ré 7&3&. & S e ‘. ', __: ..i\ c 7t DATE: SUBJECT: T0: FROM: OAK RIDGE NATIONAL LABORATORY EXTERNAL TRANSMITTAL W= CENTRAL FILES ~ Oak Ridge, Tenne.s see : 59_12_61; ‘ | (Revised) Operated by . AUTHORIZED UNION CARBIDE NUCLEAR COMPANY Division of Union Carbide Corporation _ 0 R N l NUMBER Jenuary 12, 1960 | COPY NO. 7 ,Mbltefi-Salt Breeder Reactors Distribution H. G. MacFherson iflbstract ' The problems involved in building e molten-salt thermal-breeder reactor are reviewed, and it is concluded that the most feasible construction iz an externally-cooled reactor with the fuel salt paesing through the reactor core in graphite tubes. A reactor with 15% of the core volume occupied by fuel salt and 5% occupied by fertile salt would have & net breeding ratio of about 1.06. The specific power ie about 3.0 Mw(th) per kg of U-233, U-235, and Pe in the entire reactor and chemical processing system. The resulting doubling time is 13 full-power years. The cost of the fuel eycle for & 1000-Mw(E) station with this breeding performance ig estimated to be 1.2 mills/kwhr. The performance in terms of ‘material utilization is an output of 1.18 Mw(E) per kg of U-233, ' U-235 and Pa, and 3.2 Mw(E) per metric ton of thorium. The latter | figure could be increased by & factor of two at & sacrifice of 0.01 in breeding ratio. NOTICE . This document contains information of ‘a preliminary nature and was prepared primarily for internal use at the Oak Ridge National Laboratory. It is subject to revision or correction and therefore does not represent a final report. The information is not to be abstracted, reprinted or otherwise given public dissemination without the approval of the ORNL patent branch, Legal and information Control Department. ? — ¥ MOLTEN-SALT BREEDER REACTORS The purpose of this memo 18 to examine and swmsarize the status of the molten-selt reactor as meeting the requirements of a breeder with a doubling time of not more than 25 years., . Included are a discussion of: 1, The practicebility of different types of breeder reactor construction. 2. The power density attainable in the fuel selt. 3. The status and cost of the required chemicel processing schenme. 4. The breeding gain, specific pover, and doubling time coneistent with reasonsble assumptions concerning items (1), (2), ana (3). 5. The feasibility end cost of molten-aalt‘reactors. The.contents of this memo have not been subjected to analysis by the Thermal- Breeder Eveluation group. Their work will be largely independent of this, and their results, vhen available, will take precedence over the numbers used in this memo. _ I. Reactor Construction Three'types'bf'breeder regctor construction are discussed: the unit-fuel- tube construction, the graphite-core-shell construction, and an internally- cooled construction. Unit Fuel Tube Construction - The type of construction that is believed to be most practical at present for a molten-salt breeder reactor is one In which the fuel salt passes through the reactor in graphite tubes. Graphite moderator is massed outside of the fuel tubes in the core region of the reactor, and blanket salt containing thorium surrounds the core. The blanket galt also passes through small passages in the moderator graphite and cools it. Fig. 1 (ORNL-IR-Dwg. 42242) gives a schematic representation of one edge of such & reactor, showing a single fuel tube, one of many. Although Fig. 1 shows a re-entrant graphite tube with both inlet and outlet at the bottom end to avoid problems of differential thermal expansion, it may also be possible to use a construction in which the qul tubes go straight through the reactor. The fuel tubes would be manufactured from fine-gxained extruded graphite rendered impervious by one of a number of treatments‘available. This type. of graphite has been shown to be the most impervious to molten salts; one E w9 o & +@ent of salt by volume when pressures of up to 150 psi are applied; in fact, . blenket salt will be maintained under slight pressure with respect to the such grade hae been used in contact with flowing salt streams for a year with no evidence of attack or bulk penetration by the salt. Separate tests have indicated that such a grade of graphite will soak up less than one per- | one grade picked up less than 0.2% by volume of salt. Tubes 3-3/4 #n. ID x 5 in. OD are on order and will be tested within a few months. The moderator graphite will be in the blanket salt environment, end the fuel salt so that any leskage that develops will be from blanket salt to fuel salt. Ieakage can be tolerated provided it is at a rate that 1s small compared to the rate of chemicel processing of the fuel salt. The moderator graphite could also be made from fine-grained extruded graphite to keep pick- up of salt in it a¥'es low & level as possible. By confining the fuel to tubes end pressurizing the blanket salt with respect to the fuel ealt, fis- sloning within graphite will be kept to & minimum. As a result there is little reason to expect buildup of fission-product poisons in the graphite. In the re-entrant fuel-tube construction, two metal-to-graphite connections are necessary. The connection to the central graphite tube need only be a mechanicelly sound connection, sitwh as & slip fit, since a small leakage here would only bypass s small amount of fuel from going through the reactor core. The connection of the outer fuel tube to the metal wall of the reactor should be reasonably tight, with leakage small relstive to chemical processing rate of the fuel. The three possibilities for this joint are a flanged joint - with a mechanical pressure