REACTOR PHYSICS AND FU EL-CYCLE ANALYSES

Transcription

1 REACTOR PHYSICS AND FU EL-CYCLE ANALYSES A. M. PERRY and H.F. BAUMAN Oak Ridge National Laboratory Oak Ridge, Tennessee 3783 KEYWORDS: molten-salt reactors, fels, economics, operation, breeding, thorim-232, ranim-233, performance, MSBR, fel cycle, cost, breeding ratio Received Agst 4, 1969 Revised October 9, 1969 As [Jresently canceived at Oak Ridge National Laboratory and described in this isse, the singleflid Molten-Salt Breeder Reactor, operating an the 232 Th- 233 U fel cycle and based on a reference design, has a breeding ratio of 1.6, specific fissile inventory of 1.5 kg/mw(e), a fel dobling time of 2 years, and fel cycle costs of o.7 mill/kwh(e). Start-p may be accomplished with either enriched ranim or pl:tanim, with little effect an fel cost; the breeding ratio, averaged over reactor life, is redced.1 to.2 relative to the eqilibrim cycle. operated as a canverter, with limited chemical processing, the reactor may have a conversian ratio in the range.8 to.9 with fel cycle costs of.7 to.9 mill/kwh(e). INTRODUCTION One of the most important aspects of the Molten-Salt Reactor (MSR) 'Concept is that it is well sited for breeding with low fel-cycle costs, and it does so in a thermal reactor operating on the 232Th-233 fel cycle. This is tre not primarily becase of any niqe nclear characteristics, for the reactor is similar to other thermal reactors in terms of attainable fel-moderator ratios, the navoidable presence of certain parasitic netron absorbers, and reliance on a fertile blanket to redce netron losses by leakage to an acceptably low level for breeding. Indeed, the concept might be thoght to have some a priori disadvantage, becase a sbstantial fraction of the fissile material is invested in the heat transfer circit and elsewhere otside the reactor core. The pecliar sitability of the molten-salt reactor for economical thermal breeding stems rather from the practical possibility of continos removal of fission-prodct wastes and 233Pa, and virtally arbitrary additions of ranim or thorim, withot otherwise distrbing the fel. This fndamental aspect of the molten-salt reactor, details of which are discssed in other papers of this series, has a profond effect on the relationship between ne - tron economy and fel-cycle cost. The coincidence of good netron economy with low fel-cycle cost which characteries the molten-salt reactor appears to be niqe among thermal reactors and will be described more flly in this paper. GENERAL NUCLEAR CHARACTERISTICS The LiF /BeF2 carrier salt sed in the MSR concept is not by itself a very good moderator. Its moderating power is abot half to two-thirds that of graphite (the exact vale depending on the proportions of Li and Be in the salt), while its macroscopic absorption cross section is an order of magnitde greater than that of graphite, even with the feed lithim enriched to % in the 7 Li isotope. (With this composition, <1% of the netron absorptions in the salt occr in 6 Li; nearly half are in florine, and abot a third in 7 Li.) It is evident, therefore, that an additional moderator is needed, and graphite is selected for this prpose becase of its compatibility with the salt. There is only a weak connection between the fissile fel concentration in the carrier salt and the heat transfer characteristics of the salt (arising primarily from the inflence of the thorim concentration on the physical properties of the salt), and as a conseqence one has considerable latitde in selecting the ranim (and thorim) concentrations in the salt. Becase the carrier salt itself constittes a significant netron poison, the fel concentration in the salt mst not be set 28 NCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 197

2 at too low a level, bt mst be high enogh for the fel to compete favorably (for netrons) with the lithim and the florine in the salt. On the other hand, it mst not be too high, lest the inventory of fel otside the reactor core become excessive. The optimm fel concentration, typically.2 mole% of UF4 in the salt, or ""1 kg of ranim per cbic foot of salt, is interrelated with the netron spectrm in the reactor, which is a fnction of the relative proportions of fel salt and graphite moderator in the core. Too large a proportion of salt leads to an excessive fel inventory and to a poorly thermalied netron spectrm, with a redced netron yield, 1J; too large a proportion of graphite leads to excessive netron-absorption losses in the graphite. An optimm salt volme fraction is typically fond to be ""13 to 15%. The proper balance of the above factors does, of corse, depend in part on the power density in the reactor core, which may be selected almost independently of the power density in the remaining parts of the primary salt circit. The maximm power density in the core is limited by fast netron damage to the graphite moderator, while the removal power density in the external power recovery circit is limited primarily by heat transfer and pressre-drop considerations and by reqirements for pipe flexibility in the piping rns between the reactor vessel and the heat exchangers. The necessity for maintaining a sfficiently high fel concentration