the point of resource utilization, would be one with break—even breeding performance. Calculations for a DMSR core without neutron flux flatten- ing to extend the life expectancy cf the graphite moderator showed® that break—even breeding was marginally possible with full-scale fission- product treatment of the fuel using a reductive—extraction/metal-transfer process100 similar to that proposed for the MSBR. Even if break-even performance were not attained, the initial fuel change could be "used” for several reactor plant lifetimes by feeding moderate amounts of fis- sile fuel. The next step downward in performance might be a concept involving treatment of the fuel for partial fission—product removal by chemical operations significantly different from the reference process. This ap- proach probably would lead to still lower conversion ratios, but it might permit internal recycle of the fuel through a few generations of reactors and, therefore, offer better resource utilization than the once-through fuel cycle. Some improvement in fuel utilization over current-technology LWRs could be achieved even without on—site chemical treatment for fission- product removal. Periodic replacement of the fuel carrier salt (after re- covery and return of only the uranium) with material that is free of fis- sion products and higher actinides would improve the utilization of fis~ sile fuel, though it would increase the consumption of other fuel-salt constituents. 98 All the MSR options from the breeder through the simplest comverter would take advantage of the fact that the noble—gas fission products (including the very important nuclear poison 135Xe) are very sparingly soluble in the molten fluoride fuel. Thus, they would all use simple stripping with gaseous helium to remove krypton and xenon from the pri- mary system. In addition, they would all take advantage of the fact that the noble-metal and seminoble-metal fission products do not form stable fluorides in the fuel and would precipitate as elemental species, pri- marily on metal surfaces outside the reactor core. 4.,1.1 Break-even breeding The presence of 238y ip a DMSR, combined with the effects of flux filattening, sufficiently reduces the nuclear performance so that a net breeding ratio substantially greater than 1,0 probably could not be achieved, even with full-scale fission-product processing. (A positive breeding gain presumably would be undesirable in a proliferation-resistant system because it would require the periodic "export” of excess fissile material.,) However, the studies that have been carried out indicate that break—even breeding is within the uncertainty limits of the neutronic cal- culations for a flux-flattened DMSR core with a 30-year moderator life ex- pectancy. Extended operation at break—even would require z carefully op- timized core design as well as continuous fuel-salt processing on a rela— tively short time cycle (~20 d) to remove fission products and retain {(or return) all fissile and higher—actinide nuclides. The reference fuel processing concept proposed for the MSBR could not be directly applied to a DMSR for several reasons. 1. Isolation of 233Pa would not be acceptable in a DMSR because its decay would lead to a supply of diversion-sensitive, high-purity 233y, 2. Isolation of protactinium would be accompanied by removal and loss of plutoniwm from the operating system. This would not only degrade sys— tem performance but also provide a source of plutonium that would have to be safeguarded and/or disposed of. 99 S 3. The reference system without protactinium isolation would have no means for removing fission-product zirconium, which would then reach undesirably high chemical concentrations. However, the reference process could be modified to meet the regquirements of the DMSR concept. A modified process (described in Ref. 9), in addi- tion to providing the required fission-product removal, would offer other advantages. 1. The tetal plutonium inventory would be limited because the plutonium would eventually be consumed at its production rate in the reactor. 2. The reactor would serve as its own "incinerator” for‘transplutonium actinides, which would be continuously recycled in the fuel. 3. Neither the protactinium nor the plutonium would ever be isolated from all other highly radicactive species. This modified processing concept would use all the basic unit operations proposed for the MSBR system in essentially the same sequence. However, additional, though similar, process steps would be required to remove zirconium on a reasonable time scale, and these are included in the con- ceptual flow sheet, Some removal of neptunium also might be desirable to avoid the long—term poisoning effects of 237Np and 238Pu; this probably could be included without adding significantly to the complexity of the processing facility. With full-scale fission~product removal and break—even breeding, the fuel in a DMSR could be used indefinitely. That is, at the end—ocf-life of - one reactor plant, the fuel salt could be transferred to a new plant and used without any significant intermediate treatment., During the life of any given plant, adding thorium as the principal fertile material and 238y to maintain compliance with denaturing requirements would be necessary, but no fissile additions would be required. Other routine removals of fuel-carrier salt (LiF + BeF2 + ThFy) and additions of BeF; and Tth* would be required to maintain the desired chemical composition of the salt. The removed carrier salt could be disposed of (after conversion to a suitable form) or chemically processed for recycle into other MSRs. *Lithium fluoride would be formed continuously in the salt from the lithium used in the reductive-extraction/metal—transfer steps. 100 4,1.2 Converter operation with fuel processing g Because the results of the currently completed neutronic calcula- tiong will not support any final conclusions about the breeding potential of fully optimized DMSR cores, consideration must be given to the conse-— guences of conversion ratios lower than 1.00. The evaluations were per-— formed for the two—zone flux—flattened core described for the 30-year fuel cycle with the fuel processing concept for the break—even breeder added. If this system were operated with no constraint on the enrichment of the uranium in the reactor and no 238U addition, it would gradually develop intc an MSBR as the 238U was consumed. The system would then be fully self-sustaining on thorium with a breeding ratio of about 1.03 but with a very high enrichment of fissile uranium. Breeding ratios as high as 1.1l could be attained by changing the thorium concentration and/or the size of the inner core zone. With the addition of enocugh 238 ¢q keep the in- plant uranium denatured at all times, this particular reactor system would ultimately require an additional 27 in nuclear reactivity to be indefi- : nitely operable,* This reactivity deficit, if real, could be supplied in a number of ways, A moderate feed of 235U at 20% enrichment would extend the fuel cycle : to about 300 years. At that time, the 2380 loading would become exces- sive, and the reactor could no longer be made critical. While even 300 years may be much longer than any reasonable planning horizon, this re- sult indicates that a fully denatured MSR could have a very long, if not uniimited, fuel lifetime. If the enrichment of the feed material were allowed tc rise to 33% 235U, reactor operation could be sustained indefi- nitely without fuel discard. Because the buildup of 238U is the limiting phenomenon in the fuel | cycle of any nonbreeding DMSR, any process that would have the effect of removing 238U would improve the characteristics of the cycle. With the fuel feed enrichment set at 207 235U, the buildup of 238U could be limited *"Indefinitely operable” is arbitrarily defined here as maintaining kegf 2 1.0 for 600 years or longer. In all extended fuel cycles, the fuel is presumed to be transferred without loss from one reactor plant to another as required by hardware lifetime considerations, e 101 by removing some uranium from the fuel salt and replacing it with fresh feed. 1If only 1% of the uranium inventory were removed each year and con- signed to waste or to off-site recovery, the in-plant isctopic composition would reach equilibrium within 300 years, and the fuel cycle could be con—- tinued indefinitely. An even more attractive choice would be to remove some of the uranium, strip out part of the 238U, and return the remainder to the reactor plant. To examine this case, we assumed that 2% of the re- actor inventory would be treated each year and that the returning uranium would contain one-half the original 238U or enough for denaturing, which— ever was greater. (Only 238U was extracted in this preliminary calcula- tion.) The calculation showed that this approach also would allow indefi- nite operation and would require less feed material (see the following discussion) than the other options. 4,1.3 Partial fission—-product removal Although the reference fission-product processing concept could strongly affect the very long-term viability of DMSRs, the fissiom~product process would require substantial time and effort for commercial develop- ment, and, even then, it might not be a market success. Consequently, considering alternative processes might be useful, A variety of alternative separations procedures have been examined over the years in the ORNL MSR program for possible application in fuel reprocessing operations. Possible recovery of protactinium, uranium, and other actinides by selective precipitation of oxides has been examined, though most methods have preferred removal of uranium isotopes by fluo- rination to volatile UF.. Attempts to remove the lanthanides {the most important parasitic absorbers of neutrons) have included processes based on ion exchange, precipitation of intermetallic compounds, and even vola- tilization at low pressure of the other melt constituents™ to leave the very nonvolatile lanthanide trifluorides behind. All such processes re- quire solids handling, and many also have other disadvantages. None was *such a separation might be feasible, after fluorination of the uranium, for a fuel consisting only of LiF, BeFp, and UF,, but inclusion of considerable ThF, (as in a DMSR fuel) defeats such a process. 102 developed far enough to lead to an integrated process. Further, after discovery of the reductive-extraction/metal-transfer process, which, though complex, involved handling only liquids and gases, studies of all other separations were largely abandoned. An ion-exchange process for selective removal of lanthanide ions from the molten fuel has long seemed attractive in principle,19 but no attrac-~ tive ion exchanger for these materials has been demonstrated. An obvious difficulty is posed by the aggressive tendency of the molten LiF-BeFp- ThFy~UFy system to react with most materials that are likely to be useful. Certain refractory lanthanide compounds (such as carbides, nitrides, or sulfides) could conceivably be useful and sufficiently stable. The only candidate materials to date have been materials such as CeF3 and LaF3. By virtue of the formation of nearly ideal solid soclutions among the rare earth trifluorides, these compounds are capable of removing other (higher cross—section) lanthanides from the molten fluorides. The neutron cross sections of cerium and lanthanum are not negligible; because such an ex- change process saturates the treated fuel with CeF3 or LaFj3, the resulting fuel solution still has substantial parasitic neutron absorbers.!? The CeFy (or LaF3) exchanger also would presumably remove trivalent actinides {including plutonium) from the molten fuel. This would be unacceptable for a DMSR. No overall chemical process based on such separations has been de- scribed. Obvicusly, much development would be necessary before such a process could be demonstrated. Alsc, several solids-handling cperations apparently would be required and no process based on these operations could be capable of processing a DMSR on a short time cycle. However, given the present state of knowledge, the following process can be visu- alized to operate on relatively large (1- to 2-m3) batches of DMSR fuel, possibly after cooling for 5 or 6 d. The following steps would be neces-— sary. Step 1. Treat the melt with a strong oxidant to convert UF3 to UF,, PaF, to PaFg, and PuFy to PuFy. This should ensure that cerium is present as CeF, and, probably, that neptunium is present as NpFy. Americium and curium may be present as tetrafluorides but will 103 e probably still be mostly trifiuorides. This oxidized system will be corrosive, but it should be manageable in equipment of nickel or nickel-clad Hastelloy. ; Step 2. Precipitate the insoluble oxides using water vapor diluted in helium. The oxides UO,, Pa,0g, Pul,, Ce0,, probably NpO,, and possibly AmO, and CmO, should be obtained. With the exception of Zr0, and Pa,0g, these will be largely in solid soclution. The oxide solid solution is likely to centain 15 te 20% of ThO,; this would correspond to a few (less than 5) percent of the ThF, pres— ent in the fluoride. Recover the oxides by decantation and fil- tration. Step 3. Hydrofluorinate the oxides from step 2 into the purified LiF- BeF,~ThF, melt from step 7 and reduce the melt with Hy and then with lithium, thorium, or beryllium to reconstitute fuel with the desired UF3/UFy, ratio, Step 4. Hydrofluorinate the liquid from step 2 to remove excess oxide ion. Oxidize to get samarium and (if possible) eurcpium to SmFj and EuFj. Step 5. Treat the melt from step 4 with an excess of CeF3. This might be done in a column or in a two— or three—batch countercurrent operation. This removes a major fraction of the rare earths but does essentially nothing for cesium, rubidium, strontium, and barium. (If neptunium, americium, and curium are appreciably harder to oxidize than plutonium, they should remain in the salt in step 2 and should be removed on the CeF, in step 5.) Step 6. The LiF-BeF,~ThF, melt from step 5 contains only a fraction of the rare earth poisons but, of course, is saturated with CeFj. Oxidize the Ce3t to Ce¥t, Step 7. Precipitate the Ce% as Ce0,. Some ThO, will accompany the CeQ,, but the quantity should be small. Separate the precipitate by decantation and filtration. Feed the molten LiF-BeF,~ThF, to the fissile material recovery operation in step 3. | Step 8.. Dissolve the solid CeF, (contaminated with rare earths) from step 5 in some suitable salt (preferably not 7LiF—BeF2) and oxidize the Ce3t to Ce“+. 