seal, a frozen-salt plug seal, and a brazed metal- to-graphite tube Jjunction. Babecock and Wilcox have experimented with pressure type flanged Joints with some succees, and it is presumed that this will be a feasible solution to the problem. The testing of the freeze plug technigue is under way at Ozk Ridge, and early indications are that it will be possible t0o braze graphite to INOR-8, probably by the use of pure molybdenum as en in- termediate material to provide a match to the thermal expansion coefficient to the graphite. _ Graphite Core Shell Construction - A simple construction for & small twa- ‘region reactor with a graphite core shell is shown in Fig. 2 (ORNL~-IR-Dwg. 37258). As ghown in the drawing, the core is made from three large blocks of graphite, a top header, & bottom header, and a center section. The diameter of the core is approximated 54 in., and the height of the center section would be about 40 in. It is proposed that these graphite parts be made from large-size molded-graphite blocks, that the blocks be rough machined to shape, and that they then be impregnated and tmmted to make them nearly impermeable to molten salts. It is possible that final machining on the in- ternal parts would be done before treatment and the parts clamped together during treatment to cement the headers on and yield a monolithie block con- _ struction. o L This monolithic construction is an altérnate to having the three graphite blocks as separate pieces, clamped together and held in place by springs. The monolithic structure is considered more desirable, but makes the impervious treatment more difficult. The pressure contact should be sat- isfactory, at least for the initiel reactor operation, on the basis of "Babecock and Wilcox work. It is possible that distortions produced by shrinkege accompanying radiation damage would reduce the effectiveness of the seals. A greater worry is the effect of shrinkage on the internal portion of the core. If trouble were encountered here, the interior of the core could be made up of smaller graphite pieces, such as sticks of extruded graphite. - - B . . Grephite has been made in larger sizes than called for in this reactor, but not of the small grain size required to render it impervious. It is believed that sultable material has been produced in diemeter of 39-1/2 in. and in thicknessges of up to 20 in. Samples of this graphite are now being procured, end teste of their penetrability by molten salts will be completed " during FY 1960. If this graphite’appears suitable, it is believed that it can be made in larger sizes, up to 5 £t in diameter, but at considerable cost in production equipment and in development expense. The development of the larger block graphite for molten-salt reactors is not now planned, but would ~ be a part of the cost of the first full-scale breeder reactor of this type. (There is a possibility that the development of such graphite might be under- teken by the Defense Department for other purposes before this time.) Thies design of reactor requires a seal between the '¥raphite header and the INOR-8 pipe passing into the blanket vessel. This joint need not be a her- metic one, but should limit the leakage of blanket salt into the fuel to some small fraction of the core processing rate. The problems of this joint ere the same as those involved in the fuel tube construction. The permissible pickup of molten salt by the graphite depends on the rates of diffusion of uranium into the salt that is in the graphite and of the fission products out of the graphite into the main fuel stream. Some infor- mation on this subject will be determined in FY 1960 in the capsule experiments &t the MTR. In the meantime, a reasonable assumption is that the diffusion of fission products out balances the diffusion of them into the grephite. For a poison effect of one percent, the pickup of fuel salt into the grephite should ‘be less than one or two volume percent, depending on the volume fraction of - fuel passages in the core. If the large molded graphite turns out to have a plckup of less than one percent, then it should be suitable for use as the bulk - of the core graphite. If, on the other hand, it picks up more than 2% -galt by volume, it would be preferable to use a hollow core shell with an interior - construction of extruded graphiterstickSa. As previously indicated, such ex- . ‘truded grades have been shown to pick up a satisfactorily low level of salt. -I! i‘-n--" ‘ oY .0! Internslly-Cooled Reactor - Various designs of internslly-cooled molten- salt reactors have been suggested. One of the simplest is shown in concept in Fig. 3. In this concept, the fuel is conteined in graphite tubes about 0.54n.ID x 0.7 in. OD that extend through the moderator, well into the blenket region. The tubes are connected at each end by brazed joints to e metal header system so that the fuel can be circulated slowly to keep it uniform, to remove gaseous fission products, and to allow fuel concentration adjustment as burnup proceeds. Presumably the tubes would have graphite in- serte forcing the fuel to the periphery of the tubes in the core region and occupying most of the internal volume of the tubes in the blanket region. The heat generasted in the fuel would be transferred through the tube wall to the blanket salt which is used a8 & coolant. This would probably limit the heat generation to perhaps 50 kw per tube averaged over the reector, and would therefore require 10,000 tubes for a reactor delivering 200 elec- trical megawatts of power. Although no brazed joint that is completely satisfactory in