to sppress netron losses in the carrier salt and in the moderator, together with the reqirement for appreciable core sie simply to generate the reqisite amont of power, leads to the conclsion that thorim mst be present in the core, not merely in a srronding blanket. However, the qestion of how the thorim is to be incorporated in the core is crcial to the MSBR concept. One qickly recognies several distinct possibilities, some mch more desirable in principle than others, bt fll of implications with respect to reactor design and chemical processing. We have previosly given serios consideration to a two-flid reactor in which the fissile and fertile materials are carried in separate salt streams, the bred ranim being continosly stripped from the fertile stream by the floride volatility process. Blanket regions contain only the fertile salt, while the core contains both fissile and fertile streams; these streams mst be kept separate by a material with a low-netron cross section, that is, by the graphite moderator itself. This approach appears to yield the best nclear performance, owing primarily to a combination of maximm blanket effectiveness and minimm fel inventory. It also exhibits attractive safety characteristics becase expansion of the fel salt, pon heating, removes fissile material from the core while leaving the thorim concentration nchanged. The concept does, however, involve important qestions regarding the reliability of the graphite "plmbing" in the core, the adeqate proof of which may reqire a good deal of time and testing. The present approach employs a single salt stream which contains both the fissile and the fertile materials. This concept represents a modest extrapolation of the technology already demonstrated in the MSRE. A central featre of the concept is the manner in which the single salt composition can be made to fnction adeqately both in the core and in the blanket (or oter core) regions. This is done by the simple expedient of altering the salt volme fraction, making it considerably larger in the blanket than in the core. This ndermoderation reslts in enhanced resonance captre of netrons by thorim in the oter region, gives rise to a negative material bckling in the oter region, and shold in principle case a fairly rapid decrease in power density in the blanket as a fnction of distance from the core bondary. In practice, the distinction between the core and blanket regions is not as clear ct as this argment may sggest, bt the idea works reasonably well. Figre 1 illstrates the power density distribtion for or present reference design based on the single-flid concept. The enhancement of resonance netron captre in the blanket (or oter core) region is indicated by the ratio of netron absorptions in 232 Th to those in 23 3; this ratio is abot 1. in the core, and 1.3 in the blanket. The salt annls, which is reqired to allow the periodic replacement of the moderator, fnctions as a part of the oter core region. The principal shortcoming of the single-flid concept, of corse, is the sbstantial investment of fissile material in the blanket region. This reslts in a rather different compromise between breeding gain and specific inventory than in the two-flid concept, leading both to redced effectiveness of the blanket region and to an appreciable increase in fel inventory. Fortnately, this featre of the single-flid reactor is partly offset by a redction in netron captres in the carrier salt, owing to the fact that a single carrier salt contains both fissile and fertile materials. The preceding qalitative discssion is intended to provide a general nderstanding of the interplay of factors affecting the selection of MSBR design parameters. These factors are of corse qite nmeros. They inclde core sie, radial and axial blanket thickness, reflector thickness, salt volme fractions in the core and blanket regions, thorim and ranim concentrations, NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY

3 ... 6 GI... E ' 5 "' c c >< ::::> 4 LL... E 3 >- 2 VI : LI.I c a: LI.I 3 1 c.. Fig RADIAL DISTANCE FROM CORE CENTERLINE (cm) a: 1- _, LL. a: 3 Radial power density and fast flx distribtions-single-flid MSBR. chemical processing rates, and reactor power level. Becase the interaction of all these factors is rather complex, and becase of the need to identify optimm vales of the design variables rather closely, we have fond it convenient to make se of a comprehensive, atomatic reactor optimiation procedre for arriving at that combination of design parameters that will prodce, in some sense, the best attainable performance. The Reactor Optimiation and Design code (ROD) is based on a gradient projection method for locating the extreme vale of a specified figre of merit, which may be any desired fnction of the breeding ratio, the specific fel inventory, varios elements of the fel cycle and capital costs, or any other factors important to the designer. The comptational procedre con;iprises mltigrop (synthetic), two-dimensional diffsion-theory calclations of the netron flx, an eqilibrim felcycle calclation which determines the critical fel concentration and nclide composition consistent with processing rates and other