104 Step 9. Precipitate the cerium as Ce0, and recover the precipitate by decantation and filtration. Discard a portion of the molten salt, which cofitains rare earth fissiom products, to waste stor- age. Return the remainder with the necessary makeup teo step 8. Step 10. Combine the CeC, from step 9 with that from step 7 and treat these solids with HF and H, to obtain CeF3 (plus some ThF,). Use this as the major part of the reagent for step 3. This process would have a number of disadvantages when compared with the reductive-extraction/metal-transfer process. Zirconium, cesium, ru- bidium, strontium, and barium would not be removed, though none of these is a major problem. Néptunium probably would not be removed, though am- ericium and curium may be. Iodine would be removed either during the fuel oxidation or subsequent hydroflucorinations., Selenium and tellurium — as-— suming that they arrive at the processing plant — might be volatilized as elements or as fluorides during the fuel oxidation step (and they might cause a corrosion problem for the process). Heat generation by the fuel, even after a few days cooling time, would present problems, and the com- plex process would be difficult (possibly impossible) to engineer. At best, several days would be required to get a batch of DMSR fuel solvent through the process, though the fissile materials might be returned to the reactor with a 2-d holdup. An appreciable inventory of fuel material (but perhaps not more than 5% of reactor inventory) would be cooling and in the processing area. b4,1.4 Salt replacement Even with no chemical removal of fission products, the neutron pei- soning effect in a DMSR does not begin to approach saturation until after about 15 years of power operation at a 757 capacity factor. Thus, if the fission—-product inventory could be held at or below that corresponding to a 15-year level, a significant reduction in fueling requirements could be realized. The simplest way to limit the fission-product concentration in the salt is to discard a portion of the salt on a routine schedule and replace it with clean salt. With no refinement, salt discard would re- quire replacement of the fissile material as well as the fertile component 105 G and the scolvent (or carrier) salt and, therefore, would actually require a larger uranium supply than the 30-year once-through fuel cycle proposed for the reference DMSR concept. However, uranium is easily and effectively separated from the rest of the fuel mixture, so the denatured uranium could be removed and recycled at the reactor site with a minimum of ef- fort. Depending on the rate of salt replacement, this approach would significantly reduce the requirement for fissile uranium below that for the simple once~through cycle. 4,2 Fuel Cycle Performance Of the alternate fuel cyeles considered in this section, the break- even breeder, if it were successful, would provide for the best utiliza- tion of fisgsile fuel rescurces (235U). If that system were started up on 20% enriched 23°U, it would probably require 700 to 1000 Mg of natu- ral U30g to provide the initial fuel loading for each 1 GW of electric generating capability. [The separative work to enrich this fuel to 20% 235y would be less than ! million separative work units (SWU).] However, once provided, this fuel would continue to produce electricity in an arbi- trarily long succession of power stations (or as long as fertile material was available). Thus, the effective resource requirement could be made arbitrarily small by averaging it over a large number of plants. Even if the initial fuel charge were used in only one plant, the resource require- ment would be only 10 to 20%Z of that for an LWR with similar electric gen- erating capability. The converter options with fuel processing provide other estimates of the potential performance of DMSRs with fission-product cleanup (Table 29). The options, which were described earlier, may be summarized as follows: Option Fuel cycle A Initial load is 20% 235U; makeup fuel is 20% 23%y Initial load is 20% 23°U; makeup fuel is 33% 23°U Initial load is 20% 23%U; annual discard of 1% of uranium inventory; makeup fuel is 20% 23°y D Initial load is 20% 235U; annual reenrichment of 2% of uranium in- ventory to denaturin% limit or to one-half of prior 238y content; makeup fuel is 207 2°°¢ 106 Table 29, Performance datg for long-term fuel cycle options for DMSRs™ with full-scale fission—-product removal Optionb A B C D Conversion ratio after 20 years 0.90 0.% 0.9 0.9 300 years C.74 0.92 0.93 0.89 600 years e 0.92 0,93 0.89 Requirement for initial core loading U30g, Mg 860 860 860 300 Separative work, Mg SWU 860 890 860 300 Average requirement for fuel makeup per 30-year cycle Us0g, Mg During years 0300 1000 420 580 500 During years 301600 e 460 600 600 Separative work, Mg SWU During years 0—300 1000 440 580 508 During years 301-600 e 470 600 600 Uranium reenrichment, Mg/year O 0 0 0.60 Uranium discard, Mg/year 0 0 0.24 0 Fissile inventory at equilibrium, Mg Uranium 1,22 2.9 2.7 2.8 All fissile nuclides 3.2 3.1 3.0 3.1 AFor 1 GWe at 75% capacity factor. bSee text for characterization of options. “Not operable beyond 300 years. dAt 300 years. The tabulated results show that all four of these options would maintain relatively high conversion ratios for very long times. The U30g resource requirements for the initial core loadings are all similar, and all are slightly higher than that for the once-through fuel cycle (because of the volume of fuel in the processing system). 107 The fuel makeup requirements are expressed in metric tons of Uj0g for 30 years of operation in a 1-GWe plant at 757 capacity factor and are averages for ten 30-year cycles., The effect of this averaging is most pronounced for option A; the fuel makeup requirement is only a fraction of the average for the first one or two reactor lifetimes and is somewhat greater than the average for the last cycles. Thus, while this option would require more uranium than the others in the very long term, its per- formance for the first few reactor lifetimes would be quite attractive, Even for the long—term, this resource requirement would be well below that of current—-generation LWRs, Option B illustrates the long-term sav- ing in uranium resources that could be achieved if higher enrichments could be tolerated for the relatively small amounts of makeup fuel. Be- cause the resource savings are principally long term and the required uranium enrichment exceeds currently perceived denaturing limits, this appears to be one of the less promising options. The two remaining op- tions, C and D, both show favorable resource utilization properties for long times with only minor penalties for discarded uranium {option C) or uranium subjected to reenrichment (option D}. Of these, option D clearly would be preferable if reenrichment were an acceptable procedure. The preceding four converter optiomns and/or the break—even breeder would require the availability of a complex and expensive fuel cleanup facility within the primary containment of each reactor installation.” The technology for an integrated processing facility has not been fully developed, and past work clearly indicates that a substantial development effort would be required tc produce a commercially functional system. Even then, the capital, operating, and maintenance costs oflsuch a system possibly would have a significant adverse impact on the overall economic performance of the associated DMSR., Other factors to be considered for these options included the willingness of the reactor operator to assume % . . .y s Conceivably, a single cleanup facility could serve several reactors at a common site, but such an arrangement would complicate the operation and would add problems of inventory accountability among the various units. 108 responsibility for a chemical processing facility, the sociopolitical ac~- ceptability of colocating such a facility with each DMSR, and the licens-— ing questions that may arise from such an arrangement. The other end of the range of possible fuel cycle performances for DMSRs is represented by the 30-year cycle described earlier in this report as the reference concept. Although this system, with a lifetime require- ment of 1810 Mg (2000 short tons) of U30gp, would be the largest consumer of natural uranium and separative work among the DMSR options considered, it still would require substantially less of these commodities than the once~through fuel cycle in light-water reactors. In the absence of fa- cilities for recycling the non—-SNM constituents of the fuel salt, this approach would use less of such materials than any of the other alterna- tives. However, despite the 30-year fuel cycle, this concept would not eliminate all on—-site chemical treatment of the fuel salt. The activities to maintain the desired U3*/U"" ratio in the fuel and the treatments to 1imit the level of oxide contamination in the salt would still be needed. Thus, even the "simplest” DMSR would require some equipment for and some technical competence in chemical processing, even though neither would directly involve the SNM in the system. The intermediate concepts that make use of a shorter salt discard cycle merely substitute consumption of other fluoride salts for part of the fissile uranium consumption in the reference 30-year cycle. Because these other fluorides (especially 7LiF) may also be relatively expensive, this substitution might not always be cost effective. In addition, any system that used salt discard weould have to recover uranium from the "waste” salt to prevent excessive uranium consumption. This would add yet another chemical processing operation to the reactor plant. The alternatives that rely on special treatment schemes to remove fission products from the fuel salt may have attractive fuel utilization characteristics, but they have not been analyzed in sufficient detail to permit an accurate characterization. In addition, considerable research and development would be required before such processes could be shown to be technically feasible. Consequently, little incentive is apparent at this time to propose new and different chemical processing concepts for DMSRs, 109 5. COMMERCIALIZATION CONSIDERATIONS While the technological feasibility, the overall technical perform— ance, and the proliferation resistance of the DMSR are important charac- teristics to be considered in assessing its value as an alternative nu- clear concept, an overriding consideration is likely to be the commer- cialization potential of the syétem. This general attribute includes a number of considerations, such as: 1. the probable total cost of developing a commercially ready system; 2. the time required for such development, which strongly affects the impact a system can have on power needs; 3. the probable net economic performance of commercial units, which de- termines the attractiveness of the concept to its potential users, that is, the electric power utilities; 4, the ease of licensability of the commercial plants, which is a re- flection of the concept's sociopolitical attractiveness, as well as its technical performance. Some relevent information about the DMSR with respect to each of these points is presented in the following discussion. 