terms of com- patibility with the salt has yet been demonstrated, there is little doubt but that such & joint will be demonstrated during this fiscal year. The use of molybdenum as en intermediate nipple connection has been demonstrated, end brazing materials that wet graphite and ere compatible with the salt have been found. Thus, in all probability, there will be no single technical element of infeasibility remaining by the end of this fiscal year for an internally-cooled reactor. Nevertheless, the concept of 10,000 tubes all maintaining their integrity during a long reactor lifetime is not very attrac- tive, at least to this writer. The adventage hoped for with internal cooling ie a greater specific power, but it is doubtful if the internally-cooled reactor cen achleve more than a factor of two in specific power over an externally-cooled molten-salt reactor. Summary - The unit-fuel-tube construction seems to be & feasible configuration Tor & breeder reactor. By the end of this fiscal year it should be poseible to specify suiteble types of graphite for both Prel tubes and moderator, and to specify a satisfactory end connection for the .tubes. The construction avolds most of the possible problems involved in sosking of fuel into the ‘ grephite since the fuel contacts only & small portion of the moderator graphite. Furthermore, it will use the type of graphite that is now deemed least likely to soak up fuel salt. The graphite-core shell construction reqpires graphite of & size and quality ~ that is not immediately avallable, and will probably not be available without the expenditure of a few million dollars of development money. If this expendi- ture were made, the reactor construction would have & good chance of success. However, the earlier availability of the fuel tube construction makes it the first choice., , & F 6 The internally-cooled reactor has some attraction in terms of higher specific power and-is made up of elemente that individually seem quite feseible. The great complexity of the core and the probable inacceseibility of it for minor repairs make it unattractive at this stage of the technology. ' The reference reactor for the remainder of this memo i1s then teken to be of the fuel tube construction, with 15 wol % of the core occupied by fuel salt, 5% by blanket salt, and the remaining 80% by graphite. If the fuel tubes are 3-3/4 in. ID, the fuel tubes will be spaced on 8-5/8 in. centers on & square &rray. | II. Power Densitz The nuclear celculation will yield the breeding ratio and the uranium con- centration required in the salt to make the reactor eriticel. In a circu- lating fuel reactor, the latter figure must be combined with the power that can be extracted per unit volume of fuel salt to yield a gross figure for specific power. The power obtainable per unit volume of salt cen be errived at in two ways: one is a general approach that looks at the fundamentsl factors involved,; and the other 1s to lay out specific designs end eee what thelr volumes are, and how much power they take care of. We will first look at the problem generally and then examine specific layouts that have been proposed. - o | A ressonable value of the power density in the fuel can be estimated from the total length of piping required in the system. The length of piping considered 1s that required to carry the fuel salt into end out of the reactor, through the blanket, through header connections, through the heat exchanger and the equivalent length of piping represented by the pump volute and expansion k. An average fluid velocity will be assumed through this riping, and a saije._t_e_mperature range between reactor entrance and exit. This | information combined with the volumetric specific heat of the palt determines ‘the amount of heat transferred per unit volume of salt. " The composition of the fuel salt will be sbout é27mole % IiF, 37 mole % BeFp, and 0.3 mole % UFy. The volumetric heat ecapacity of this mixture is about 1»2S‘ea1/cce°c at reactor temperature, or 7705_Btu/eu £t-OF. The melting 1 point of the fuel is about 850°F, and a figure about 100°F above this should ‘ be used as the minimum bulk fluid temperature. However6 the fertile salt has \ a melting point of 9759F, go that this, rather than 950°F, will be taken as the minimum temperature of the fuel salt returning to the reactor. The J maximm temperature of the salt leaving the reactor should be limited by the corrosion tolerance of the metal alloy system, and with present knowledge Gy thie is set as 1300°F since very few loqps have been run ag yet at a higher temperature. This 1e probably also & practical limit as set by the creep resistance of the alloy INOR-8. The temperature range of 975 to 1300°F 1is 3250F, but & value of 300°F will be used as & reasonsble limit, presumably from 975 to 12759F. With this AT and heat capacity, each cubic foot of fuel transporte 23,200 Btu of heat each time it makes the heat transfer circuit. It 1s difficult to set & natural 1limit on the maximum average flow velocity that can be allowed for the salt in traversing its circuit. The only known limiting factors are the pressure drop developed by the pumps and the pro- portion of power one wishes to expend in pumping. A figure .of 17 ft per - second has been picked somevhat arbitrarily for an average fluid velocity, with velocities of up to 20 ft per second in the external system and & lover velocity inside the reasctor. The major Justification for selection of this figure is that it ylelds reasonable pressure drops (v100 psi) with reasonably sigzed heat exchanger tubes and other plumbing