variables, and the gradient projection calclation for moving the clster of independent variables in the direction that most rapidly improves the figre of merit. The optimiation may be constrained by limiting the allowed range of the independent variables, or by selecting in advance the desired vale (or a limiting vale) of certain derived qantities, sch as the maximm power density. The figre of merit sed here in determining reactor design specifications is related to the capability of a reactor type to conserve fel spply in an expanding nclear economy. For the special case of a linear increase in power generation, the total amont of natral ranim that mst be mined p to the point when the system becomes self-sfficient (i.e., independent of any external spply of fissionable material) is proportional to the prodct of the dobling time and the specific fel inventory. We have chosen to optimie or MSBR design primarily on the basis of a qantity which we call the fel "conservation coefficient," defined as the breeding gain times the sqare of the specific power, which is eqivalent to the inverse of the prodct of the dobling time and the fel specific inventory. Therefore, a maximm vale of the conservation coefficient is soght in the optimi.ation procedre. EQUILIBRIUM FUEL-CYCLE RESULTS The reslt of a reactor optimiation calclation is a set of specifications for the optimm reactor configration, sbject to any imposed constraints, together with a complete description of its eqilibrim fel cycle. This description incldes the mltigrop netron flx distribtions, the reslting power distribtion, and the consistent set of concentrations of all nclides present in the reactor. We have imposed constraints on maximm power density (i.e., minimm graphite life), on overall reactor vessel dimensions, and on chemical processing rates which we believe will reslt in near-minimm power cost. Althogh we lack specific information as to the cost of chemical processing as a fnction of fel processing rate for the liqid-metal extraction process, it appears that processing eqipment sies and operating costs will be comparable with those for the floride-volatility /ranim-distillation p r o c e s s considered for the two-flid reactor. We have therefore fixed the processing rates, listed in Table I, at vales fond to be essentially optimm in stdies of the two-flid reactor, with minor adjstments appropriate to the extraction process. While sbseqent improvements in processing cost estimates may sggest some change in optimm processing rate and some change in fel cost estimates, we do not expect that these will reslt in any major revision in performance estimates for the reactor. The reference reactor configration wmch reslts from these and other (engineering) considerations is described by Bettis. 1 A smmary of its nclear design characteristics is given in Table I. 21 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 197

4 TABLE I Characteristics of the One-Flid MSBR Reference Design A. Description B. Performance Identification Power, MW(e) MW(th) Plant factor Dimensions, ft Core one 1 Height Diameter Region thicknesses Axial: Core one 2 Plenm Reflector Radial: Core one 2 Annls Reflector Salt fractions Core one 1 Core one 2 Plena Annls Reflector CC Conservation coefficient, [MW(th)/kg]2 Breeding ratio Yield, % per annm Inventory, fissile, kg Specific power, MW(th)/kg Dobling time, system, year Peak damage flx, E >5 kev, n/(cm2 sec) Core one 1 Reflector Vessel Power density, W / cm3 Average Peak Ratio Fission power fractions by one Core one 1 Core one 2 Annls and plena Reflector xl14 4.2xl13 3.7x Salt composition, mole% UF4 ThF4 BeF2 LiF Processing cycle times for removal of poisons a 1. Kr and Xe; sec 2. Se, Nb, Mo, Tc, R, Rh, Pd, Ag, Sb, Te, Zr; sec 3. Pa; Cd, In, Sn; days 4. Y, La, Ce, Pr, Nd, Pm, Sm, E, Gd; days 5. Sr, Rb, Cs, Ba; year 6. Br, I; days mance of the MSBR. By far the most important effect is the ncertainty in the average vale of 1J of 233 in the MSBR spectrm, which leads to an ncertainty in the breeding ratio of ±.12. Uncertainties in the cross sections of other important MSR nclides (sch as F) make a relatively small contribtion to the overall ncertainty in the breeding ratio, which is estimated to be ±.16. A detailed discssion of cross-section ncertainties is given in a report by Perry. 2 Eqilibrim Fel-Cycle Costs As stated before, the molten-salt breeder reactor exhibits nsally low fel-cycle costs in comb1nat1on with good breeding performance. This a According to or present flow sheet, Zr' Cd, In, and Sn will be removed on a 2-day cycle, and Br and I on a 5-day cycle. The additional poisoning, however, is negligible. A netron balance for this case is given in Table II, in which the normaliation is to one netron absorbed in 233 pls 235U. Uncertainties in Netron Cross Sections We have estimated the effect of ncertainties in netron cross sections on the calclated perfor- reslts primarily from the low specific fel in- ventory and from a small bt non-negligible ex- cess prodction of fel, which reslts from the ability to process the fel rapidly at what appears to be a very low nit cost. The inventory of fissile material in the reactor and chemical processing plant amonts to some 148 kg, inclding ""'1 kg each of 235 U and 233 Pa; when valed at $13./g for 23 3 and 233 Pa and $11.2/g for 235 U, this material is worth $19 million. With an effective annal inventory charge rate of 1%/year and a.8 plant factor, the fel inventory ths contribtes.27 mill/kwh(e) to the NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY

5 TABLE II Netron Balance, Single-Flid MSBR Absorptions Fissions 2ss Th Pa.17 2ss Np.61 6Li.49 7Li.159 9Be.71 (.46) 8 l9f.25 Graphite.519 Fission prodcts.196 Leakage TIE a (n,2n) reaction. fel-cycle cost. The fel salt, with a composition LiF/BeF2/ThF4 = 72/16/12 mole%, respectively, is estimated to be worth $3 million, inclding the thorim. At 1%/year, this contribtes.4 mill/ kwh( e) to the fel cycle cost. For a conversion ratio of 1.62, fel prodction reslts in a redction of.9 mill/kwh{e) in the fel cycle cost. The cost of thorim brnp, in contrast, is negligible [....2 mill/kwh{e)]. The chemical process for removal of fission prodcts, which is nder development for se with the single-flid MSBR, involves the accmlation of rare-earth fission prodcts in a portion of the salt stream in the liqid bismth extraction tower. The concentration of rare-earth triflorides in this salt is limited by solbility to mole%; it is presently planned to limit this concentration by discarding....5 ft3/day of carrier salt having,...1 times as high a concentration of rare earths as the salt circlating in the reactor. The makep of carrier salt (inclding ThF.J reqired to compensate for this discard ths contribtes....5 x $1846/ft3 = $932/day to the fel cost, i.e.,.5 mill/kwh(e) at.8 plant factor. The cost of processing the fel for removal of fission prodcts and for isolation of 233Pa from the circlating salt stream is difficlt to assess precisely. Or tentative estimate of these costs, inclding both capital and operating expense, is....3 mill/kwh{e), based on the rapid processing rates indicated in Table I. 3 In smmary, therefore, we estimate that the eqilibrim fel-cycle cost will be... o. 7 mill/kwh(e), as shown in Table m, for a single-flid MSBR of the reference design. EFFECT OF CHANGES IN REACTOR DESIGN PARAMETERS Althogh or comptational procedre is designed to lead directly to the optimm combination of reactor parameters, it is nonetheless a matter of some interest to see how deviations of these parameters from their optimm vales will affect the performance of the reactor. The inflence of these parameters on reactor performance may be investigated by assigning specific pertrbed vales to each parameter in trn, the others retaining their reference vales, and performing the flx and eqilibrim fel-cycle calclations for the pertrbed cases. In some instances, a selected sbset of the npertrbed variables may be allowed to be reoptimied, sing ROD, if there is reason to sppose that sch reoptimiation will partially compensate for any adverse effect of the pertrbation. The effect of several specific departres from the reference 1 MW(e) design given in Table I is discssed in the following paragraphs. Reactor Plant Sie The effectiveness of the blanket (oter core) region depends very mch on its thickness. Normally, the blanket will contain a larger fraction of the salt inventory for a small than for a larger reactor. Ths, both the fel specific power and the breeding gain, for an optimied reactor, increase as the reactor plant sie is increased, as shown in Fig. 2. This is tre when the reactors are compared at eqal core life (the solid crves) or at eqal average core power density (the dashed crves). A brief listing of dimensions and other parameters for 5, 1, 2, and 4 MW(e) reactors is given in Table N. TABLE III Eqilibrim Fel-Cycle Cost Cost Element Cost mills/kwh(e) Fel inventorya.27 Salt inventorya.4 Salt makep.5 Moderator replacement.1 Processing.3 -- Sbtotal.76 Fel prodction credit Total fel-cycle cost.67 a Inventory charge 1% per annm. 212 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 197

6 LL. 1 CONSTANT CORE LIFE CONSTANT RATIO OF REACTOR 4 POWER TO CORE VOLUME ' I- : ' 7 3 <..> <..> cc - tj -LL ' >- 5 %/year 2 : :;;: -I- 4 ct: t!l L, years >.; ' V) ' 3 1 : I- tj : > 2 : ----I, kg/mw(e) l Fig. 2. Graphite Moderator Life REACTOR POWER [13 MW(e)] Effect of power level on MSBR performance. The sefl life of the graphite moderator is limited by radiation damage effects, cased by fast netrons. As discssed by Eatherly, 4 we have for the present adopted a limiting fast-netron flence (E > 5 kev) corresponding to ero net graphite volme change at the end of exposre. Since the fast-netron flx is almost entirely determined by the local power density (per nit vol- me of salt-pls-moderator), there is a nearly niqe relationship between maximm core power density, plant tiliation factor, and sefl core life, for a specified maximm flence. While one expects a higher power density to be accompanied by a redced fel inventory, there is in fact a power density above which increased netron leakage losses and other associated losses in breeding gain more than offset the redced inventory, and the fel yield and the conservation coefficient then decrease. These trends are exhibited in Fig. 3, which shows breeding