5.1 Research and Development Since MSR research and development has been under way for some 30 years, the basic technology is well understood. However, much of it has not been developed to the stage and scale that would be required for the construction of large reactor systems. Thus, a significant R&D effort would be an important part of any program to commercialize MSRs. In ad- dition, until recently, development was concentrated on reactor concepts with a good breeding gain and a low fissile inventory so that the result- ing thermal breeder reactor system would have a reasonably short doubling time and could be considered a viable alternative (or complement) to fast breeder systems., The technology needs of the modified reactor concept that has been developed in response to the recent emphasis on prolifera- tion resistance differ from those of the nominal breeder concept. 110 5.1,1 Current status s MSR development has been carried through the design and operation of a proof-of-principle test reactor, the MSRE, which was an 8-MWt reac- tor that operated at ORNL from 1965 to 1969. This reactor demonstrated the basic reliability of a molten—salt system, stability of the fuel salt, compatibility of fluoride salts with Hastelloy-N and graphite, re-— 1iability of molten—-salt pumps and heat exchangers, and maintenance of a radicactive fluid-fueled system by remote methods. The reactor was crit- ical over 17,000 h, circulated fuel salt for nearly 22,000 h,'and gener— ated over 100,000 MWh of thermal energy. The MSRE had achieved all the objectives of the reactor test program when it was retired in 1969, After the successful operation of the MSRE, the reactor concept ap- peared ready for commercial development. In preparation for further de- velopment, three major reports were prepared: a conceptual design study of an MSBR in 1971 (Ref. 8), a review of the status of development in 1972 (Ref. 101), and a program plan for development in 1974 (Ref. 21). For reasons other than technological, the govermment decided mot to fund further development of MSRs, The program was cancelled in 1973, restarted in 1974, and finally terminated in 1976. The development of a proliferation—resistant DMSR would require basi- cally the same techmnological development program as was proposed for the MSBR, but the emphasis would be on reliability, ease of commercialization, licensing, and prcoliferation resistance rather than on high breeding per- formance. With these objectives in mind, the 1972 status—of-development report has been updated, and the program plan for development has been 102 (While the main outline of DMSR development modified for the DMSR. requirements will be presented in this report, the reader is referred to Ref. 102 for greater detail.) 5.1.2 Technology base for reference DMSR The base technology for MSRs is well established and has been largely “proven in principle” by the cperation of the MSRE. While no major un-— resclved technical issues exist at the present time, a large R&D effort would be required to bring molten-salt technology to commercialization. 111 s At the close of MSRE operation, two major technical issues appeared unresolved. The first was the control of tritium, which is produced in fairly large quantities in a molten—salt system and which is known to dif- fuse through metal walls. Subsequent engineering-scale tests have demon— strated that tritium is oxidized in sodium fluorcborate, the proposed sec- ondary salt for the DMSR, and appears toc be handled readily. However, this process is not yet well understood, and the effects of maintaining an ade- quate concentration of the oxidant on the long—term compatibility of the salt with the structural alloy are unknown. The seceond issue involved the compatibility of Hastelloy-N with fuel salt. Dperation of the MSRE showed that the general corrosion of Hastelloy~N and graphite in an operating MSR was near Zero, as expected. However, metal surfaces that had been exposed to fuel salt containing fission products were unexpectedly found to exhibit grain-boundary attack, which was subsequently shown to be caused by reac- tion with the fission product, tellurium, Further work has shown that tel- lurium attack can be controlled by either a modification of the Hastelloy—-N alloy or by contrcl of the oxidation potential of the fuel salt. The major areas of research required for commercialization of MSRs would involve improvement of the materials of construction (Hastelloy-N and graphite), the design of in-line instrumentation for high—temperature use, and the development of fuel processing (at least for the end of reac- tor life and possibly also for use on-line). The major areas of develop- ment involve the scale-up of reactor components {(e.g., pumps) and the de- sign and development of components that were not present in the MSRE (e.g., steam generators and mechanical valves). In addition, we antici- pate that the design of some components such as the fuel drain system and the reactor cell with its insulation, heating, and cooling requirements would be extensively modifed to meet currently unspecified licensing re- quirements, Another large area of development would be the control of the temperatures and flows in the primary and secondary salt systems and in the steam system to avoid salt freezing and excessive thermal stress. Alternatively, some components might be designed to accommodate such freezing. Still another area of development would be advanced remote maintenance techniques, including the replacement of components using remote pipe cutting and welding. 112 The concept of the DMSR has emphasized proliferation resistance, e and further design efforts would be expected to adhere to proliferation- resistance criteria. However, nc major areas have been identified in which the R&D requirements for a DMSR would be substantially different v from those for other versions of the MSR concept. Essentislly the same R&D program would be pursued as was planned for the MSBR. The selection of a low-power—-density core for the DMSR has relieved the requirements for core graphite (especially for gas permeability) and has simplified vessel design (because graphite replacement is not required). The selection of a reference DMSR without on—line fuel processing has removed the develop— ment of on—line processing from the expected critical path for reactor de- velopment. Processing development should proceed, however, to meet two closely related objectives: (1) development of on-line reprocessing to obtain the improved fuel utilization of the break—-even breeder DMSR op— tion as scon as possible and (2) development of a process {(probably using the same basic technology) for eventual central processing of fuel from once—through DMSRs, possibly in secure fuel service centers, 5.1.3 Base program schedule and costs An R&D base program has been presented in some detail in the program plan..102 The projected cost schedule (in 1978 dollars) for each major development activity annually from 1980 to 1994 and as a total for 1995 through 2011 is given in Table 30. The complete base program is pro—~ jected to cost about $700 million over about 30 years. Some of the costs are targeted for either the Molten—-Salt Test Reac- tor {(MSTR) or the demonstration DMSR {(as discussed in the following sec- tion), while other costs apply generally to the MSR development program. However, these costs do not include design and construction costs for the reactor plants. The schedule of fuel processing technology development was set up for the concurrent development of on-line processing. This schedule could be stretched if the once~through cycle were chosen for the first DMSRs. How— ever, the development of processing technology is an important goal of the program in any event. Table 30, Projected research and development costs for MSR base development program (Thousands of 1978 dollars) Cost by fizscal year . Target Development antivity Type fund reacror . ' 1980 1981 1982 1383 1984 1585 1986 1987 1988 1989 1990 991 1992 1993 Reacter design and Operating MSTR 430 1,279 1,100 720 9330 l.lZOa @70 9]Ua EEIN 880, BZUQ 570, 320, 180 analysis Operating Demo 290 200" 200 2009 200" s00° soo” s00” gz Reactor and component Operating MSTR 530 1,050 1,260 1,413 2,630 4,270 5,920 6,610 7,970 9,2lfla Sflfiu 5,000'{'I 3,0003 E,GOOa terhnology Uperating Demo ) Y . 100 so0” 1,000% 2,100 21002 Capital all 40 @d 150 80 5,3307 79,4007 26,100“ 57 840 1,130 L, 400 won” gon? 900a Safery and licensing Operating MSTR 117 303 351 468 397 576 5§29 97 1,100 1,235 1,300 l,SGUa 1,530: 1,5005 Operating Demo 1007 1007 Fuel and coeolant Operating MSTR 695 990G 1,125 1,230 1,345 1,360 1,430 1,475 1,30C 35, 5604 4b5 465 250 chemistry Operating Demo 65" 440" 535" 5357 750 Capital all 95 205 335 310 180 410 325 350 L85 53 50 50 lDUa lOUa Analytical chemistry Operating MSTR 260 405 485 570 670 715 765 760 695 615 480 435 385 275 Qperating Demo " . 1157 2259 Capital ALl 35 295 290 210 185 255 120 n 40 507 sg® 07 s0? Process materials Operating MSTR 425 610 820 950 1,050 930 765 400 400 205 205 180 1067 180a Capital All 100 1,175 2,070 1,560 1,380 700 400 350 250G 100 Fuel processing Operating MSTR 1285 2,170 , 480 2,455 Z,SODa 2,8@0a J,OOOa 3,250 3,670 3,670 3,510 Z,000 300 2 technology Operating Demo 5 5 se0® 1,000% 15007 Capital 411 75 1,060 12,750 o 7,000 510 G 260 400 515 500 200 150a 1507 Structural alley Operating MSTR 2200 2,800 3,025 3,590 1,910 1,785 1,612 1,560 1,534 1,560 1,326 as0 Operating Demo 742 ma® 10007 15007 Capital All 955 1,170 1,502 507 98 169 15G 176 137 150 137 82 10&Z 1504 Moderator graphite Operating MSTR 300 300 450 600 620 500 630 650 550 500 400 400 200 300 Operating Demo 1007 2007 Capital All 106G 75 166 150 1590 G0 130 100 100 75 75 75 50 100% Total fundsc 7462 13,968 23,293b 14,810 2 ,4159 95,370b 43,296b 18,836 20,281 21,440 21,827 15,590 13,470 14,470 ™ L] 1,500? 156% 300, 2007 1507 14,870 13,100, 4, 340" 62,9203 8,900 118,530 13,761? 1,000% 13,675 3,275 23,670 ZS,E?ZP 4,87&; 5,631 6,750 s 6001 1,500° 370,878 Cost from 1‘)95{I threugh 2011 1,000 25,008 20,000 80,000 8,000 3,000 40,001 5,005 15,0600 2,500 2,000 5,00 1,040 1,825 325 12,000 50,000 5,000 10,900 30,000 3,800 3,000 8,000 1,900 331,485 alncludes costs estimated without detailed program analysis. Includes funds authorized for major development facility. “Total funds through 201i: $702,563. ( 114 5.2 Reactor Build Schedule 5.2.1 Reactor seguence In addition to the program of hase technology cutlined previously, a series of three developmental reactors culminating with a standardized commercial plant are proposed for construction. The proposed development plan is given in considerable detail in Ref. 102, The development sequence would start with a preliminary conceptual design for a 1000-MWe DMSR (which would actually be the second reactor in the series) to further define the development problems. This would be followed by considerable component development (in the base technology program), after which the MSTR would be designed. The MSTR is proposed to be in the 100- to 250-MWe size range, which would employ components in the one—~fifth to full-scale range (based on the 1000-MWe conceptual design); Most of the MSTR components weuld be tested in an MSTR non- radioactive mockup before the MSTR was actually assembled. During construction of the MSTR, component development and design of the prototype DMSR would proceed, with detailed design and construc- tion coincident with operation of the MSTR. This would allow close feed- back from MSTR experience into the design and construction of the proto- type. Finally, detailed design of the first standard DMSR would proceed during construction of the prototype so construction of that reactor could begin shortly after the prototype started operation. 