fixtures, and with ‘it pumping powers are low. This velocity is well below the maximum velocity assumed in the reactor portion of sodium-cooled reactors. It should be recognized ;howvever, that this fluld velocity 1s not derived from any basic constants of pature, but it does seem a reasonable essumption on the besis - of present experience. The effective length of the plumbing circuit will depend on the requirements - for maintenance and on the necessary allowances for thermsl strains. These, of course, depend on the type of reactor layout. Here, we will consider pipes coming from the reactor to & pump, the pump feeding into & heat exchanger, and the exit of the heat exchanger going back to the reactor. There will be ' header connections joining a number of fuel tubes together to feed into each punp end heat exchanger. It would seem possible to have a total circuit length of 80 ft for a reasonably sized power reactor, broken down approximately as followss sl In reactor core = - . 10.0 ‘Through blanket ST 5.0 End connections to reaetor . 15.0 Heat exchanger : 0 15.0 “Allowance for pump and | L expansion tank = - Te5 Miscellaneous SR 2.5 Connecting pipes 25.0 o (2 ok If this is the length of the circuit, the fuel salt will traverse the circuit in k.7 seconds and the heat transfer rate will be 4940 Btu/sec-cu ft or 5.2 Mw/cu £t of fuel, or 184 kw/liter of fuel. The remote maintenance demonstration facility in the 9201-3 building in Y-12 provides one check on the length of piping required for & proper layout of a. ~ reactor system. In the facility as it stands,; with its pump, dummy reactor, riping layout with flange connections and dummy heat exchanger, the total - effective length of the fuel galt circult is about k2 £t. If a true breeder resctor were installed (with ite blanket dictating & larger path), and if a full-scele real heat exchanger were installed, the piping length would in- crease by ebout 30 £t to a total of 72 ft. Thie system would then be main- tainable, but its power would be limited by the pump and piping presently installed to about 67 Mw thermal., It would probably be desirable to have ebout 100 Mw (thermal) in each pump-heat exchanger circuit for this type of layout, so that with the increased capacity system, 80 £t is probably a practicel length for the circuit piping in a reactor maintainable by the canyon type equipment installed in this facility. Another type of layout that is currently in favor is that celling for top ~ maintenance, and in which the entire fuel salt circuit is contained in a single large vessel. This type of system allows greater compactness. B. W. Kinyon has analyzed severel cases in a memo (reproduced ag Appendix I), in which credit was taken only for a 200°F AT in the fuel. The results of hie study indicate that with the 2009F AT and a fuel velocity of 15 ft per second in the piping, power densities of sbout 5 Mw/cu £t can be obtained. The total pressure drop in thie system is &bout 115 pei, of which 105 is across the heat exchanger. This enalysislends further credence to the belief that ebout 5.2 Mw/cu. £t or 184 kw/liter can be obtained for a molten- salt breeder reactor. III. Chemical Processing ~ The ohemical processing scheme proposed for the core circult is as follow5° A small eide stream of the fuel salt will be fluorinated to remove Ufls the fluoride volatility process. The UFg will be burned in Hp to UF, and - will be placed in the reserve storage of UF) for reactor feed. The molten~ ~ salt carrier, with most of the fission producte, will be stored for decay of radicactivity to a suitable level, and then processed by the HF dissolu- tion process, recovering the IiF and BeF D9 end eliminating most of the fission products. The IdF-Ber will be adJusted in composition and added to the reactor core stream again. The blanket will be processed by the fluoride volatility process to remove the UFg on & frequent basis to keep the uranium inventory in the blanket low. The UFg is burned to UF),, end the UF), produced is added to n » the reserve supply pending reactor feed or sale. The Pa does not come out of the blanket salt, but returns with it to the reactor system. By keeping the ureanium et a low level in the blanket, the buildup of fission-product poisons i very small and the blanket ie reprocessed completely only a few times dnring reactor life. _ In practice, the freqnency of processing of. the blanket and fuel will be determined by an economic belance. This balence is not struck here, the approach being to see what processing rates are reqnired to achieve certain nuclear aims end to examine the cost of these rates. A 1000-Mw(E) station vill have a heat output of about 2500 Mw(th). At 5.2 Mw/cu ft, the circu- leating fuel volume will be about 480 cu ft of fuel salt. When the fuel comprises about 15% of the volume of the core, the uranium concentration’: in the fuel salt ie sbout 1.2 kg/eu ft, ylelding a total uranium content of the circulating fuel system of 575 kg of uranium (fissionable). About 3 kg will be burned per dzy, so that 191 days is the burnup time for the fuel. TFor the variable fission-product poison to be kept at one percent, & ten percent burnup is allowed before reproceseing, so that the entire 480 cu ft of core must be reprocessed every 19.l1 days of full-power operation. This requires a chemical plant with e capacity of 9200 cu ft/yr, if the same rercentage load fector is assumed for both reactor and chemical plant. A rough calculation indicates that the required blanket or fertile stream volume 1s between 2000 and 3000 cu ft for & system of resctors