gain, specific inventory, fel yield, and conservation coefficient as a fnction of core life. In this comparison, blanket, reflector, and plenm thicknesses were held constant, and the core sie was specified. Only the salt volme fraction was reoptimied, and it changed very little. Thorim Concentration The thorim concentration in the fel salt primarily inflences the ranim inventory and the breeding ratio. For a reactor configration very similar to or present reference design (bt having a slightly lower estimate of the reqired external salt volme) we have examined the inflence of thorim concentration on reactor performance over the range 1 to 14 mole% ThF4 Cross sections were careflly compted for each region in each case and iteratively adjsted to allow for resltant changes in reactor configration. In these calclations, core sie, radial and axial blanket thicknesses, and core salt volme fraction were all sbject to reoptimiation. TABLE IV Performance of Single-Flid MSBR's as a Fnction of Plant Sie Reactor Power, MW(e) Core height, ft Core diameter, ft Salt specific volme, ft 3 /MW(e) Fel specific inventory, kg/mw(e) Peak power density, W/ cm Peak flx ( > 5 ke V), 1 14 n/ (cm sec) Core life, years at.8 PF Leakage, n/ fissile absorption x Breeding ratio Annal fel yield, %/year Conservation coefficient c( t!l 6 GI :>... l>'l Cl _, LL.I... > Fig. 3. <..>... LL. t t----t---,.-t----tl8;:; LL.I c..,..., (/') (IJ 1- t ,,.."" "----t----t----t O> t t----t GRAPHITE CORE LIFE (year) LL. LL.. LL.I O> 14 Zl- <..> OZ... LL.I y I-> c(z >... er: t t t----t 12 :7: <..> Performance of 1 MW(e) MSBR as a fnction of core life (at.8 plant factor). NUCLEAR APPLICATIONS & TECHNOWGY VOL. 8 FEBRUARY

7 The basic interplay, of corse, is between rising breeding gain and rising inventory, as thorim and ranim concentrations are increased. Both the annal fel yield ( y) and the conservation coefficient (CC) shold exhibit a peak, when plotted as a fnction of thorim concentration, bt the peaks will occr at different places becase of the difference in weight assigned to the specific power. These trends are shown in Fig. 4. It may be seen that there is qite a broad maximm in the conservation coefficient in the vicinity of 12 mole% ThF4. Salt Volme Fractions Cre. In all of or calclations, the optimm salt volme fraction in the core has fallen in the range 12 to 15 vol%, with a carbon/fissile-ranim atom ratio close to 9. As indicated earlier, the volme fraction is rather closely determined by a balance between fel inventory, degree of netron moderation, and netron absorptions in the moderator; for the reference design, the optimm salt fraction was.132. Blanket. The volme fraction of salt in the blanket (oter core) is central to the whole concept of a single-flid, 1 MW(e) molten-salt breeder reactor. We have tested the effect of variations in salt fraction (in the radial blanket) nder the special assmptions of constant overall salt volme and constant oter diameter of the blanket region. Reslts of these calclations show that a broad optimm exists in the range of.35 to.6 for the salt fraction. The choice of.37 for the reference reactor was initially selected to permit, if desired, the se of a randomly packed ball bed in the blanket.!---+cc..._,..., 4;" 6 t=:=t==::;f====f-;;7xto-1 16 e; "' c ::; G (x 1)! a _ > tt... L r--=:::::=tY 8 c I 1) 8 >- 2 l l ;:;: LLl <>::::::>......_ REFERENCE DESIGN ; l l l s 1 Fig. 4. THORIUM CONCENTRATION (mole%) Inflence of thorim concentration on the performance of a single-flid MSBR. "' 233pa Removal Rate Rapid and inexpensive isolation of 233 Pa from the circlating fel stream is essential for economic breeder operation of the single-flid molten-salt reactor concept. The necessity for removing the Pa and allowing it to decay otside the netron flx may be seen from Fig. 5, in which we show the approximate loss in breeding ratio associated with Pa absorptions as a fnction of specific power and of the processing cycle time for continos isolation of Pa from the circlating salt stream. (This loss incldes both the netron losses in the 233 Pa and the destrction of latent 23 3.) As one wold expect, the Pa removal cycle mst be short compared to the 4-day mean decay life, if significant losses are to be avoided. The effect of the Pa removal rate (along with certain fission prodcts) on the breeding performance of MSR's is shown in Fig. 6. Note that for no Pa removal, the reactors considered are high performance converters rather than breeders. Removal of Fission Prodcts The fission prodcts may be divided into several grops, according to their chemical behavior in molten salt systems. The main grops, roghly in order of importance as netron poisons, are the noble gases, the rare earths, the noble and seminoble metals, the volatile florides, and the stable florides not readily separable from the carrier salt. ::.1.8 c g.6 LLJ LLJ a: a>.4 I.I) I.I).2 Fig. 5. l PROCESSING CYCLE TIME FOR Pa REMOVAL (days) Approximate loss in breeding ratio associated with netron absorptions in 233 Pa for a singleflid MSBR. 214 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 197