5.2.2 Schedule and costs A potential reactor build schedule is given in Fig. 20. This is the same development schedule as was proposed for the break—even breeder DMSR option.9 In the latter case, the assumption was that development of the on-line reprocessing system proceeded in parallel with development of the reactors. Therefore, no credit can be taken for omitting process design in the schedule for the once-through DMSR. However, removal of the pro- cess development from the expected critical path for reactor development removes a major source of uncertainty and potential delay from the devel~- cpment schedule. MSTR { LEAD COMMERCIAL — pmsr (1980 (PROTOTYPE) e \CONCEPTUAL ORNL -DWG 78-7062R TECHNOLOGY DEVELOPMENT v _OPERATE FOR DEVELOPMENT & AUTHORIZE DESIGN TN 1985 CONCEPTUAL DESIGN DESIGN STANDARD | DESIGN DMSR Fig. 20. Potential reactor constructicn schedule for commercializa- tion of DMSRs. CTT 11é The construction cost of the MSTR may be estimated by updating the g cost estimate for the MSTR prepared in 1975 for the MSBR program. Using a construction materials and labor increase of 12%/year gives a multi- 'plier of 1.4 and a cost for the MSTR in 1978 of about $600 million. A detailed estimate of the cost of a 1000-MWe DMSR based on a mature technology is given in a following section. From this, we have estimated the cost of a first standardized DMSR by applying a factor of 1.5 to al- low for increased first-of-a-kind costs and the cost of a lead commercial prototype by applying another factor of 1.5 to allow for increased proto- type costs. Using this procedure, the cost of a 1000-MWe prototype DMSR is estimated to be $1470 million and the first standardized DMSR $980 million The prototype DMSR need not be as large as 1000 MWe; the cost could be reduced some, for example, by building a 500-MWe prototype with two steammgenerator loops rather than four. Estimating the probable cost of experimental and prototype reactors in advance of design is exceedingly difficult. The cost estimates pre- sented were made by a staff that has had experience in the design and op— eration of experimental reactors, particularly the MSRE, The MSRE was constructed and operated within budget, which is an indication that the technoiogy is reasonably well understood and that cost estimates for fu- ture reactors are probably realistic, if not absolutely accurate. Con— versely, the proposed reactors are a large step up in scale from the MSRE and would be subject to the vagaries of the licensing process for a new reactor type. These factors introduce uncertainties into the cost esti- mates which are beyond evaluation at the present time, 5.3 Economic Performance of Commercial DMSR The projected cost of power from a commercial DMSR can be estimated only approximately because the DMSR is in the conceptual stage and not even a detailed conceptual design has been prepared. However, by taking advantage of the similarity of the DMSR to the MSBR (for which a detailed ‘ conceptual design and cost estimate have been prepared) and by carefully comparing the DMSR with LWR and coal-fired power plants (which would have 117 some components in common with or similar to those of a DMSR}, a reason- able cost estimate can he made. A cost estimate was prepared for the MSBR in 1970 and appears in Ref. 8. These costs were taken (for most accounts) as the basis for the DMSR estimate using the following method. 1. The costs were adjusted to take into account the differences in size or other requirements for the DMSR, For example, the reactor ves- sel cost was increased to take into account the larger size of the DMSR vessel, 2. The 1970 costs were increased by a multiplier based on the in- crease in construction materials and labor costs from 1970 to 1978. The multiplier was calculated to be in the rafige 2.3 to 2.5; to be conserva- tive, the multiplier 2.5 was used. This represents an annual rate of in- crease of about 127, In addition, the costs of a pressurized—-water reactor (PWR), a boiling~water reactor (BWR), and a coal-fired plant in 1978 were esti- mated using the CONCEPT V code, 103 Where appropriate, some DMSR cost accounts were estimated based on the analogous account in one of the CONCEPT estimates. For example, the turbine-generator cost was based on the coal-fired plant estimate because the same type of supercritical steam turbine would be used. In reporting the results, the cost estimates for the PWR and (in the appendix) the coal-fired plant are given for comparison. (The costs of the BWR were not substantially different from the PWR.) The results of the estimates for capital, nonfuel operation and maintenance, and decommissioning costs are presented in the next three sections. The fuel cycle costs were calculated independently and are presented in a fourth section. 5.3.1 Capital costs The bases for the capital cost estimates are given in Table 31. Some minor adjustments in the code of accounts used in the 1970 MSBR cost estimate were required to conform to the present code of accounts. Table 31. 118 Bases for capital or investment cost estimates e Bases Fxcluded costs Plant site, Middletown (New England area) Code of accounts, NUS-5331 and NUREG-0241, ~0242, ~0243 {direct and indirect accounts) Cost date, 1978.0 Regulation codes and standards, 1876 Evaporative cooling Commercial plant size (optimum), 1000 MWe Cash flow, 1978 to commercial operation in 1988 Capital costs in 1978 dollars/kWe Development costs and first-of-a- kind ' Switchyard (including main transformer) Nuclear liability insurance Interest during construction Escalation during construction Contingency allowance Owners costs, including expenses for taxes and property insurance, spare parts, staff training General and administrative, site selection, and other owner-related expenses Salt and fuel inventory, including chemical processing system The capital costs estimated for the DMSR and the PWR by major (two~- digit) accounts are given in Table 32, A further breakdown (for three- digit accounts), which also includes the estimate for the coal-fired A cumulative cash flow sched- ule for the DMSR (adapted to the CONCEPT V cash flow schedule} is also plant, is given in Appendix A (Table A.1). given in Appendix A (Table A.2). The capital costs of the fuel treatment facilities are not included here but are taken into account in the fuel cycle cost (see Sect. 5.3.4). However, space and equipment for handling the coolant salt are included in the reactor plant estimates. The DMSR is estimated to cost $653 million, or about $650/kWe in 1978 : dollars. This compares with about $600/kWe for a PWR plant and $380/kWe for a coal plant without flue-gas cleaning. 119 e Table 32. Capital cost estimate of commercial 1-GWe DMSR and PWR plants (Expressed in millions of 1978.0 dollars) Account No. ITtem DMSR PWR Direct costs 20 Land and land rights Z 2 21 Structures and improvements 124 111 22 Reactor plant equipment 180 139 23 Turbine plant equipment 100 113 24 Electric plant equipment 54 44 25 Miscellaneous plant equipment 17 13 26 Main condenser heat rejection system 14 22 Total direct costs 491 L4 Indirect costs 91 Construction services 75 70 g2 Home office engineering and service 53 53 » 83 Field office engineering and service 34 30 Total indirect costs 162 153 Total plant capital cost 653 597 A discussion of some of the important assumptions and results for the major accounts follows. Account 21, Structures and improvements, The primary and major structure in the DMSR plant is the reactor contaimment building. While layouts of internal areas would differ, comparable costs with the PWR are expected. An allowance has been entered for plant lifetime storage of radicactively contaminated items including provisions to facilitate decommissioning operations. The other structures parallel the PWR structures in cost. The tur- bine rooms are considered comparable. A supercritical steam turbine for the DMSR is considerably smaller than a PWR turbine, but space was allowed for extra piping and equipment that may be required to adapt the super- critical system to a molten-salt steam generator. Space was also allowed for handling and storing coolant salt for normal operations and for 120 e storing the blowdown material that would result from a major steam leak in the steam generator. Account 22, Reactor plant equipment. The reactor and associated heat-transfer system costs have been updated from the 197C estimate using a multiplier of 2.5. About 10% of this total (accounts 221 and 222} has been added for engineered safety features (which were not previously con- sidered) and for larger salt volumes., This amount also covers external heat dissipation equipment for engineered safety features., Radioactive waste handling in the DMSR was estimated to cost about the same as radio— active waste processing for the PWR. Fuel handling and storage and maintenance equipment were updated from the MSBR estimate using the 2.5 multiplier, An allowance was made for the control features necessary for making the plant operate on the salt-steam cycle. Account 23, Turbine plant eguipment., This account parallels the coal plant case, which uses supercritical steam~cycle equipment. The feed~heating account was increased 507 to allow for operating design fea- tures peculiar to the salt-loop application. Account 24. Electric plant equipment. Except for provision of about . 25 MWe of electric heating associated with the salt loops, this account is similar to the PWR case. Account 25, Miscellanegus plant equipment. The auxiliary steam sup- ply cost for the PWR has been increased to adjust the PWR cost to the DMSR basis. Account 26. Main condenser heat rejection system. Design for the coal plant is comparable to the DMSR; therefore, the same costs have been assumed . Accounts 91, 92, and 93. Indirect costs. The DMSR costs are based on PWR costs adjusted upward for those accounts in which higher labor re- quirements are anticipated. 5.3.2 Nonfuel operation and maintenance cost Estimates of nonfuel operation and maintenance (0&M) costs of 2.82 mills/kWh are based on the single-unit base~load plant. The procedure 121 T is based on the OMCOST code. %" Annual expenses are derived for staff, maintenance materials, supplies and expenses, nuclear liability insur- ance,* operating fees, and general administrative activities. Operation v and maintenance costs are presented in 1978 dollars and are divided into fixed {demand related) and variable {energy-related) components, Staff requirements are given in Table 33 for a one-unit plant. An- nual costs have been derived from the 0&M cost code with modifications to adjust the maintenance-labor ratio to 70:30. Estimates were that a DMSR might require major plant work at ten—-vear intervals {over and above PWR réquirements), for which maintenance labor was increased ~50%. The sumnary of annual O&M costs is given in Table 34. 5.3.3 Decommissioning and disposal cost Costs for decontamination and decommissioning of the facilities would be incurred at the end of plant life., A nuclear waste working group com— ’ prised of DOE, Nuclear Regulatory Commission (NRC), and Environmental Pro- tection Agency (EPA) officials is working to identify legislative needs on handling nuclear wastes. Preference is now given for dependence on : "engineered and natural barriers" for contrcl of on—-site material after decommissioning with fall-back dependence on institutional controls for a "finite time.” The group alsc copts for dismantling a decommissioned site after a short decay period, rather than either of two other options, which are entombing and mothballing nuclear facilities, The cost of dismantling a DMSR is expected to be greater than for an LWR because the activity level of components in the primary circuit is higher. A number of estimates of the decommissioning cost of LWRs have been prepared; as a basis for our estimates, we have selected a representative recent (1978) estimate by the Tennessee Valley Authority (TVA) for their Yellow Creek plant early site review. The estimated de-— commissioning cost for this plant was $/8 million for a BWR. If we as— sume that the cost for a DMSR would be about 107 greater, then the esti- - mated decommissioning cost for a DMSR would be about $86 million. A *This is excluded during construction period when no fuel is on site. 122 Table 33. Staff requirement for one 1-GWe DMSR power plant Employee type Number ° Plant manager's office Manager 1 Assistant 1 Quality assurance 3 Envirommental control i Public relations i Training 1 Safety 1 Administration and services 13 Health services 1 Security 66 Subtotal 89 Operations Supervision (nonshift) 2 Shifts 33 Subtotal 35 Maintenance Supervision 8 ? Crafts 16 Peak maintenance, annualized 96 Subtotal 120 Technical and engineering Reactor 1 Radiochemical 2 Instrumentation and control 2 Technical support staff 17 Subtotal 22 Total 266 less security 200 Less security and peak 104 maintenance 123 Table 34. Summary of annual nonfuel 0O&M costs for base-load steam~electric power plants in 1978.0% Plant type DMSR with evaporative cooling towers Number of units per station 1 Thermal input per unit, MWt 2270 Plant net heat rate 7755 Plant net efficiency, 7% 44,00 Power output, net, MWe 1000 Annual net generation, million kWh 6570 Plant factor 0.75 Annual costs, thousands of dollars Staff, 266 persons at $23,412 6228 Maintenance material ' 6555 Fixed 6555 Variable G Supplies and expenses 3317 Fixed, plant 3000 Variable, plant 317 Insurance and fees 408 Commercial liability insurance 284 Govermment liability insurance 18 Retrospective premium 6 Inspection fees and expenses 10C Administrative and general 2367 Total fixed costs 18,500 Total variable costs 317 Total annual 0&M costs 18,875 Unit costs, mills/kWh(e) Fixed unit O&M costs 2.75 Variable unit O&M costs 0.07 Total unit O&M costs 2.82 %Excludes the salt inventory losses; assumes nuclear insurance at LWR rates. 124 large uncertainty is present in this estimate, of course, because of the = ™% limited experience in decommissioning. However, the cost ¢f decommission- ing does not appear to be a large fraction of the cost of the construc- tion, and the present worth of expenditures to be made in the future is small. The present worth of the estimated cost of decommissioning a DMSR 40 years after start—up, discounted to the start—up date at a 4.57% dis-— count factor, would be about $15 million. 