yielding 1000 Mw(E). . This volume is calculated on the basis of adequate coverage of the reactor cores. However, if it is desired to keep the Pa absorbtion down to 0.005, corresponding to a loss of breeding ratio of 0.01, then there will have to be gbout 310,000 kg of thorium in the blanket system. Since the blanket salt contains about 50 kg of thorium per cubic foot, this re- quires a blanket volume of about 6200 cu ft. Thus the blanket volume can be arbitrarily set to yield the desired Pa 1osses, and for this analysis, 6200 cu ft and 0.01 Pa loss is assumed. The frequency of chemical processing of the blanket is set (aside from eccnomics) by the deslre to keep the uranium 1nventory low and by the desire to keep the fission-product buildup in the blanket small enough so that com- " plete reprocessing of the blanket will not be required frequently. A desir- able goal is.to keep the U and Pa in the blanket down to 30% of the fuel dreuit inventory.. Reprocessing in ebout a 20-day cycle is required to accom- plish this, and there is little benefit to faster processing because the Pa holdup 1is limiting. Processing at this rate will keep the U in the blanket circuit to about one-tenth that in the core eircuit, or about 60 kg, Since uranium has = about fifty times the neutron cross section of thorium, and since there " » «\ 10 are about 300,000 kg of Th in the blanket, there will be ebout one percent as many U absorptions as thorium absorptions in the blanket. Thus after ebout %en burnups of the core, the fission product level in the blanket will glve sbout & one percent poison there. At an 80% load factor, this would be after ebout 6.5 yr. Thus, in a 20-yr life, the fission-product poison level in. the blanket might rise to 3% poison, and complete reprocesesing of the blanket salt should be considered at that time. The estimate of costs for chemical processing of the fuel and blenket at the above rates (9200 cu ft per yr for the fuel and 113,000 cu ft per yr ‘for the blenket salt) is based on e report by Weinrich and Associates to ORNL on "Process Design and Estimated Coste of Chemical Plants for Procese- ing Molten Salt Fuels”. The larger plant estimated by them had & capacity of 10,000 cu £t per yr of fuel salt, which is about the size required here for the fuel salt circult. For the fuel processing plent, Weinrich and Associates estimate & cost of $3,455,000, plus about §1,500,000 of shared fecilities with the reactor plant. Crude adjustments to these figures made by Osk Ridge personnel revised them upward to & total of about g9,830,000. A much cruder estimate has been made of the additional plant cost to provide for the rapid fluorination of the blanket salt. This was made by assuming that multiplying the cost of the portion of the plent involved in fluorina- tdon of the core salt and UF), recovery by five would give & plant of eleven times the capacity. On thig assumption, the complete chemical plant for treatment of both core and blanket salts would cost sbout $18,000,000 for the 1000-Mw(g) plent. At & 29% annual charge and en 80% load factor, the cost of the chemical plant, together with its operation, would be about 0.75 mill/kwhr. - The total inventory of uranium and protectinfum in the reactor system is estimated as follows: In resctor fuel 575 kg In blanket 180 kg In chemical processing 30 kg In storage | , 60 kg - Total Bh5'kg The uranium inventory at §15/g is $12,680,000, or $12.70/kw. At 4% this is 0.07 mill/kvhr, or at 124, 1t is 0.22 mill/kwhr. The blanket and core salts, including thorium inventory, will cost sbout $25,000,000, or £25/kw.. At 144 per year and 80% load factor, this amounts to 0.5 mill/kwhr. With a net breeding ratio of 1.06, there would be 52.5 kg of fissionable uranfum ”» ;] &) 11 produced per year, which would yleld ebout £790,000/yr or ebout 0.11 mill/kwhr. Thus the total fuel cycle cost would be about.l.a.mdlls/kwhr on the basgis of present uranium use charges. It is obvious that considerable savings in fuel cycle cost can be made by sacrificing doubling time. It is probable that if a breeding ratio of 1.00 wvere satisfactory, at least half of the salt and thorium inventory charge could be avoided and the chemical plant chargee would be considerably reduced, probably by at least one-third. Thue, & hold-own breeder might have power costs as much es 0.5 mill/kvhr less than the doubling reactor. In this enalysie of chemical processing, only processes on which there is a faeir amount of laboratory data have been considered. With the fluid blanket, an easy means of removing Pa 18 being sought. If it is found, then the blanket holdup will be reduced, and the thorium inventory cen be reduced appreciebly. IV. Performance as & Breeder A number of grephite-moderated molten-salt resctor configurations have been subjected to multigroup nuclear calculations with the Cornpone and Sorghum codes.