8 r I- : 3-day Pa CYCLE "" l V'l "' > : Fig. 6. NO Pa REMOVAL RARE EARTH REMOVAL CYCLE (days) Conversion ratio vs processing cycle timessingle-flid MSBR. An otstanding characteristic of the molten salt reactor is the relative ease with which the noble gases-notably 135 Xe-can be removed from the salt stream. This aspect is discssed in detail by Scott 4 ; sffice it to say here that we estimate a residal poisoning effect (redction in breeding ratio), de to 135 Xe together with other noble gas fission prodcts and their daghters, of not more than.5. The effect of the rare earth removal cycle on the breeding performance of single-flid MSR's (at varios Pa removal rates) is shown in Fig. 6. The ability to process the fel rapidly for rare earth removal is one of the most important characteristics of the molten salt reactors. While rapid rare earth and Pa removal are reqired for good breeding performance, there is considerable flexibility to trade off nclear performance for decreased processing rate (and processing cos at the discretion of the designer. The "noble metals," which do not form stable florides and do not remain in the fel salt, wold have a serios adverse effect on the breeding ratio if they simply attached themselves to the graphite moderator in the core. The nclides in this grop, which incldes Nb, Mo, Tc, R, Rh, and several others of less importance, have a combined yield of.35 and, if they accmlated in the reactor core over a long enogh period of time, wold eliminate any chance of breeding. Most of the nclides in this grop, however, have small netron-absorption cross sections and satrate very slowly, or else they have very low yields. It is primarily 95 Mo, 99 Tc, 11 R, and 1 3Rh Perry and Baman whose potential poisoning effect is of real concern. The chemical behavior of this grop of fission prodcts is discssed in detail by Grimes. 5 For the present prpose, it is sfficient to note that we believe "'1% of these nclides will remain in the core regions of the MSBR. On this basis, the poison fraction, P(t ), of all nclides in the grop and the time-averaged poisoning, P(t), p to time t after start-p with clean graphite, are shown in Fig. 7. Since we anticipate replacement of the graphite after... 4 years, we may see from Fig. 7 that the average redction in breeding ratio attribtable to the accmlation of these nclides is.4. This effect is more than compensated by the brnot of 1 B in the graphite, and these effects are assmed to cancel, in or calclations. The "semi-noble metals," inclding Zr, Cd, In, and Sn, are assmed to be removed in an adjnct to the Pa removal process. The poison effect of this grop is small and they cold also be removed, if necessary, by a gradal discard of fel salt (from which the ranim wold be recovered by florination). The volatile florides of Br and I are relatively nimportant and are removed in the florination -Ic:( ex:.1 <.!>.8 : -Cl ex: co : <.!> : c:( ::i:::.6.4 Fig. 7. TIME (calendar years) Change in breeding ratio de to noble-metal fission prodcts in MSBR. NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY

9 step of the Pa removal process. The stable florides of Rb, Sr, Cs, and Ba are removed by the discard of salt from the rare earth process on a long cycle, typically eight years. REACTOR START-UP AND APPROACH TO EQUILIBRIUM The preceding discssion has dealt only with the eqilibrim fel cycle, in which all nclide concentrations-and particlarly those of the ranim isotopes-are time-independent and have their asymptotic vales. This condition is in fact approached rather rapidly in an MSBR, becase of its relatively high specific power. Nonetheless, it is of interest to examine the transient fel-cycle characteristics of the reactor when it is initially feled with enriched ranim (2:1;U) or with pltonim. There is considerable latitde in specifying the initial fel loading for the reactor, even when the eqilibrim fel cycle is well defined, since the initial thorim concentration may, at the discretion of the operator, be qite different from the eqilibrim vale. We have, in fact, tried different initial thorim concentrations, and find no advantage in choosing a vale different from the eqilibrim one. In the following discssion, therefore, it will be nderstood that the thorim concentration is held constant at 12 mole%. 235U Start-Up Start-p of the reactor on enriched ranim ( 93% 235U) reslts in an initial conversion ratio sbstantially less than nity. However, becase 233 is mch more reactive than 235U in the MSBR spectrm, the export of fissile material from the reactor plant may begin qite soon after start-p, and indeed well before the conversion ratio actally exceeds nity. Time-dependent inventories of the ranim isotopes are shown, for the reference single-flid 1 MW(e) MSBR, in Fig. 8, while the conversion ratio and the cmlative net feed of fissile material to the circlating fel salt are shown in Fig. 9. It may be noted that very little additional fel is spplied to the reactor in excess of the initial critical mass reqired for start-p. There is, of corse, some loss in performance, averaged over the life of the reactor, associated with this initial phase of the fel cycle. The average redction in breeding ratio, over the life of the plant, is "'.18, while the increase in the present vale of the 3-year fel-cycle cost is +.2 mill/kwh(e). P Start-Up A sefl attribte of the molten-salt reactor is the capacity of the LiF /BeF2/ThF 4 carrier salt to contain pltonim in the form of PF3, with a solbility in excess of 1 mole% at the operating temperatres of the MSBR. We have considered the start-p of the MSBR on pltonim whose composition is typical of the material discharged from a "' >- :: 1- LI.I > ::;: LI.I 1- Vl > Vl 1.2 L O.Sl.