5.3.4 Fuel cycle costs At this stage of development and optimization of a oncemthrough DMSR, several assumptions are necessary if a fuel éycle cost is to be estimated. One assumption is that the initial fuel charge will consist of 74 mole % 7LiF, 16.5 mole % BeF,, 8.23 mole % ThF,, and 1.27 mole % UF, plus UF3. If, as seems reasonable, an allowance is made for an additional 27% of molten fuel in the drain tank, the initial fuel solvent will require 149 metric tons of ThF,, 113 metric tons of LiF, and 45.6 metric tons of BeF,. 1In addition, 23.5 metric tons of UF, enriched to 20% in 2357 is re- quired; this is equivalent to 804 metric tons of U0, and would require 8.05 x 105 SWU for its enrichment. The purified fuel delivered to the re- actor storage tank would consist of 331 metric toms (7.30 x 10° 1b) of material. Total cost of the initial fuel is based on U,0, at $35/1b, separative work at $80/SWU, BeF, at $15/1b, thorium at $15/1b ThO,, and LiF avail- able at $3/g of contained Li.” An additional $2/1b of tetraflucride has been allotted for comnversion of ThO, to ThF, and for conversion of UFg to UFye The fuel, made by migxing the powdered ingredients, must be purified in the molten state before use (as described in previocus sections). Given the component fluorides at the prices above, the assumption is that this purification can be performed for $6/1b of finished product, With these assumptions, the total cost of the initial DMSR fuel charge is near $225 million (see Table 35). If we assume that the annual *This is the official price for small quantities of 99.99" 7Li for use in PWRs. It is almost certainly toco high (probably five-fold) if lithium were actually used in such quantities. 125 o Table 353. Cost of initial fuel charge for once-through DMSR (331 metric tons fuel) Cost . Fuel (dollars) Fuel solvent Materials 7LiF (30.46 metric tons lithium at $3/g) 91.38 x 10° ThF, (127.68 metric tons ThO, at $15/1b) 4,22 x 108 BeF, (45,60 metric tons at $15/1b) 1.51 x 10 Uranium (803.8 metric tons U30g at $35/1b) 62,03 x 108 Separative work (8.05 x 10° SWU at $80) 64,40 x 10® Conversion and purification Thorium (148,96 metric tons ThF, at $2/1b) 6,57 x 10° Uranium (23,50 metric tons UF, at $2/1b) 1.04 x 10° Fuel mixture (331.2 metric tons at $6/1b) 1.99 x 106 Total cost of initial fuel 224,30 x 10° Annual charge (12%) 26,92 x 10° use charge is 12%, this initial fuel contributes $26.9 millien/year to the fuel cycle cost. A once-through DMSR must add uranium at more or less regular inter-— vals over its operating iifetime. Though other modes of addition are pos-— sible, for this assessment we assumed that the uranium additions will be made as a liquid ’LiF-UF, mixture containing 30 mole % UF, (melting at about 540°C). Such additions would appreciably increase the ’LiF con- centration of the fuel. Adjustment of the UF3/UF, ratio is assumed to be done by in situ reduction of UF, with metallic beryllium. The fuel stream is assumed to be treated (once in each 1000 full-power dayé) with an an- hydrous HF-Hs mixture to remove inadvertent oxide contamination; the re- sulting oxidation of UF3 is managed by additional reduction with beryllim. With this mode of opération, BeF» equivalent to nearly 67 of that in the original fuel charge would be added over the reactor lifetime. Such ad- ditions of ’LiF and BeFo dilute the fuel and appreciably increase its volume so that an increasing (though relatively small) fraction of the 126 fuel would remain in the reactor drain tank. The dissolved parasitic s neutron absorbers, of course, would also be diluted. At this stage of DMSR development, no detailed optimization for such fuel dilution has been made. For this assessment, we assumed that, as a consequence of = this dilution effect by fuel maintenance, the uranium additions shown in Table 17 plus uranium (at 20%Z enrichment) and thorium equivalent to 3% of the initial inventory would be required over the 30-year operating life of the reactor. Thorium is assumed to be added as a molten mixture of 7LiF—Tth containing 28 mole 7 ThFy (melting point of 370°C). Table 36 shows the average annual cost of these additions and fuel maintenance. Costs ¢of ThFy, UFy, LiF, and separative work are those de- scribed previously. Metallic beryllium is assumed to cost $75/1b. Cost of preparing the 7LiF—Tth and 7LiF--UFq. mixtures was assumed tc be $20/1b (plus the cost of the solid raw materials). Table 36. Average annual cost of fuel additions and maintenance Cost ' " (dollars) Materials 7L1F (62.4 kg Li% at $3.00/g) 1,87 x 10° 0 (16.2 kg at $75/1b) 2.7 x 103 ThF4 (0.149 metric toms at $15/1b ThO,) 4,2 x 103 UF, (34.83 metric tons Us0g at $35/1b) 2.689 x 106 Separative work {3.48 x 10% SWU at $80) 2.784 = 10% Conversion and purification Thorium (0.14% metric tons ThF, at $2/1b) 7 x 102 Uranium (1.03 metric toms UF, at $2/1b) 4.5 x 103 L1F-ThF4 (0.181 metric tons at $20/1b) 8.0 x 103 L1F~UF4 (1.2209 metric tons at $20/1b) 5.42 x 10% HF-H, treatment of fuel Fixed charges on equipment {at 107%) 1.5 x 10® Annuwal operating cost 5.0 x 107 Total average annual cost 7.73 x 10% 127 i There are no detailed estimates of the capital or operating costs of the equipment for HF-H2 treatment to remove 02" from the small batches of fuel. For this assessment, we assumed that the capital cost is $15 x 10° and that its operation costs $500,000/year. As a consequence of the assumptions and the estimates described, the cost of producing 6.57 x 107 kWh/year (operation at 757 plant factor) ap- parently averages $34,650,000, and the resulting fuel cycle cost is about 5.3 mills/kWh. Note that, if the price of 7Li were lowered by five—-fold [to $0.60/g (8272/1b)]1, the resulting fuel cycle cost for the once-through DMSR would fall to slightly below 4 mills/kWh. 5.3.5 Net power cost Because the return on the plant capital investment would be a sub- stantial factor in the net cost of power from a DMSR and because a number of terms that would be important in a commercial plant were omitted in de- veloping the capital cost estimate, projecting a potential net cost for DMSR power is not appropriate. Substantially more design and development would be required to support a reasonably reliable estimate. However, the previcus discussions suggest that the cost of power from a DMSR would not be greatly different than that from other nuclear systems. 5.4 Licensing | Although two experimental MSRs have been built and operated in the United States under govermnment ownership, nomne has ever bheen subjected to formal licensing or even detailed review by the NRC. As a consequence, the question of licensability of MSRs remains open; the NRC has not yet identified the major licensing issues and the concept has not been con- sidered by various public interest organizations that are often involved in nuclear plant licensing procedures., Further, the licensing experience of solid-fueled reactors can be used as only a general guide because of significant fundamental differences between those systems and MSRs. Pre- sumably, MSRs would be required to comply‘with the intent, rather than 128 the letter, of NRC requirements, particularly where methods of compliance e are concept—-specific. Any special issues that might arise from public consideration of an MSR license probably would be ciosely associated with those features of ® the reactor concept that affect its safety and envirommental attributes, ‘A number of these features and attributes have been identified in earlier sections. One major difference between more conventional reactors and MSRs is in the confinement of radicactive fuel and fission products. The barriers to fission-product release in LWRs are (1) the fuel element clad- ding, (2) the reactor coolant pressure boundary (RCPB) (i.e., the primary- loop vessels, components, and piping), and (3) the reactor containment. This arrangement relies heavily on the ECCS to prevent cladding failure in the event of coolant loss by failure of the RCPB.* Without adequate ECCS performance, a failure of the RCPB conceivably could leave the fis- sion products with only one level of confinement intact. A different situation would prevail in an MSR because the fission—- product confinement barriers are different. The relevant barriers in an MSR are (1) the RCPB, (2) the sealed reactor cells or primary containment, and (3) the reactor containment building or secondary contaimnment. Be- cause the fuel is 3 circulating liquid that is alsc the primary cooclant, there is no thin fuel clad that could fail quickly on loss of coocling or in a reactor power/temperature transient. Thus, an entire class of po- tential accidents could be eliminated from the licensing consideration. Failure of the RCPB in an MSR would cause no short—term threat to either of the remaining two barriers to fission-product release. The ultimate requirements for longer—term protection of the fission-product barriers cannot be defined without extensive system design and safety analysis, but preliminary considerations suggest that the requirements may not be extensive. Although radiocactive materials would have three levels of confine- ment during normal operation, a different condition could exist during maintenance operations that required opening of the primary containment, * Failure of the RCPB is one of the mechanisms for initiating a loss- of-coolant accident (LOCA). 129 particularly if such activities were undertaken after an RCPB failure. However, in a shutdown situation, substantial confinement can be achieved through access limitation and controlled ventilation because, as shown by MSRE experience, fission products are not readily released and dis- persed from stagnant salt. Thus, whether fission-product confinement would be a net favorable or unfavorable factor for a DMSR in a licensing proceeding is not clear at this time. At the end of reactor life, a DMSR without fuel processing would contain the entire fission-product inventory associated with the 30-year operating history of the plant., Some of the volatile nuclides, especially 85kr and 3H, would have been accumulated in storage containers outside the primary circuit, and the noble metals would have plated out on surfaces in the primary circuit. The inventories of these nuclides, which would not be strongly affected-by nuclear burnup, would be about the same as those produced in a solid—fuel reactor with the same thermal power level and duty factor. However, because the DMSR would generate only about two— thirds as much thermal power as an LWR for the same electrical output, it would produce a correspondingly smaller inventory of fission products, Most of the other fission products and all the transuranium nuclides would remain with the fuel salt in a DMSR. The inventories of these mate- rials would be further reduced by nuclear burnup resulting from exposure of the nuclides to the neutron flux in the reactor core. This effect would be particularly important for the high-cross—section nuclides such as the major plutonium isotopes. Consequently, the net production of plu- tonium would be much smaller for a DMSR than for a comparable solid-fuel reactor, but the production of higher actinides would be much greater be- cause of the long effective fuel exposure time, Although a DMSR would produce a much smaller total inventory of some important nuclides over its lifetime than an LWR, the actual in-plant in- ventory could be substantially higher for the DMSR because there would be no pericdic removal during refueling operations, (There would also be no major shipments of highly radiocactive spent fuel from the plant during its lifetime and no out-of-reactor storage of such materials until after the final shutdown.) Thus, if a major release of in-plant radionuclides could occur, the consequence might be more seriocus in a DMSR than in an 130 LWR. However, considering the mechanisms and probabilities for release events, along with the consequences, would be necessary in assessing any effect on system licensability. Before any MSR is licensed, we probably will need to define a com- plete new spectrum of potential transients and accidents and their appli- cable initiating events that are to be treated in safety analysis reports. Some of the more important safety—significant events for an MSR were umen— tioned earlier, but even routine operational events may have a different order of importance for this reactor concept. TFor example, moderate re- actor power disturbances would not be very important because one of the principal consequences, fuel cladding failure, is a nonevent in an MSR. Conversely, a small leak of reactor cooclant would be an important event because of the high level of radiocactivity in the MSR coolant. The above examples of significant differences between MSRs and other licensed reactors illiustrate why a substantial design and analysis ef- fort would be required — first to establish licensing criteria for MSRs in general and a DMSR in particular and second to evaluate MSR licens— ability in relation to that of other reactor types. This requirement, with no a priori assurance that an MSR could be licensed, makes it un- likely that private organizations in the United States would undertake the development and commercialization of MSRs. Instead, if such develop~ ment were pursued, government funding probably would be required, at least until the licensing issues could be resclved and near-commercial units could be constructed. ......... 131 6. SUMMARY AND CONCLUSIONS The technology of MSRs was under development with U.S. government funding from 1947 to 1976 with a nominal one-yéar interruption from 1973 to 1974, Although no significant effort to commercialize MSRs was in- volved in this work, a very preliminary conceptual design was generated for a 1000-MWe MSBR, and some alternate fuel cycles were examined. The current study of denatured MSRs was supported by the program (NASAP) to identify, characterize, and assess proliferation-resistant alternatives to currently projected nuclear power systems. In principal, MSRs could be operated with a number of fuel cycles ranging from plutonium—fueled production of denatured 233U, to break-even breeding with Th-233y fuel, to high-performance conversicn of thorium to 233 a %33y makeup fuel. The last of these cycles currently U with denature appears to be the most attractive and is the one chosen for characteriza- tion in this study. The fuel cycle would invelve an initial loading of denatured 235U; operation for 30 years (at 75% capacity factor) with 235y makeup, no fuel discharge, and no chemical treatment for fission-product removal; and end-of-life storage/dispcsal of the spent fuel. The resource utilization of this cycle could be significantly enhanced by end-of-life recovery of the denatured uranium in the fuel salt via fluorination. 6.1 Reference-Concept DMSR The differences between a DMSR and the conceptuzl design MSBR in- volve primarily the reactor core and the fuel cycle. Thus, the rest of the primary circuit (e.g., pumps and heat exchangers) and the balance of the plant would be very similar for both concepts, and the descrip- tions developed for the MSBR are presumed to be applicable to the DMSR. Minor variations that might be associated with design optimization are not considered. The reactor vessel for the DMSR, about 10 m in diameter and 10 m high, would be substantially larger than that for the high-performance breeder. This would permit the low power density required to allow a 30~year life expectancy for the reactor graphite and would also reduce 132 neutronic losses to 233pa. Other effects of the low power density in- s clude reduced poisoning effects from in-core fission products and an in- creased fissile inventory. The reactor core would consist of a central region containing 20 - vol % fuel salt and a larger surrounding zone containing 13 vol ¥ salt,. Neutron moderation would be provided by vertical cylindrical unclad graph- ite "logs,” with fuel salt flowing upward throuéh central passages and between the moderator elements. The core would‘befsurrounded first by salt plenums and expansion spaces and then by a graphite reflector and the reactor vessel, With this core design, a 1-GWe plant would require an initial fis- sile loading of 3450 kg 2351 at 20% enrichment {extractable from about 870 short tons of Ujg0g). Over 30 years at 75% capacity factor, the fuel makeup requirement would be about 4470 kg of 20% enriched 235y (from 1125 short toms of U30g) for a lifetime U,04 demand of 2000 short tons. How- ever, at the end of plant life, the fuel salt would contain denatured fis- sile uranium (233U and 235U) equivalent to at least 800 short tons of natural Ug0g. If this material could be recovered (e.g., by fluorination) and reeunriched, it would substantially reduce the net fuel requirement of the DMSR. Preliminary calculations of the kinetic and dynamic characteristics of the DMSR system indicate that it would exhibit high levels of control- lability and safety. The system would also possess inherent dynamic sta- bility and would require only modest amounts of reactivity control capa- bility,. A first—-round analysis of the thermal-hydraulic characteristics of the DMSR core conceptual design indicated that the cylindrical moderator elements would be adequately cooled by the flowing fuel salt and that reasonable salt temperature distributions could be achieved with some orificing of the fuel flow passages. While some uncertainties about the detziled flow behavior in the salt-graphite system remain which would have to be resolved by developmental testing, the results would not be expected to affect the fundamental feasibility of the concept. The primary fuel salt would be a molten mixture of LiF and Ber con— taining ThF,, denatured UF,, and some PuF3. Lithium highly enriched in 133 the 'Li isotope (>99.99%) would be required, and the mixture would gradu- ally build up a significant inventory of fission-product and higher- actinide fluorides. This mixture would have adequate neutronic, physical, thermal-hydraulic, and chemical characteristics to function for 30 years as a fuel and primary reactor coolant. Routine maintenance of the salt would be required to keep some of the uranium in the partly reduced ydt state for the preferred chemical behavior. Although severe contamination of the salt with oxide ion could lead to precipitation of plutonium and uranium oxides, the solubility of these oxides is high enough that an increase in oxide ion concentration probably could be detected and stopped before such precipitation cccurred. In ad- dition, cleanup of the salt on a routine basis to maintain the required low oxide concentration would be relatively easy. The fuel salt is alsc highly compatible, both chemically and physically, with the proposed structural alloy, Hastelloy-N, and with the propecsed unclad graphite mod- erator. The radiatioen resistance of the fuel salt is well established, and no radiation decomposition would be expected except at very low tempera— tures (below ~100°C). The noble~gas fission products, xencn and krypton, are only sparingly soluble in fuel salt and would be removed continucusly during reactor operation by a helium sparging system. Portions of some other volatile fission products might also he removed by this system. Another class of fission products, the noble and seminoble metals, would be expected to exist in the metallic state and to plate out mostly on metal surfaces in the primary circuit. Keeping tellurium, which can be harmful to Hastelloy-N when deposited on its surface, in sclution in the salt may be possible by appropriate control of the reduction/oxidation potential of the salt. Most of the fission products would remain in solu- tion in the fuel sait. It appears (but must be demonstrated) that a full 30-year inventory of these materials could be tolerated without exceeding solubility limits. Because routine additions of uranium would be required to maintain criticality in the reactor, additions of lithium and beryllium would also be required to maintain the desired chemical composition. Some of these additions, conceivably, could be used to help control the oxidation state 134 of the salt, which would have to be adjusted routinely to compensate for p the oxidizing effect of the fission process. Also, the total salt inven- tory possibly would have to be limited through cccasional withdrawals of some salt. . The DMSR, in common with other systems that would use molten fluo- ride salts, would require a special primary structural alloy and, pos- sibly, special graphite for the moderator and reflector. The alloy that was originally developed for molten—-salt service, Hastelloy-N, was found to be excessively embrittled by neutron irradiation and to experience shallow intergranular attack by fission—product tellurium. Subseqguently, minor composition medifications were made which appear to provide ade- quate resistance to both radiation embrittlement and tellurium attack, While extensive testing and development would still be required to fully qualify the modified Hastelloy-N as a reactor structural material, the fundamental technical issue of an adequate material appears to be re- sclved. The requiréments imposed on the graphite in a DMSR are much less se- vere than those that would apply to a high—-performance breeder reactor. The low flux levels in the core would lead to damage fluences of less than 3 x 1026 neutrons/m2 in 30 years, so some current technology graphites could last for the life of the plant. In addition, the low power demsity may eliminate the need to seal the graphite surfaces to limit xenon in- trusion and poisoning. This would substantially reduce the technology development effort associated with the manufacture of DMSR graphite. The generic safety features of a DMSR would differ significantly from those of other reactor types primarily because of the fluid nature of the fuel and the circulating inventory of fission products. Because the fuel in a DMSR would be unclad, the three levels of fission—product confinement for this system would be the RCPB and two separate levels of containment. The primary contaimment would be a set of sealed and in- erted equipment cells that would be inaccessible to personnel after the onset of plant operation. These cells would provide the principal con- finement of radioactivity in accidents involving failure of the RCPB. They could also provide auxiliary cooling of spilled fuel salt if that salt failed to flow to the cooled drain tank. Loss of coocling accidents 135 k4 with reactor scram may be relatively mild in DMSRs because of the large heat capacity and low vapor pressure of the fuel salt which inherently retains most of the fission—-product decay~heat generators. However, loss - of cooling because of blocked core fuel passages at full power could lead to some local salt boiling. A full safety analysis of the DMSR has not been performed because it would require a much more comprehensive design than is currently available. Preliminary consideration of the environmental effects of DMSRs sug- gests that such effects would generally be milder than for currently oper- ating nuclear systems. There would be little or no routine gaseous and liquid radiocactive effluents, less waste heat rejection, no shipment of radicactive spent fuel during the normal plant life, relatively little solid radioactive waste, and less impact from uranium mining. In con- trast to these more favorable features, the DMSR at end-of-life would involve a more complex decommissioning program and a larger solid waste disposal task. In addition, during operation, the retention of tritium and the relatively larger inventory of radionuclides may require extra efforts to avoid possibly unfavorable effects. In general, the antiproliferation features of the once-through DMSR appear to be relatively favorable. The entire fissile uranium inventory would be fully denatured, and there would be no comvenient means of iso- lating 233p, for decay to separated 233y, The fissile plutonium inven- tory would be small, of poor quality, and difficult to extract from the large mass of highly radicactive fuel salt. In addition, no shipments of spent fuel from the plant would occur except at the end-of-life. 6.2 Alternate DMSR Concepts Although a DMSR operating on a 30-vear, once-through fuel cycle ap- pears to have a number of attractive features, the basic concept could be adapted to a number of alternative fuel cycles., If full-scale, on- line processing of the fuel salt to remove fission products were adopted, : some likelihood exists that break-even breeding performance could be achieved. However, even without break-even breeding, the fuel charge could be recycled through several generations of reactors to greatly 136 reduce the average demand for mined uranium. Other performance improve- ments {short of break-even breeding) could be achieved by combining the on—-line fuel processing with periodic removal or reenrichment of part of the active uranium inventory. In all these coptions, the net consumption of natural uranium would become a minor factor in the application of DMSRs. Some consideration was given to fuel processing concepts that would remove only part of the soluble fission products. Such processes appear to offer few (if any) advantages over either the unprocessed or the fully processed apprcaches. 6.3 Commercialization Considerations Since the MSR concept was under study and development for nearly 30 vears, most of the relevant areas of the required technology have received at least some attention. After the successful operation of the MSRE, a limited amount of design effort was expended on a commercial-size MSBR; that effort was discontinued in 1973, The technology development work proceeded in parallel with the design studies up to that time. A small development effort (without design support) was resumed in 1974 and can- celled again in 1976, This work, despite its limited scope, provided an engineering-scale demonstration of tritium management in the secondary salt and significant progress toward the definition of an acceptable structural alloy for molten-salt service. Work was under way toward dem- onstration of some of the chemical processing operations when the program ‘was ended. Aside from the technical progress, the last development activity produced a comprehensive plan for the further development of MSRs, which served as the basis for the proposed DMSR development plan and schedule, This plan suggests that the commercialization of DMSRs could proceed via three reactor projects: (1) a moderate-sized (100~ to 200-MWe) molten— salt test reactor that could be authorized in 1985 and become operational in 1995, (2) an intermediate-sized commercial prototype plant authorized in 1995 and operating in 2005, and {(3) a first standard-design DMSR to operate in 2011. A preliminary estimate for the cost of this program, 137 kT including $700 million for the concurrent base development work, is $3750 million (in 1978 dollars). A preliminary estimate of the construction cost for a "standard” DMSR {(neglecting contingencies, escalation, and interest during construction) yielded about $650/kWe in 1978 dollars. This compares with about $600/kWe for a PWR and $380/kWe for a coal plant (without flue-gas cleaning) esti- . mated on the same basis., The DMSR capital estimate did not include the cost of on-site salt treatment facilities or the costs of salt and fuel inventories; these quantities are all included in the fuel cyecle costs. The estimated nonfuel O&M costs were 2.82 mills/kWh, and fuel cycle costs were 5.3 mills/kWh. The cost of decommissioning a DMSR was estimated to be about 10% higher than that for a comparably sized LWR. The licensing of MSRs has not been seriously addressed because no proposal to build a reactor beyond the MSRE was ever supported and none cf the conceptual design studies proceeded tc that level. However, a number of new licensing issues clearly would have to be addressed. Be- cause the three levels of fission-product confinement in a DMSR would differ from those in a solid-fueled system, demonstrating compliance with the risk objectives rather than specific hardware designs in established licensing criteria presumably would be necessary. Preliminary studies suggest that the risks associated with the operation of MSRs may be lower than those for LWRs, while risks during maintenance and inspection of the reactor system may be higher. 6.4 Conclusions - The preliminary studies of DMSRs described previously indicate that these reactors could have attractive performance and resource utilization features while providing substantial resistance teo the further prolifera- tion of nuclear explosives. In addition, the environmental and safety features of DMSRs generally appear to be at least as favorable as those of other nuclear power systems, and the system economic characteristics 5 - are attractive. While a substantial RD&D effort would be required to commercialize DMSRs, there are no major unresclved issues in the needed technology. Thus, a commercial DMSR without on—-line fuel processing 138 probably could be developed in about 30 years; with additional support St for RD&D, the technology for on-line fuel processing coculd be develbped on about the same time schedule. Although the DMSR characterizations presented in this report are ap- proximate, they provide as much detail as is justified by the very pre- liminary status of the system conceptual design. Any effort to substan- tially improve the quality and detail of the characterizations would have to be accompanied by a significant system design effort oriented toward é specific DMSR power plant. Costs and times required for such studies would be several times as large as those for the preliminary work and probably could be justified only if a national decision were made to re- establish a federally funded MSR program. Any MSR program of substantial size presumably would include an RD&D effort of some size to support effective pursuit of program goals. This work, in turn, would be complemented by the design studies which would help to define RD&D tasks and focus the entire effort. The combination would allow the attaimment of objectives on the shortest practical time schedule. 7 From the preliminary studies reported, a once-~through DMSR without on-line fuel processing apparently would be the most reasonable choice for development if an RD&D program were established. However, parallel development of the techmology for continuous fuel processing would add only moderately to the total program cost and could provide the option of a more resource-efficient (and possibly a cheaper) fuel cycle. i 139 G ACKNOWLEDGMENT The authors want to acknowledge the extensive and able support of the Union Carbide Corporation Computer Sciences Division, especially J. R. Knight, N. M. Greene, 0. W. Hermann, R. M. Westfall, W. E. Ford III, B. R. Diggs, and L. M. Petrie. The nuclear calculations depended heavily on their advice and help. The estimate of capital, operating, and decommissioning costs for the DMSR were prepared and reported by M. L. Myers. 10. 11, 12, 13. 14, 140 REFERENCES i L. E. McNeese and M. W. Resenthal, "MSBR: A Review of the Status and Future,” WNucl. News 17(12), 52—58 (September 1974}, R. W. Roberts, Energy Research and Development Administration, let— ter to H., Postma, Oak Ridge National Laboratory, March 1, 1976. H. F. Bauman et al., Molten-Salt Reactor Concepts with Reduced Po- tential for Proliferation of Special Nuclear Materials, Institute for Energy Analysis, ORAU/IEA(M) 77-13 (February 1977). Gerald R. Ford, "Statement by the President on Nuclear Policy,” Of- fice of the White House Press Secretary, October 28, 1976. Department of Energy, Nowmproliferation Alternative Systems Assess- ment Program Plan, Assistant Secretary for Energy Technology Nu-— clear Energy Programs, Office of Fuel Cycle Evaluation {(April 1978). Jimmy Carter, "Statement on Nuclear Power Policy,” press release dated April 7, 1977, Dept. State Bull. 76, 42933 (May 2, 1977). Executive Office of the President, The WNational Energy Plan, April 29, 1977. Roy C. Robertson (Ed.), Conceptual Design Study of a Single-Fluid Molten-Salt Breeder Reactor, ORNL-4541 (June 1971). J. R. Engel et al., Molten-Salt Reactors for Efficient Nuclear Fuel Utilization Without Plutonium Separation, ORNL/TM-6413 (August 1978). G. T. Mays, A. N. Smith, and J. R. Engel, Distribution and Behavior of Tritium in the Coolant-Salt Technology Facility, ORNL/TM-5759 (April 1977). N. M. Greene et al., AMFX: A Modular Code System for Generation Coupled Multigroup Neutron-Gamma Libraries from ENDF/B, ORNL/TM-3706 (March 1978). N. M. Greene, Unicon Carbide Corperation, personal communication to We A. Rhoades, Oak Ridge National Laboratory, September 19877. J. R. Knight, Union Carbide Corporation, perscnal communication to W. A. Rhoades, Oak Ridge Natiomal Laboratory, June 1978. D. S. Bost, Selected Computer Codes and Libraries: Volume II, APC, £ Multigroup Transport Buckling Iteration Code, Rockwell Inter- s national, AI-AEC-13076, vol. II (May 1973). 141 s 15. M. J. Bell, ORIGEN — The ORNL Isctope Generation and Depletion Code, ORNL-4628 (May 1973). 16. J. R. Knight, Union Carbide Corporation, personal communication to W. A. Rhoades, Oak Ridge National Laboratory, June 1978, 17. Argonne National Laboratory, Reactor Physics Constants, ANL-5800 (July 1963), 18. F. W. Dittus and L. M. K. Boelter, University of California Publs. Eng. 2, 443 (1930). - 19. W. R. Grimes, "Molten-Salt Reactor Chemistry," Wuecl. Appl. Technol. 8, 137 (February 1970). 20, W. R. Grimes et al., "Fuel and Coolant Chemistry,” chap. 5, pe 95 in The Development Status of Molten-Salt Breeders, ORNL-4812 (August 1972), 2l. L. E., McNeese et al., Program Plan for Development of Molten-Salt Breeders, ORNL=5018 (December 1974). 22, P. N. Haubenreich and J. R. Engel, "Experience with the Mclten-Salt Reactor Experiment,” Nucl. Appl. Technol. 8, 118 (February 1970}. 23. 8. Cantor et al., Physical Properties of Molten-Salt Reactor Fuel, Coolant and Flush Salt, ORNL/TM-2316 (August 1968). 24, S. Cantor, Density and Viecosity of Several Molten Fluoride Mix- tures, ORNL/TM-4308 (March 1973). 25, C. F. Baes, Jr., p. 617 in "Symposium on Reprocessing of Nuclear Fuels,” ed. by P. Chiotti, vol. 15 of Nuelear Metallurgy, USAEC- CONF-690801 (1969). 26, C. F., Baes, Jr., "The Chemistry and Thermodynamics of Molten—Salt Reactor Fuels,"” J. Nuel. Mat. 51, 149 (1974), 27. C. E.-Bamberger, R. G. Ross, and C. F. Baes, Jr., "The Oxide Chemis- try of Plutonium in Molten Fluorides and the Free Energy of Forma- tion of PuF, and PuF,," J. Inorg. Nucl. Chem., 33, 776 (1971). 28. J. K. Dawson et al., "The Preparation and Some Properties of Pluto- nium Fluorides,” J. Chem. Soc., 558 (1954). 29, C. E. Bamberger et al., "Absence of an Effect of Oxide on the Solu- bility and the Absorption Spectra of PuF; in Molten LiF-~BeF,~-ThF, and the Instability of Plutonium (III) Oxyfluorides,” J. Inorg. Nuel. Chem. 33, 3591 (1971). 30, "Molten Salt Reactor Concept,” Quarterly Report for Period Ending July 31, 1971 , Atomic Energy Commission, Bombay, India, NP-19145. 31, 32, 33. 34, 353 35, 37. 38. 39. 4G, 415 42, 43. 142 C. Je Barton et al., Solubility of Cerium Trifluoride in Molten Mix- tures of Lithium, Beryllium, and Thorium Fluorides," Inorg. Chem. 9, 307 (1970). C. ¥, Baes, Jr., J. H. Shaffer, and H. F. McDuffie, UG02 and Zr0Q; Be- havior in Molten Fluorides," Trans. Am. Nucl. Scc. 6, 393 (1963}, K. A. Romberger, C. ¥. Baes, Jr., and H. H. Stone, "Phase Equilib- rium Studies in the UO,-Zr0, System,” J. Inorg. HNuel. Chem. 29, 1619 (1967). C. E. Bamberger and C. F. Baes, Jr., "The Exchange of U“T and Th“T Between Molten LiF-BeF,-ThF, -UF, and (U-Th)oz Solid Solutions,” Jo Nuel. Mat. 35, 177 (1970). C. E. Bamberger, R. G. Ross, and C. F. Baes, Jr., "Oxide Chemistry of Plutonium in Molten Fluorides and the Free Energy of Formation of PuF3 and PuF,," J. Inorg. Nuecl. Chem. 33, 767 (1971). C. F. Baes, Jr., p. 617 in "Symposium on Reprocessing of Nuclear Fuels,” ed. by P. Chiotti, vol. 15 of Nuelear Metallurgy, USAEC-CONF 690801, 617 (1969). R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr., "The Oxide Chemis- try of Protactinium in Molten Fluorides,” J. Inorg. Nucl. Chem. 35, 43349 (1973)., C. E. Bamberger, R. G, Ross, and C. F. Baes, Jr., Molten-Salt Reac- tor Program Semiarmual Progress Report for Period Ending February 29, 1972, CRNL-4782, p. 72. C. E. Bamberger et al., Molten-Salt Reactor Program Semiarnmual Progress Report for Period Ending August 31, 1871, ORNL-4728, p. 62. C. E. Bamberger, R. G. Ross, and C. F. Baes, Jr., Molten-Salt Reac- tor Program Semiannual Progress Report for Period Ending February 28, 1971, ORNL-4676, p. 119, C. E. Bamberger, R. G. Ross, and C. F. Baes, Jr., Molten-Salt Reac- tor Program Semiannual Progress Report for Period Ending August 31, 1970, ORNL-4622, p. 92. C. E, Bamberger, R. G. Ross, and C. F. Baes, Jr., Molten-Salt Re- actor Program Semiannual Progress Report for Period Ending February 28, 1970, ORNL-4548, p. 142, M. J. Bell, D. D. Sood, and L. E. McNeese, Molten-Salt Reactor Program Semiannual Progrese Report for Period Ending February 29, 1972, ORNL-4782, p. 234. 143 S 44, 0. K. Tallent and L. M. Ferris, Molten-Salt Reactor Program Semi- annual Progress Report for Period Ending February 29, 1972, ORNL- 4782, p. 210. 45. A. L. Mathews and C. F. Baes, Jr., "Oxide Chemistry and Thermo- dynamics of Molten LiF-BeF, Solution,” Inorg. Chem. 7, 373 (1968). 46, H. E. McCoy, Jr., Status of Materials Development for Molten-Salt - Reaetors, ORNL/TM=-5920 (January 1978). 47, H. E. McCoy et al., "New Developments in Materials for Molten-Salt Reactors,” Nucl. Appl. Technol. 8, 156 (February 1970), 48. H. E. McCoy, "Materials for Salt-Containing Vessels and Piping,” chap. 7, p. 195 in The Development Status of Molten-Salt Breeders, ORNL-4812 (August 1972), 49, G. M. Adamson, R. S. Crouse, and W. D. Manly, Interim Report on Cor- rogion by Alkali-Metal Fluovrides: Work to May 1, 1953, ORNL-2337. 50. G. M. Adamson, R. S. Crouse, and W. D. Manly, Imterim Report on Cor- rogion by Zireconium-Base Fluorides, ORNL-2338 (January 3, 1961), 51. W. B. Cottrell et al., Disassembly and Postoperative Examination of the Aireraft Reactor Experiment, ORNL-1868 (April 2, 1958). 52. W. D. Manly et al., Aireraft Reactor Experiment — Metallurgical As- pecte, ORNL-2349, pp. 2=24 (December 20, 1957). 53. W. D. Manly et al., "Metallurgical Problems in Molten Fluoride Sys- tems,” in Progr. Nucl. FEnergy Ser. IV Technology, Engineering and Safety 2, 164 (1960). 54. W. D. Manly et al., Fluid Fuel Reactors, pp. 595604, ed. by James A. Lane, H. G. MacPherson, and Frank Maslan, Addison-Wesley, Read- ing, Pa. (1958}. 55. Molten-Salt Reaetor Program Status Report, ORNL-CF-58-5-3, pp. 112—-13 (May 1, 1958). 56. J. H, DeVan and R. B. Evans, III, pp. 55779 in Conference on Cor- rosion of Reactor Materials, June 4—8, 1962, Proceedings, Vol. IT, International Atomic Energy Agency, Vienna, Austria, 1962. 57. F. F. Blankenship, Molten-Salt Reactor Program Semiannual Prog- ress Report for Period Ending July 31, 1964, ORNL-3708, p. 252Z. 58. S. S. Kirslis and F. F. Blankenship, Molten-Salt Reactor Program Semiarnual Progress Report for Pertod Ending August 31, 1968, ORNL-4344, p. 115. 59. 60. 6l. 62. 63, 64, 65, 66. 67. 68, 69. 70. 7%, 72, 73, 144 L. M. Toth and L. 0. Gilpatrick, "Equilibria of Uranium Carbides in Molten Solutioms of UF3 and UF, Contained in Graphite at 8530°K," J. Inorg. Nuel. Chem. 35, 1509 (1973). | L. M. Toth and L. O. Gilpatrick, "Temperature and Solvent Effects on the Equilibrium of Dilute Uranium Trifluoride Sclutions Contained in Graphite,"” J. Phys. Chem. 77, 2799 (1973). L. M. Toth and L. 0. Gilpatrick, The Equilibrium of Dilute UF; Solu~ tions Contained in Graphite, ORNL/TM-4056 (December 1972). E. K. Storms, The Refractory Carbides, vol. 2, pp. 171—213, Academic Press, New York, 1967. W. R. Grimes, "Materials Problems in Molten-Salt Reactors,” in Mate- rials and Fuels for High Temperature Nuclear Energy Applications, ed, by M. T. Simnad and L. R. Zumwalt, The MIT Press, Mass., 1964, W. R. Grimes, Molien-Salt Reactor Program Semiannual Progress Report for Period Ending July 31, 1964, ORNL-3708, p. 214. W. R. Grimes et al., "Chemical Aspects of Molten Fluoride—-Salt Re- actor Fuels,” in Fluid Fuel Reactors, ed. by J. A. Lane, H. G. MacPherson, and F., Maslan, Addison-Wesley, Reading, Pa., 1958. Molten-Salt Reactor Program Progress Report for Period from Novem- ber 1, 1960 to February 29, 1961, ORNL-3122, p. 100. Molten-Salt Reactor Program Progress Report for Period from Mareh 1 to August 31, 1961, ORNL-3215, p. 117. Molten-Salt Reactor Program Semiannual Progress Report for Period Ending August 31, 1962, ORNL-3369, p. 100. Molten-Salt Reactor Program Semiannual Progress Report for Period Ending January 31, 1863, ORNL-341%9, p. 80. W. R. Grimes, Radiation Chemistry of Molten-Salt Reactor System, ORNL/TM-500 {(March 31, 1963). Reactor Chemistry Division Annual Progress Report for Period Ending January 31, 1962, ORNL-3262, p. 20. E. L. Compere et al., Fission Product Behavior in the Molten-Salt Reactor Experiment, ORNL-4865 (October 1975). F. F. Blankenship et al., Reactor Chemistry Division Annual Progress Report for Period Ending January 31, 1963, ORNL=-3417, p. 17. 145 E 74, W. R. Grimes, N, V. Smith, and G. M. Watson, "Solubility of Noble Gases in Molten Fluorides:; I. In Mixtures of NaF=ZrFy, {(53-47 mole %) and NaF-ZrF,-UF, (560-46-4 mole %)," J. Phys. Chem. 62, 862 (1958). 75. M. Blander et al., "Solubility of Noble Gases in Mclten Fluorides; II. In the LiF-NaF-KF Eutectic Mixtures,” J. Phys. Chem. 63, 1164 (1959). 76. G. M. Watson, W. R. Grimes, and N, V. Smith, "Solubility of Noble Gases in Molten Fluecrides; II. In Lithium—Beryllium Fluoride,” J. Chem. Eng. Data 7, 285 (1962). 77. R. B. Briggs and R. B. Korsmeyer, Molten-Salt Reactor Program Semi- annual Progress Report for Period Ending February 28, 1970, ORNL- 4548, p. 53. 78, A. P. Malinauskas and D. M. Richardson, "The Sclubilities of Hydro- gen, Deuterium, and Helium in Molten Li,BeF,," Ind. Eng. Chem. Fund. 13(3), 242 (1974). 79. D. R. Cuneo and H. E. Robertson, Molten-Salt Reactor Program Semi- annual Progress Report for Period Ending August 31, 1368, ORNL-4344, pp. 14146, 80. H. E. McCoy and B. McNabb, Imtergranular Cracking of INOR-§ in the MSRE, ORNL-4829 (November 1972). 81. D. L. Manning and G. Mammantov, Molten-Salt Reactor Program Semi- annual Progress Report for Period Ending February 29, 1976, ORNL- 5132, p. 38. 82, J. R. Keiser, Status of Tellurium-Hastelloy-N Studies in Molten Fluoride Salts, ORNL/TM-6002 (October 1977). 83. W. R. Grimes et al., "High Temperature Processing of Molten Fluoride Nuclear Reactor Fuels,” Nuel. Eng. Part VII 55, 27 (1959). 84. G. J. Nessle and W. R. Grimes, "Production and Handling of Molten Fluorides for Use as Reactor Fuels,” Chemical Engineering Progress Symposium Series 56(28) (1960). 85. J. H. Shaffer, Preparation and Handling of Salt Mixtures for the Molten-Salt Reactor Experiment, ORNL-4616 (January 1971). 86. R. B. Lindauer, Molten-Salt Reactor Program Semiannual Report for Period Ending August 31, 1971, ORNL-4728, p. 226, 87. R. B. Lindauver, Molten-Salt Reactor Program Semiannual Report for Period Ending February 28, 1971, ORNL-4676, p. 269. 88. 89. 90. 91. 92. 93O %. 95. 96. 97. 98. 99. 100, 101, 146 A. D. Kelmers et al., Evaluation of Alternate Secondary (and Ter- tiary) Coolants for the Molten-Salt Breeder Reactor, ORNL/TM-5325 (April 1976). C. J. Barton et al., Molten-5alt Reactor Program Semiarmual Progress Report for Period Ending February 29, 1968, ORNL~4234, p. 166. E. L. Compere, H. C. Savage, and J. M. Baker, "High Intensity Gamma Irradiation of Meclten Sodium Fluoroborate—Sodium Fluoride Eutectic Salt,"” J. Wuel. Mat. 34, 97 (13%70). J. W. Cooke, Molten-Salt Reactor Program Semiannual Progress Re- port for Period Ending August 31, 1969, ORNL-4449, p. 92. D. M. Richardson and J. H. Shaffer, Molten-Salt Reactor Program Semiannual Progress Report for Period Ending February 28, 1870, ORNL-4548, p. 136, S. Cantor and R. M. Waller,'MbZten—SaZt Reactor Program Semiannual Progress Report for Period Ending February 29, 1872, ORNL-4782, p. 63. G. T. Mays, Molten-Salt Reactor Program Semiannual Progress Report for Period Ending February 28, 1975, ORNL-5047, p. 8. S. Cantor and R. M. Waller, Molten-Salt Reactor Program Semiannual Progress Report for Period Ending August 31, 1970, ORNL-4622, p. 79. J. B, Bates et al., Molten-Salt Reactor Program Semiannual Progress Report for Period Ending February 28, 1971, ORNL-4676, p. %4, S. Cantor and R, M. Waller, Molten-Salt Reactor Program Semiannual Progress Report for Period Ending February 28, 1971, ORNL-4676, p. 88. J. B. Bates, J. P. Young, and G. E. Boyd, Molten-Salt Reactor Pro- gram Semiannual Progress Report for Period Ending February 29, 1972, ORNL-4782, p. 59. - J. P. Young, J. B. Bates, and G. E. Boyd, Molten-Salt Reactor Pro- gram Semiannual Progress Report for Period Ending August 31, 1872, ORNL—4832 2 p ° 52 ® Wo L. Carter and E, L. Nicholson, Design and Cost Study of a Fluo- rination—Reductive Extraction-Metal Trangfer Processing Plant for the MSBR, ORNL/TM-3579 (May 1972). M. W. Rosenthal et al., The Development Status of Molten-Salt Breeder Reactors, ORNL-4812 (August 1972), 147 Hi 102, J. R. Engel et al., Development Status and Potential Program for Development of Proliferation-Resistant Molten-Salt Reactors, ORNL/ TM—~6415 (March 1979). 103, C. R. Hudson II, CONCEPT-5 User's Manual , ORNL=5470 (January 1979). 104, M. L. Myers and L. C. Fuller, 4 Procedure for Fstimating Nonfuel Operation and Maintenance Costs for Large Steam-Electriec Power Plants, ORNL/TM-6467 (January 1979). 149 Appendix A COMPARATIVE REACTOR COST ESTIMATES 2 151 e Table A.l. Cost estimates for the DSMR, PWR, and coal plants {(Thousands of 1978.0 dollars) Account No. Twoe Three- Ttem DMSRY PWR? Coal?s? digit digit Direct costs 20 Land and land rights ' 2,000 2,000 2,000 21 Structures and improvements 211 Yard work 10,000 10,103 5,990 212 Reactor containment building 44,000 39,017 Cmit 213 Turbine building ' 14,000 12,820 10,337 215 Auxiliary building(s) 24,000 9,297 Cmit 216 Waste process building {(in 215} 8,841 Cmit 217 Fuel storage building Nad 4,928 Omit 218 Other structures? 30,000 25,952 Cmit 219 Stack (heat rejection) 2,000 NA 2,203 Account 21 subtotal 124,000 110,958 Omit 22 Reactor plant equipment 220 Nuclear steam supply system (in 221 67,111 Omit and 222} 221 Reactor equipment 45,000 3,727 Omit 222 Main heat—-transfer system 63,000 9,873 Omit 223 Safeguards system 6,000 11,582 Omit 224 Radwaste processing 10,000 10,042 Omit 225 Fuel handling and storage 10,000 3,405 Omit ? 226 Other reactor plant equipment 30,000 19,822 Cmit 227 Reactor instrumentation and control 10,000 7,779 Omit 228 Reactor plant miscellaneous items 6,000 5,497 Omit Account 22 subtotal 180,000 138,838 Omit 23 Turbine plant equipment ‘ 231 Turbine generator 43,000 61,943 42,299 232 {Changed to account 26) 233 Condensing systems 12,000 15,257 12,022 234 Feed-heating system 21,000 15,315 14,519 235 Other turbine plant equipment 18,000 15,496 14,924 236 Instrumentation and control 2,000 1,336 837 237 Turbine plant miscellaneocus items 4,000 3,590 3,190 Account 23 subtotal 100,000 112,937 87,791 24 Electric plant equipment 241 Switchgear 6,000 5,739 4,081 242 Station service equipment 14,000 9,419 3,949 243 Switch boards 1,000 701 721 244 Protective equipment 2,000 1,770 1,879 245 Electrical structure and wiring 12,000 10,215 9,422 246 Power and control wiring 19,000 15,737 10,805 Account 24 subtotal - 54,000 43,581 30,857 e 152 Table A.l (continued) Account No. b b,e Twom Throo— Item DMSR? PWR Coal digit digit 25 Miscellaneous plant equipment 251 Transportation and life equipment 3,000 2,617 1,606 252 Air, water, and steam service 10,000 7,664 6,034 system 253 Communication equipment 2,000 1,524 691 254 Furnishings and equipment 1,000 1,041 898 255 Waste water treatment equipment 1,000 NA 1,351 Account 25 subtotal 17,000 12,846 10,580 26 Main condenser heat rejection 14,000 21,968 14,003 Accounts 2026, total direct 491,000 443,128 Omit costs Indirect costs 91 Construction services o911 Temporary construction facilities 26,000 25,801 14,348 912 Censtructicn tools and equipment 24,000 21,878 11,285 913 Payroll, insurance, and social 25,000 22,460 13,363 security taxes Account 9! subtotal 75,000 76,139 38,996 92 Home-office engineering services 921 Home-cffice services 49,008 14,917 922 Home-office quality assurance 2,333 NA 923 Home-office construction management 1,338 1,192 Account 92 subtotal 53,000 52,678 16,106 93 Field-office engineering and services 931 Field-office expenses 3,000 3,180 824 932 Field job supervision 22,000 19,188 8,732 333 Field quality assurance/quality 5,000 4,683 180 centrol 934 Plant start-up and test 4,000 2,853 343 Accounts 91-93 34,000 28,904 10,079 Total indirect costs 162,000 152,722 65,184 Total capital costs, direct and 653,000 595,850 Omit indirect ?Estimated by M. L. Myers. bEstimated from CONCEPT V. Cselected accounts. dNot applicable. e . . s . . For example, control room, administration building, fire tunnels, sewage, holding pond, diesel-generator building, receiving, and guard. 153 Table A.2. Cumulative cash flow for DMSRZ Date Cost to dateb (millions of dollars) 1978.0 G 1979.0 5 1980.0 10 1981.0 23 1982.0 55 1983.0 143 1984.0 313 1985.0 458 1986.0 586 1987.0 637 1988.0 653 Ynit 1; 1000-MWe DMSR power plant at Middletown; cost basis is year of steam supply system purchase (1978.0); con- struction permit is 1978.0; commercial operation is 1988.0. bTotal cost incurred to date excludes interest and es- calation charges. A, 1. 2e 3. 4-8. i0. 11. 12, 13. 14-18, 19-23. 24, 25. 26-30. 31. 32. 33. 34. 35. 36. 67. 68. 69. 70. 71, 72, 73. 74, 75, 76. 155 ORNL/TM=7207 Dist. Category UC-76 Internal Distribution T. D. Anderson 37. R. E. MacPherson Seymour Baren 38-42, H. E. McCoy D. E. Bartine 43, L. E. McNeese H, F. Bauman 44-48, W, A. Rhoades H, W. Bertini 49, J. L. Rich H. I. Bowers 50. P. S. Rohwer Je C. Cleveland 5l. M. W. Rosenthal T. E., Cole 52. R. T. Santoro S. Cantor 53. Dunlap Scott J. F. Dearing 54, 1. Spiewak J. R. Engel 55. H. E. Trammell D. E. Ferguson 56, D. B. Trauger M. H. Fontana 57, J. R. Weir W. R. Grimes 58. J., E. Vath R. H. Guymon 59. R. G. Wymer W. O, Harms 60. ORNL Patent Office J. F. Harvey 61-62. Central Research Library H. W. Hoffman 63. Document Reference Section P. R. Kasten 64-65 Laboratory Records Department Milton Levenson 66. Laboratory Records (RC) External Distribution Director, Nuclear Alternative Systems Assessment Division, Depart- ment of Energy, Washington, DC 20545 E. G. Delaney, Nuclear Alternative Systems Assessment Division, Department of Energy, Washington, DC 20545 H. Kendrick, Nuclear Alternative Svstems Assessment Division, De- partment of Energy, Washington, DC 20545 Of fice of Assistant Manager for Energy Research and Development, Department of Energy, ORO, Oak Ridge, TN 37830 Director, Nuclear Research and Development Division, Department of Energy, ORO, Oak Ridge, TN 37830 G. Nugent, Burns and Roe, Inc., 185 Crossways Park Drive, Wood~- bury, NY 11797 R. A. Edelman, Burns and Roe, Inc., 496 Kinderkamack Road, Ora- dell, NJ 07649 W. R. Harris, Rand Corporation, 1700 Main Street, Santa Monica, CA 90406 S. Jaye, S. M. Stoller Corp., Suite 815, Colorado Building, Boul- der, CO 80302 W. Lipinski, Argonne National Laboratory, 9700 South Cass Ave., Argonne, IL 60439 156 77. H. G. MacPherson, Institute for Energy Analysis, Oak Ridge Asso— w ciated Universities, Oak Ridge, TN 37830 _ 78. R. Omberg, Hanford Engineering Development Laboratory, P.C. Box 1970, Richland, WA 99352 79. E. Straker, Science Applicaticns, 8400 Westpark Drive, McLean, VA i 22101 80, A. M. Weinberg, Institute for Energy Analysis, Oazk Ridge Associ- ated Universities, Oak Ridge, TN 37830 81-181, For distribution as shown in DCE/TIC-~4500 under category UC-76, Molten—Salt Reactor Techmnology # U.S. GOVERNRENT PRINTING OFFICE: 1380-640-245/150