* The eriticality calculations cen be correlated quite well if one plote the concentration of uranium in the core against the carbon absorptions in the core. This is done in Fig. 4 (ORNL-IR-Dwg. 42240)., The plot shown comes from reactors of equivalent spherical core diemeters of 3 ft, L ft, 5 £t and 14 ft. The fuels in the core have thorium concentration of 1 mole 4 ThF),. % mole % ThF), 7 mole % ThFy, and 13 mole % ThF,, and the volume frac- tion’of fuel represented in the various calculations of 10%, 12.5%, 15%, 18.3% end 20%. It includes calculations of both the initial state of the reactor with pure U-233 and the state achieved after 20 yr of operation with e near equilibrium mixture of U-255 and U-235. Although the relationship is not mathematical, there is a good empirical fit and the curve can be used with fair confidence in predicting the uranium toncentration required in the fuel salt. For the reference system with & 15 vol % fuel fraction in the core and 0.04 neutron absorption in carbon, the concentration of fissionable uranium required in the fuel salt ie 1.2 kg per cu ft. | This figure, combined with the mumber of 5.2 Mv per- cu £t developed in Section II sbove, ylelds a specific power in the fuel stream of L4.33 Mw/kg. ¥ 'MSR Quar. Prog. fipts._, om 268k, ORNL 2723, ORNL-2799- 3 » @ 12 The chemical processing and blanket holdupe of fuel lead to a total fission- able uranium inventory of 845 kg, as described in Section III, so that the over-all specific power is 2.96 Mw(th)/kg. The effective value of eta for the fuel mixture will depend on the thermality of the spectrum, which is related slso to the carbon gbsorptions per sbsorption in fissionable urenium. At 0.0k ebsorption in carbon, the value of eta for the isotope mixture is sbout 2.22, based on & thermal velue of 2.28 for U-233. From the nucleer celeulations cited above, one can correlate the neutron absorp- tions in the carrier salt in the core with the absorptions in carbon. This is done in Fig. 5 (ORNL-IR-Dwg. 42239). The volume fraction of carrier salt is 20%, compriced of 15% for the fuel and 5% for the fertile stream, so that the absorp- tions in the carrier salt in the core are 0.04 for carbon sbsorption of 0.0k, As described in the preceding section, the reference chemical processing plant provides for keeping the variable fission-product polson fraction down to 0.01, and the Pa losses (2 x absorptions) down to 0.0l. Uranium-236 will, of course, build up from radietive captures in U-235. With a breeding ratio of 1.06, the removal of U-236 by the salé of excess fuel will approximately equal the removal by neutron absorptions, go that the U-236 poison will be approximately 0.0l. The ~ neutron losses to saturable non-volatile Pission products will be about 0.006, and 1f Xe-135 losses can be kept to 0.00k, the total saturable fission-product losses will be limited to 0.0l. To keep the Xe-135 losses to 0.004 requires its removal on a time eycle of about 6 minutes. The off-gas system can be designed to accom- plish this by bypassing 2% of the pump flow through & degasser. This was the degassing bypass flow in the ART pump. Other neutron losses are estimated to be 0.03 in the blanket salt, 0.003 for - delayed neutrons, and 0.002 for leakage. Considering the efficiency of the fluoride volatility process, 0.005 may be adequate for chemicel processing losses. Fission-product pickup by the graphite, assuming that the fuel tubes soak up-<:0 5% by volume of fuel, givee a negligible loss unless there is pre- ferential fission-product absorptionu The total neutron losses now add up to about 0.16, which subtracted from an ete of 2.22 yields & net breeding ratio of 1.06. It should be noted that higher breeding ratios can be obtained by decreasing the volume fraction of fuel in the core and by increasing the uranium-to-carbon ratio in the core. However, these both lead to higher uranium inventories and consequently no great improvement, if any, in doubling time. Furthermore, if thermal eta for U-233 i 2.29, as is belleved in Osk Ridge, instead of the 2.28 assumed, the breeding ratio is improved by nearly one point. With an over-all breeding ratio of 1,06, ‘the doubling time is gbout 13 yr of full-power operation. "V,_;FEasibiIity and Cost of Molten-Salt Reactors The basic feasibility of molten-sslt reactors has been discussed in a section of the book "Fluid Fuel Reactors®., This and later information have been » ) 15 reviewed by the Fluid Fuel Reactors Task Force, and it was the concensus of the group that, with minor exceptions, the feasibility of the molten-salt reactor was established as far as materiels compatibility and handling ie concerned. These exceptions concern the pickup of fuel salt by the graphite and the possible .precipitation of UO> by gasee adsorbed on the graphite. Since the time the Fluld Fuels Tesk Force met, the results of a one-year circulating sslt loop containing grephite and of graphite impregnation studies bave shown that the graphite is steble in contact with the salt, and that there are varieties of graphite that will soak up less than 0.2% by volume of fuel salt, It has also been found possi- ble to prevent UO2 precipitation by pretreatment of the graphite. The Fluid Fuel Reactors Task Force further expressed doubt as to the economic maintenance of fluid fuel reactors in general. For the molten-salt reactor, this can be answered finaslly only in & reactor experiment, which has been pro- posed. In the meantime, good progress has been made in devising suitable main- tenance procedures for one type of reactor construction. Since the time of the tasgk force, the design effort on molten-salt reactors has been directed toward breeder resctors that take advantage of the compactness that is possible as a result of the high temperature and good heat transfer pro- perties of molten salts. Most of the designs developed have a compact primary system, such as that described in Appendix I and illustrated in Fig. 6. In these designs, the entire primary fuel circuit is contained inside a reactor vessel, With this construction, a parallel comparison with solid-fuel-element reectors is evident, in which the tubes of the primary heat exchangers of the salt reactor are compared to the fuel tubes in the core of a solid-fuel-element reactor; both contain fuel, both constitute the primary heat exchanger surface, and in each case they are contained within the primary reactor vessel enclosure. In a gimilar way, the pumps for circulating the fuel are compared to the control rod mechanisms (and fuel shuffling mechaniem for the fast reactor). Both involve - moving parts inside the reactor enclosure, and the pump, though bulky, is cer- tainly simpler. Maintenance of pump and heat exchanger in the salt system is by overhead withdrawl and replacement, and the operations required are comparable to those required for the replacement of core sssemblies and repair of control mechanisms in the reactors with solid fuel elements, particularly those cooled with sodium: There is thus no reason to expect maintenance costs for the MSR to be higher than, say, for the fast reactor. As for capital costs, the higher temperature of the heat source and the very - high heat capacity per unit volume of the salt (approximately k.4 times that of sodium) lead to compactness of the entire system. The following table com- pares pertinent factors of complexity and heat transfer with four reactors using & sodium coolant. Comparing the MSR primary heat exchanger with the reactor cores, it is simpler by virtue of having fewer tubes, and has about the same efficiency as the fast reactore in terms of surface area. The avoidance of an intermediate heat exchafiger for fhe'MBR, possible because there is no violent water reaction and because the induced radiocactivity is very short lived; 1s a further factor reducing capital cost. In the steam » " 5 14 - generator portion, the higher temperature of the salt coolant gives an - edvantage by about a factor of two. The lest two rowe of the table indicate how the high heat capacity of the salt, even using a conegervative bulk AT, can reduce the pump capacity end eystem piping requirements by at least a factor of two. On the basis of this analysis, even after allowing for the high cost of INOR-8 and of the salt coolant, the ecapital costs of a molten-salt reactor should be less than for the sodium-cooled reactors. Sodiuwmn-Cooled Reactors ' Advanced MSR Hellam P/604* Fermi Fast Reactor Net electrical Mv assumed for reactor 333 80 205 ok 283 Fuel tubes per MwE 58 28 1%0 264 Primary heat exchsnger - tubes per MwE 19 Primary heat transfer 22 111 38 14.6 25 surface per MVE (fuel tubes in case of sodium- cooled reactors) sq ft per MwE Intermediate heat exchanger sq £t per MwE - | 935 170 160 g2 Steam generator, super- 120 237 -:214 345 200 " heater and reheater .surface gq ft per MWE - Coolant flow data (avg) L | o Bulk AT assumed . -~ 150°F 338°F 275°F 250°F 350°F “gal/min flow per MwE 106 2k 263 318 214 * 8. Ievy et al, "Advanced Design of a Sodium-Cooled Thermal Reactor for Power Generation", 1958 Geneva Conference Paper P/60L. o b » 15 The moet careful cost estimates of molten-salt reactor construction have been made by G. D. Whitman and ere included in: (1) ORKRL 2634; (2) ORNL- CF-59-1-26; and (3) ORNL 2796. The three cases include two power reactors of 640- and 860-Mw (thermal) caepacity, and an experimental reactor of 30-Mw capacity. A reasonable extrapolation of these costs to the 2500-Mv (thermal) etation required for 1000 Mw(E) ylelds capital costs of from $170 to $200 per kw. This estimate 1s for & first plant, but does not include development coets. When these capitel coste are combined with the fuel eycle costes esti- mated in Section III of 1.2 Mw/kvhr and & reasonable operation and mainten- ance estimate of 1 mill/kwhr, one gets & power cost in the neighborhood of 6 mills/kxwhr for the first such large breeder reasctor plant. Presumably one could expect lower coste than thies as a result of prior prototype reactor construction and operation. It is difficult to attempt to prediect ultimate costs, however, until experience ‘has been had with at least an experimental reactor. . » 16 APPENDIX I. INTRA-LABORATORY CORRESPONDENCE Oak Ridge National Iaboratory October 22, 1959 To: H. G. MacPherson ce: L. G. Alexander J. W, Miller File (BWK) Subject: Volumes and Pressure Drops for Molten Salt Breeder Reactors The following table is a comparison of two reactor sizes, each with two flow velocities in the external piping. The heet exchanger has been designed on the basis of 0.300 in. inside diasmeter x 0.035 in. wall tubing in a 45° helix, with edjacent coils wound opposite hand. Fuel temperatures are taken as 1275 and 1075°F; coolant temperatures as 1150 and 10000F. The use of 1/4 in. ID x 5/16 in. OD tubing would decrease the heat exchanger length by 21%, increase the number of tubes by 50%, and inereasge the. diameter by about 15%. The fuel volume external to the core would be reduced by about 10%, which might overweigh the problems introduced by the other changes. The fuel volume might be reduced by considering the entire flow in the center of the heat exchanger as "pump suction" and using & higher flow velocity. This would be about 10% of the fuel outside the core for the higher flow rate cases. : The attached sketch (Fig. 6) is approxihately to scale for the smaller reactor with 20 ft per second fuel -veloéity in thé piping. /e/ B. W. Kinyon :nh S Enclosure Qi 17 VOLUMES AND PRESSURE DROPS FOR MOLTEN SALT BREEDER REACTORS Reactor Power, MWwE (net) Station Efficiency,% Reactor Power, MwT Blanket Power, % Core Power, MwT Fuel Tempersture Exit, °F o Fuel Tempersature Entrance, F AT in Fuel, °F 5 Volume Flow of Fuel, ft”/sec Flow Velocity,in Core, ft/sec Flow Area, ft Volume Fraction in Core Core Cross Section, ££2 Core Diameter, ft Diameter of Equ%valent-Sphere, £t Core Volume, ft Fuel Volume in Core, £t Blanket Thickness, ft Blanket Volume, £t2 Cu Ft of Blanket per MwT MwT per Cu Ft of Blanket Fuel Velocity in "piping", ft/sec Flow Area, ft° . Heat Exchanger Area, ft Bundle Area, ft° - Coolant Flow Area, ft Total Aresa Bundle Height, ft Bundle ID, ft Bundle OD, ft Cartridge OD, £t 335 40.85 81 750 1275 1075 200 55.2 20 2.765 0.25 22,08 5430 6.07 117 29.25 2.5 T4l 0.910 1.10 o \.)II\) * 8 I =3 co-J = o * * > \ no 19.27 8.75 1.875 L,56 h.gs . O\ o 33T B 500 40.85 1223 8 1125 1275 1075 200 82.8 20 4.140 0.25 33.12 6.50 T.45 215 53.75 2.5 91 0,750 1.333 15 #flBm e o e o SR 3 gffigf Ha L= O\ D O vl O 3 OO It ¢ P C (continued) Fuel Volumes, ft5 per reactor "Piping Volume 57.0 k2.9 85.5 Pump 3.6 3.6 5¢3 Shell Annulus 1’-01 5.9 h-o9 ~ Expansion and Off-Gas (8.8%) 13.6 12.2 21.2 Upper Plenum 7.9 7.8 11.8 . Heat Exchanger (0.3 in. ID, 105 psi) 32.0 32.0 48.0 Iower Plenum T.2 Tel 10.7 - External to Core 195.5% 109.5 187.4 Core 29.3 29.3% 5%.8 Total 1554.7 138.8 241.2 Ratio of external fuel to core fuel 4.28 BT 3.48 Retio of total fuel to core fuel 5.28 L.7Y L.48 Cu Ft of Fuel per MwT External to Core 0.157 0.137 0.156 Core 0.036 0.0%6 0.0k Total 0.193 0.173 0.200 MwT per Cu Ft of Fuel _ External to Core 6.37 T30 6.41 Core 27.8 27.8 22.8 Whole System 5.18 5.78 5.00 Pressure Drop Velocity head losses in: Core to pump 0.11 0.11 0.08 Heat Exchanger to core 0.06 0.07 0.05 - Plenums , 1 - 0.50 0.50 0,50 Total . - _ . . 0067 B 0_06.8 - 0065 AP/velocity head =~ ~ 3.00 5.33 3.00 Piping AP; psi - B ' 2.0 3.6 1.9 Velocity head losses: for: 2 core passes | 0.62 0.62 0.78 ~Inlet and Exit 0.25 0,25 0.25 1 Bottom Turns 0.50 0.50 0.50 - Total - | | 1.37 | 1.37 - 1.55 AP/velocity head, psi 5033 5,33 5¢33 Core AP, psi Te3 T3 8.2 N & < < - 9 ° b = qogfimpm O & OOWM = o \N n B N4 ° \Oj £\ o0 = O olo O o 18 ot 3 ° o \O O?O 3 & - . WM\ s » A W\ |C>C>c> s & o A 1O —3 wilow @ oV H o i L\ b A {continued) Total Pressure Drop Outside Hea;fgs;-@xchanger, psi 9.3 Presspre Drop Acrose Heat Exchanger, pel 105.0 Total AP in System, psi - 114.3 S 19 10.1 11.7 105.0 105.0 115.1 116.7 ORNL-LR-DWG. 42242 UNCLASSIFIED ~ BLANKET n w m o P w 9 < L] T O = % z < RO g o 7///////////// A \l""'fli!!iiifl*ii!‘fli!i e e By g g g e eyl et st ll"l'l'llllllllll!l.vllllllllll"lllll-llrlllll o V_ | \\\\\\\\V\\\\\\\\\\\\\\\\\k\Lfl \\\\\\\ \\\\\\\\ LLLLLLLLL L \»v ,\\\\L\ L \\\\\ LLLLLL \\\\\ 2L \\\\& -SLIP JOINT, GRAPHITE TO METAL 20 FUEL INLET +FUEL EXIT Figure 1 GRAPHITE TO METAL N i g ey ey s SEAL OR FLANGE 21 UNCLASSIFIED ORNL-LR-DWG 37258 TITIIT T IT I T I T I T T T T T T T T T I T T I TS T T . P LR el s L Pl el e 22 L2l BLALKET ’IIIIIIIIII ’ B E ] \ | E H L] i E N A A N \K §r1 N N N N FUEL DRAIN fll/l//’///[///f//////}//////////I/A A \\\\\ X b h) fl D] . ””/”.””’””/A ’”I/”””/”"’I”’f"’/f2 AN ’.’f/’”’.”””l—"’” 3 . b O Q Vi s ) \ B2 //t/////////////////////////////// b ! ] GRAPHITE // g f ] K N \ § § \ N \ N ) N N\ N N\ ) N N N ) N N 1 N 5 U A Y % I L 1 INOR - 8 Fig. 2 ) Heterogeneous, Graphite-Moderated, Two-Region, Molten-Salt, Thorium Breeder. 22 ORNL-LR-DWG 44463 UNCLASSIFIED w—pe-TO FUEL EXIT BLANKET | ——— SALT BN GRAPHITE REFLECTOR (COOLED BY | ANKET BLSALE) GRAPHITE MODERATOR ,442 GRAPHITE _——~7 TUBES LA ,”/ ' et lleLankeT | - | | o o . BRAZED JOINTS, ’:éfSALT -L— —J - =GRAPHITE TO INOR-8 # : 5 =n§ S T———FROM FUEL ENTRANCE Fig. 3. Internally Cooled Reactor kg.FISSIONABLE URANIUM PER ft° OF CORE e3 UNCLASSIFIED -. ORNL-LR-DWG, 42240 0.70 0.50 0.30 0.20 0‘.'10. 001 | 002 003 0.05 0.07 0.0 | NEUTRON ABSORBTIONS IN CARBON PER | | ABSORBTION IN U- 233 + U-235 FUEL CONCENTRATION IN CORE VS. CARBON ABSORBTIONS Figure 4 ABSORBTIONS IN SALT CARRIER 24 UNCLASSIFIED ORNL-LR-DWG. 42239 35 V% 0.10 ] 25 V% ¢ | | .20 V% /| { ‘ 8.3 V% 711 15 V% | . v | | 0.03 . : / | ) v - l ¢ v / - 2 10V% A . & 0.02 2.5 V% 0.07 0.05 0.01L— — —1 11 _ 0.010 002 003 0.05 0.07 0.0 ABSORBTIONS IN CARBON Figure 5 25 CRANL-AR- DWGE 42363 PUMP HEAT EXCHANGER SALT )/4VESSEL | -BLANKET ‘ “.“" : —— | | “ w;-l . \ \ \\\\\\\\\\\\\\\\\\\.\\\\.\\\\.\\\.\\.\\\\\\ m /‘ &\\\\\\\\\\\\\\\\\\\\\\\\\\\\\Q\\\\\\\\\\\\\\\\\.\\.‘\ ] e \\\\;a‘n“naaoaaanmnnaammuuunaaaan“aaaaasnfim > ; : “ / ) e \ \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ ] - Mw~\\;eaassaauaa““aaaaanaaaanaaassvcecaaaaoanamm :--...ssa«ga““»““eooeauouoaueaassxsuueenaansv“n“m y , -~ 2 Z “ S = 7 - Ll < Y - ,um___.- LIS SIS LTSS AT IS 117777777777 (] bz : T o ] ' e T — TR / MOLTEN SALT BREEDER REACTOR Figure 6 ~ Y . pistributi on 1- 15. 16. 17. 18, 19. 20, 21. 22, 23, 2k, 25, 26. 27. 28. 29, 30. 31, 32, 53 3L, 35 > / 4 37. 38. 39. k0. hlfl k2, 43, bk, 45, 46. Ho L. C, A. E. Fo A. Wo E. R. We De We Ve Go Re. R Ro D. Wo Ao Wo W. Je He Wo P. Go Vo B. M, Je G. Go Je Lo Se F. L. F. Jo Bo E. 0. H. L. I. E. A. Re Ao K. P, R. Re P. W. H. R. We E. Vo E. Ao * MacPherson Alexander Barton Benson, AEC-ORO Bettise Blankenship Boch Boudreau Breeding Briggs Browning Cempbell Carr Carter Cathers Chapman Charple Coveyou Dougles Ergen Fraae Gall Grimes Harmond Hoffman Jordan Kasten Keilholtz Kinney Kinyon Iackey Iane 47, L8, 9. 50. 51. 52. 53 54, 554 56 3T 58. 59. 60. 61, 62. 63, 6k, 65. 67. 68.. 69. T0. T1. T2 73 71"'0 75 76-17- 78-92. R. N. Lyon W. D. Ma.nly E. Rs Mann L. A. Mann W. B. McDonald He J. Metz R. P. Milford A+ Jo Miller Je Wo Miller G J. Nessle C. W. Restor W. R. Osborn As M. Perry R. M. Plerce J. T. Roberts H. W. Savage F. P, Self Ms J. Skinner J. A. Swartout A. Taboada Re E. Thoma D. B. Trauger Fo Co Vonderlfl-ge G. M. Watson A. M. Weinberg Je. Ho Westsik G. D. Whitman J. Zasler Isboratory Records, R.Ce. Iaboratory Records TISE-AEC