-IW A.47 \ / s ar.:::: '-'--'--=--' EXPOSURE TIME- EQUIVALENT FULL-POWER YEARS > s..j :::> i l.8 (,,) EXPOSURE TIME - EQUIVALENT FULL-POWER YEARS Fig. 8. Uranim isotope inventories-msbr start-p with enriched ranim. Fig. 9. Conversion ratio and cmlative fissile feedenriched ranim start-p. 216 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 197

10 PWR or BWR with a fel exposre in the neighborhood of 2 MWd/T, specifically, 59. 7% 239P, 24.1% 24 P, 12.2% 241 P, and 4.% 242 P. A notable difference between the pltonim and ranim start-p cases is the mch lower initial inventory of pltonim, which is de to the very large effective cross sections of the pltonim isotopes in the MSBR spectrm. These isotopes brn ot very rapidly, however, and additional pltonim mst be spplied for a time ntil nearly enogh 23 3 is prodced to sstain the chain reaction by itself. Thereafter, the concentrations of 239P, 24 P, and 241 P decrease rather sharply, while that of 242 P decreases more slowly ntil the pltonim ceases to make a net positive contribtion to the reactivity. At this point, the pltonim (now mainly 242 P) is isolated from the circlating fel salt and discarded. If this is done, the reslt is a self-sstaining 23 3-feled reactor in which the 235 U and 236 U concentrations are temporarily below their eqilibrim vales, and the asymptotic breeding ratio is approached from the high side. These featres of the pltonim start-p are exhibited in Figs. 1, 11, and 12. The average breeding ratio, over a 3-year plant life at.8 plant factor, is redced by,...,8 relative to the eqilibrim cycle. The present vale of the fel-cycle cost is actally decreased by,4 mill/kwh(e) if one assigns to the fissile pltonim a vale f that of enriched 235 U. OPERATION WITH LIMITED FUEL PROCESSING The combination of good breeding performance with low fel-cycle cost, which is associated with economical processes for the rapid removal of fission prodcts and protactinim, is an important featre of the molten-salt reactor concept. Nevertheless, qite satisfactory fel costs can be achieved in the absence of fel processing for removal of protactinim or fission prodcts, althogh the reactor will of corse not breed when operated in this fashion. In examining this alternate mode of operation, we postlate the occasional batch discard and replacement of the entire inventory of carrier salt, inclding the thorim, bt not inclding the ranim; the latter is to be recovered by floride volatility (as was done with the MSRE fel salt) and recycled to the reactor. The removal of noble-gas fission prodcts from the salt by gas sparging and of noble-metal fission prodcts by.. en 4..,, c::: I- > :::;;: I-..,, >-..,, 3 2 OPERATING TIME - EQUIVALENT FULL-POWER YEARS 1.6 en 2...,, 1.2 c::: ( I- >.8. R,, :::;;: I-..,,..,, > L ' OPERATING TIME - EQUIVALENT FULL-POWER YEARS Fig. 1. Pltonim isotope inventories-msbr start-p with PWR pltonim. Fig. 11. Inventories of ranim isotopes and total pltonim for pltonim start-p of MSBR. NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY

11 en....,, I- :::> i... :::> :E :::> ""T" ;::.8 I- cc _. CONVERSION RATIO.,, 8 "' URANIUM SOLD ( ) o, 4..w:; a J..6 J2 8 OPERATING TIME - EQUIVALENT FULL-POWER YEARS Fig. 12. Cmlative prchases and sales of fissile material and conversion ratio MSBR with pltonim start-p. deposition. and by escape to the off-gas system, as observed m the MSRE, are assmed to occr. An optimm salt discard rate exists, for which the cost of replacing the salt (and recovering ranim) is balanced against the fel makep cost which depends on the conversion ratio and henc on the level of fission-prodct poisoning. In addition, however, there is a minimm discard rate reqired to limit the concentration of rare-earth triflorides in the circlating fel salt to an acceptable level (<1 mole%). This consideration per se will place a limit of ""1 years on the salt replacement interval. We have investigated the trend in fel-cycle costs, as a fnction of the salt replacement interval, for a molten-salt converter reactor which is defined in terms of its general nclear characteristics, sch as netron leakage and fel inventory, bt whose engineering design has not been specified in detail. We fond the fel costs to be rather insensitive to the salt replacement interval from 3 years to 8 or 1 years, with a broad minimm at 5 to 6 years. The conversion ratios ranged from.84 for a 4-year salt-replacement interval to. 78 for a 1-year interval. We have not yet made any detailed estimates of costs for the florination plant and for the chem - ical treatment plant necessary to control the composition of the salt over long periods of time. Allowing.1 mill/kwh(e) for this eqipment and its operation, and.1 mill/kwh(e) for graphite replacement in the reactor core, we estimate the fel-cycle costs for the converter reactor with a salt discard. cycle, to be O. 7 to.8 mill/kwh{e). ' Compared with the MSBR, the principal cost differences are a saving of "".2 mill/kwh(e) in fel processing, an increase of.2 to.3 mill/kwh{e) for fel brnp, and possibly some redction of fel and salt inventory charges [<.1 mill/kwh{e)]. Ths, the fel-cycle cost for the converter reactor, withot chemical processing for removal of protactinim or of fission prodcts, appears to be within.1 mill/kwh( e) of the cost for the breeder reactor. TEMPERATURE COEFFICIENTS OF REACTIVITY Expansion of the single fissile-fertile salt in the one-flid reactor redces the density of most of the absorbing materials in the same proportion, so that one might expect a very small prompt temperatre coefficient of reactivity. The density coefficient of the salt is in fact very small. In addition to this, however, there is both a positive contribtion to the salt-temperatre coefficient associated with the shift in thermal-netron spectrm with increasing salt temperatre, and a negative contribtion associated with the Doppler broadening of resonance captre lines in thorim. The latter predominates, reslting in a prompt negative coefficient of -2.4 x 1 5 ok/k/ C. The graphite moderator contribtes a positive com.ponent to the overall temperatre coefficient, attribtable to an increase in the relative cross section of 23 3 with increasing netron temperatre. This reslts in an overall coefficient which is very small, thogh apparently negative, i.e., -.5 x 1-5 / C. The prompt negative salt coefficient will largely govern the response of the reactor for transients whose periods are several seconds or less. The small overall coefficient will provide little inherent system response to impressed reactivity changes, and it will conseqently be necessary to provide control rods to compensate any reactivity changes of intermediate dration. Long-term reactivity effects, sch as those associated with the fel cycle, are compensated by adjstment of the fel concentration in the salt. SUMMARY AND CONCLUSION While a fll economic optimiation of the single-flid molten-salt breeder reactor has not yet been ndertaken, it is apparent that good breeding performance can be achieved in conjnction with nsally low fel-cycle costs, sbject to the sccessfl completion of chemical processing developments now nder way. That is, a breeding 218 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 197

12 ratio of > 1.6, together with a specific fel inventory of <I.5 kg/mw{e), will be attainable with fel-cycle costs of "". 7 mill/kwh{e). The associated fel dobling time is ""2 years. The MSBR fel cycle may be initiated either with enriched ranim or with pltonim discharged from light-water reactors. The redction in breeding ratio, averaged over the life of the reactor, is <.2 and.1, respectively, for the ranim and pltonim start-p cases relative to the eqilibrim case. The present vale of the fel-cycle cost is increased, relative to that for the eqilibrim cycle, by "".2 mill/kwh{e) for the ranim-start-p case, while for the pltonimstart-p case the cost appears to be.4 mill/kwh less, based on a fissile pltonim vale eqal to that of 235 U. Development of the technology for molten-salt reactors themselves and for the associated chemical processes for removal of protactinim and fission prodcts need not necessarily be carried ot on precisely the same time schedle, since an economically attractive fel cycle may be achieved with a salt-discard cycle, reslting in a conversion ratio of.8 to.9. A given reactor, operated initially in this way, cold sbseqently be operated as a breeder by the addition of appropriate processing eqipment to the reactor plant. It is ths apparent that there is considerable latitde in the mode of operation of molten-salt reactors, with only small variations in fel-cycle cost. We believe that this can facilitate the devel- opment of the molten-salt reactor and of the associated chemical-processing technology on time schedles appropriate to each, and will encorage an orderly progress toward the achievement of an economical breeder reactor. ACKNOWLEDGMENTS The athors wish to acknowledge the contribtions made by many of their colleages to the work reported here, and in particlar those made by R. S. Carlsmith, W. R. Cobb, E. H. Gift, and. L. Smith. This research was sponsored by the U.S. Atomic Energy Commission nder contract with the Union Carbide Corporation. REFERENCES 1. E. S. BETTIS and R. C. ROBERTSON, "The Design and Performance of a Single-Flid MSBR,'' Ncl. Appl. Tech., 8, 19 (197). 2. A. M. PERRY, "Inflence of Netron Data in the Design of Other Types of Power Reactors," ORNL-TM- 2157, Oak Ridge National Laboratory (March 8, 1968). 3. M. E. WHATLEY, L. E. McNEESE, W. L. CARTER, L. M. FERRIS, and E. L. NICHOLSON, "Engineering Development of the MSBR Fel Recycle," Ncl. Appl. Tech., 8, 17 (197). 4. E. P. EATHERLY and D. SCOTT, "Graphite and Xenon Behavior and Inflence on MSBR Design,'' Ncl. Appl. Tech., 8, 179 (197). 5. W. R. GRIMES, "Molten Salt Reactor Chemistry," Ncl. Appl. Tech., 8, 137 (197). NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY