GRAPHITE AND XENON BEHAVIOR AND THEIR INFLUENCE ON MOLTEN-SALT REACTOR DESIGN DUNLAP SCOTT and W. P. EATHERLY Oak Ridge National Labovatory, Nucleayr Division, Oak Ridge, Tennessee 37830 Received August 4, 1969 Revised October 2, 1969 Existing data on dimensional changes in gra- phite have been fitled to parabolic lemperature- sensitive curves. From these, the graphite life, radiation-induced stresses, and permissible geometries have been calculated. Il is concluded existing materials can be utilized in a molten-salt reactor which has a cove graphite life of about four years, without serious cost penally. Fission product xenon can be vemoved bY sparging the fuel salt with helium bubbles and removing them after envichment. With reasonable values of salt-to-bubble transfer coefficient and graphite permeability, the penalty lo breeding ratio can be reduced to <0.5%. | INTRODUCTION One of the attractive aspects of the molten-salt reactor concept is that even the most stringent of the present materials or process limitations per- mit reactor designs having acceptable economic performance. In this paper we consider, as examples, the effects of finite graphite lifetime and xenon poisoning on MSBR design. graphite lifetime implies periodic replacement of the graphite; neutron economy requires the re- moval of the bulk of '*°Xe from the core to keep the xenon poison fraction below the target value of 3%. The major economic penalty associated with graphite replacement would be the load factor penalty associated with taking the reactor off- stream. This cost can be circumvented by assur- ing that the graphite will maintain its integrity for at least the time interval between normal turbine NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 Finite FEBRUARY 1970 KEYWORDS: molten-salt re- actors, breeder reactors, ra- diation effects, temperature, graphite moderator, stresses, porosity, xenon-135, poison- ing, helium, bubbles, design, economics, reactor core, MSBR maintenance requirements; i.e., the downtime re- quired for scheduled maintenance should coincide with that for graphite replacement. Hence, to avoid load-factor penalties associated with graph- ite replacement, we have set as a minimum re- quirement that the graphite have a life of two years; at the same time, a longer life would be desirable and consistent with the power industry’s objective of increasing the time interval between turbine maintenance operations. Thus, in the reference design the reactor performance is con- strained to yield a graphite life of about four years, and this paper points out the basis for that value. The removal of *°Xe from the core also puts a constraint on the graphite. The fuel salt must be excluded from the graphite to prevent local over- heating and also to decrease fission-product poi- soning, and this, in turn, requires that the graphite pore diameters not exceed one micron. This, however, is not a limitation, for the xenon removal will be shown below to require a gas permeability of the order of 10”° cm?/sec, and such a value requires pore diameters of ~0.1 p. Even though xenon is excluded from the graph- ite, it needs to be removed from the salt stream if the desired neutron economy is to be attained. This removal is accomplished by injecting helium bubbles into the flowing salt, which transfers the xenon from the salt to the bubbles and effectively removes xenon from the core region. In the following sections we shall discuss in some detail: the method of analysis of existing data on radiation damage to permit prediction of the graphite lifetime in MSBR cores; the applica- tion of these damage rates and the radiation- induced creep to calculate induced stresses in the graphite; the considerations involved in the distribution and removal of !*°Xe and other noble gases; and last, the method proposed for safely 179 Scott and Eatherly collecting and disposing of these gases. We con- clude that graphite and the removal of xenon present no questions of feasibility, but require only minor extensions of existing technology. GRAPHITE LIFETIME It has been recognized for several years that under prolonged radiation exposure, graphite be- gins to swell extensively, even to the point of cracking and breaking into fragments. However, data at high fluences and within the operating temperature ranges anticipated for molten-salt breeder reactors were largely nonexistent. Nevertheless, by using existing data, it was possi- ble to estimate graphite behavior over the range of MSBR conditions. For a large group of commercial graphites—including British Gilso- graphite, Pile Grade A, and the American grades AGOT and CSF-it was found that the volume distortion, v, could be related to the fluence, &, by a parabolic curve, v =Ad + Bd® (1) where A and B are functions of temperature only. The fit of this equation to the experimental data is excellent for the isotropic Gilso-graphite, but the relation is approached only asymptotically for the anisotropic graphites. We may write the fluence as the product of flux and time, i.e., & = ¢f. The behavior of v is to decrease to a minimal value, v,, and then increase, crossing the v = 0 axis at a defined time, 7, given by 0= Agr + Bl¢r)? . 2) Clearly, in terms of v, and 7, Eq. (1) can be rewritten as v = 4y, t (1 - —t—) T T For isotropic graphite, the linear dimensional changes will be given approximately by one-third the volume change. If the graphite is anisotropic, the preferred c-axis direction will expand more quickly, the other directions more slowly. This will induce a more rapid deterioration of the material in the preferred c-direction. As a consequence, we require that the graphite be iso- tropic, and can rewrite Eq. (2a) in terms of the linear distortion, G, G=_1.V=4_Vflt(1__t.> - (2a) ~ 2 3 3 7T (2b) and this relation will be used hereafter. The best values for the parameters ¢7 and v, were found to be o7 = (9.36 - 8.93 X 10™° T) x 10°* nvt (E > to keV) 180 GRAPHITE AND XENON BEHAVIOR NUCLEAR APPLICATIONS & TECHNOLOGY and v, =-12.0+8.92x107° T % where T is the temperature in °C, valid over the range 400 to 800°. At 700°C, these yield ¢r = 3.1 X 10*# with a 90% double-sided confidence limit of +0.2 X 10*2, and v, = -5.8%, with the limit +0.2. Figure 1 shows the behavior of the linear distor- tion as a function of fluence with temperature as a parameter. As pointed out in the introduction, it is neces- sary that the graphite exclude both salt and xenon. The first requires the graphite to have no pores larger than ~1pu in diameter; the second, as will be seen below, requires the graphite to have a permeability to xenon of the order of 10~® cm?2/ sec. Clearly, as the graphite expands and visible cracking occurs (v > +3%), these requirements will have been lost. Lacking definitive data, we have made the ad hoc assumption that the pore size and permeability requirements will be main- tained during irradiation until the time when the original graphite volume is reattained, namely, when ¢ = 7 as defined above. To estimate the lifetime of an MSBR core, we must take into account the strong dependence of T +2 [ 850°C /800° /750° / 700°/ +1 | S o Z o o 0 / / / 650 - @ o - : / / a O © 1 600/ < / / / w Z - 700 - \;/// - -3. } 0 1 2 3 FLUENCE ® x 10722 (E >50 ke V) Fig. 1. Graphite linear distortion as a function of fluence at various temperatures. VOL. 8 FEBRUARY 1970 on temperature, and the temperature of the core graphite will depend on the fuel salt temperatures, the heat transfer coefficient between salt and graphite, and the gamma, beta, and neutron heat generation in the graphite. In all core designs which have been analyzed, the power generation (i.e., fission rate) has been found to vary closely as sin(rz/1) along the axial centerline, where z2/1 is the fractional height, and I is the effective total core height. Thus, the heat generation rate in the graphite will vary as q= 4o + 41 sin(rz/1) , 3) where ¢, approximates the rate due to delayed gamma, beta, and neutron heating, and ¢, approxi- mates the maximum prompt heating. Represent- ing the actual graphite core prisms as cylinders with internal radius @ and external radius b, we can readily calculate the internal temperature distribution assuming g is not a function of radius. The result is’ q 2 2 log(7 /a) T = Ta - IR' (’V -a ) + —__mg(b/a) X [Tb - T, + 74%{. (b* - az)] , (4) where T, and T, are the surface temperatures at a and b, respectively, and K the graphite thermal conductivity. Since the axial heat flow is negli- gible, the heat, @, that must cross the graphite surfaces per unit area becomes __9a _ K _ d (p? _ g2 Q“—-2+alogb/a|:Tb T“+4K(b a)] qb K _ oo n _ 4 2 _ 2 Qb_-_2_+blogb/a [Tb Ta+4K(b a)](5) and if % is the heat transfer coefficient between the salt and graphite interface, then also Q. =h(Ta - TO) Qb =h(Tb = TO) ’ (6) where T o is the bulk salt temperature. The coef- ficient % is calculated from the turbulent-flow Dittus-Boelter equation® 04 Ko h = 0.023 Re®® Pr 57 (7) where Re is the Reynolds number Pr is the Prandtl number K, is the salt thermal conductivity. We are assuming the coolant channels at the outer surfaces of the cylinder will have the same effec- tive hydraulic radius as the interior channel of NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 Scott and Eatherly GRAPHITE AND XENON BEHAVIOR radius ¢. Finally, since the heat capacity of the salt is a weak function of temperature, in keeping with the sinusoidal variation of power density along the axis, the salt temperature, T, becomes 1 T2 T0=§[Tf+T,~-(T/-T,-)cos—l—:| (8) in which T, and 7, are the entering and exiting temperatures, respectively, of the salt. Equations (3) through (8) completely determine the temperatures in the graphite. As we shall see below, the radiation-induced strain in the graphite along the z-axis, i.e., A2/z, is given by Az 2 b 24 _ T ~ (—z——z—b_a)faG( Yv dr where G(T) is the damage function of Eq. (2b). A similar expression applies to the radial strain. A negligible error is introduced by a_pproximating the right-hand side by G(T), where T is the aver- age temperature over the cross section; there- fore, Az — = G(T) . 9) Since T varies with z/l, each point along the cylinder will have a different life, 7(T), defined by Az/z = 0. The minimum value of these 7(T) thus defines the time at which the cylindrical prism should be replaced based on our criterion. The properties of the fuel salt and graphite which are required in the above equations are given in Table I. The graphite is assumed to be similar to the British Gilso-graphite, although there are other coke sources than Gilsonite which lead to isotropic materials. Calculations have been made on two core con- figurations. Case No. CC-58 is the reference design discussed elsewhere’ in this series of TABLE 1 Materials Properties Used in Graphite Lifetime and Stress Calculations Fuel Salt Thermal conductivity, W/(cm °C)| 1.29 X 10~2 Specific heat, Wsec/(g °C) 1.36 Viscosity, cP (7.99 x 107%) exp(4342/T °K) Graphite 37.63 exp(-0.7T °K) 5.52 % 10~% + 1.0 X 10°(T °C) 1.9 x 10° 0.27 (5.3 -1.45x 107> T + 1.4 x 107°T2%) x 107" k in psi~* n/(cm® sec) T in °C. Thermal conductivity, W/(cm °C) Thermal expansion, (°C)™" Young’s modulus, psi Poisson’s ratio Creep constant, k FEBRUARY 1970 181 Scott and Eatherly papers. This design optimizes the fuel conser- vation coefficient® with the peak power density constrained to a value of 63 W/cm?® to prolong graphite life. Case No. CC-24 is the identical core scaled to a smaller geometry (half the volume of No. CC-58) and with the constraint on graphite lifetime raised. In both cases the fuel conservation coefficient is 15.1 [MW (th)/kg]?. The damaging flux in case No. CC-24 is also more typical of the case when the core is optimized with no constraints. The pertinent core param- eters® are given in Table II. TABLE II Relevant Core Characteristics for Calculation of Graphite Lifetime and Stresses Case No. Case No. CC-24 CC-58 Peak flux (E > 50 keV), n/(cm® sec)| 5.15 x 10** 3.2 x 10" Salt flow per unit area, g/(cm® sec)| 1.39 x 10° 8.21 x 10° Heat generation, delayed, W/cm® 1.16 0.71 Heat generation, prompt, W/cm® 7.17 4.39 Salt inlet temperature, °C 550 950 Salt outlet temperature, °C 700 700 \Q= 3'2 X ]Ol“ 4 L\" GRAPHITE AND XENON BEHAVIOR CORE LIFE AT 80% PLANT FACTOR (year) 3 2 0.60 0.75 0.90 INTERNAL RADIUS, a (in cm) Fig. 2. Core lifetime as a function of graphite prism dimensions for cores with peak damage fluxes of 3.2 and 5.15 x 10™ nv (E > 50 keV). In all cases the ratio of the radii, b/a, is 6.67. 182 NUCLEAR APPLICATIONS & TECHNOLOGY Since the salt-to-graphite ratios in the core are determined by nuclear requirements, and the salt flow by cooling requirements, the only vari- able left at this point is the absolute value of the internal radius g, which scales the size of the graphite prisms. Figure 2 shows the lifetime of the central graphite prism as it is affected by the radius a; there is an obvious decrease in core life as the radius ¢ increases and the internal graphite temperatures climb. We note that the ratio of fluxes in the two cases is 1.61; ata= 0.9 cm, the corresponding reciprocal ratio of lifetimes is 1.74, the additional gain in lifetime at the lower flux being due to decreasing graphite tempera- tures. The latter case has been studied in more detail since it corresponds to the current reference design concept. For this design the equivalent radii are a = 0.762 cm b = 5.39 cm . The associated temperature distributions for the central core prisms are given in Fig. 3, and the local lifetimes 7(T) as a function of z// in Fig. 4. The life of the prism, i.e., the minimal 7(7), occurs at 2/l = 0.55 and has a value of 4.1 years at 80% plant factor. The entire prism will change length as given by the integral of the right-hand side of Eq. (9) over the length of the prism; this is shown as a function of time in Fig. 5. The radial distortion for various times is shown in Fig. 6, and gives the prism a double hourglass shape toward the end of life. 800 — = 700 MAXIMUM INTERNAL P s TEMPERATURE=_ |~ = = P INNER AND OUTER S / 7z é SURFACE g / 7 TEMPERATURE 5 00 A BULK SALT =4 TEMPERATURE = 500 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 FRACTIONAL HEIGHT (z/1) Fig, 3. Temperatures associated with the central gra- phite prism as a function of vertical position for the reference core design. VOL. 8 FEBRUARY 1970 7(T) AT 80% PLANT FACTOR (year) Scott and Eatherly N ™™ | \ / N — 4 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 FRACTIONAL HEIGHT (z/1) Fig. 4. Local lifetime 7(T) as a function of vertical position for the central graphite prism of the reference core design. The cost of replacing graphite will include both material procurement and labor. We have esti- mated these costs based on an industry supplying the order of ten or more reactors. The total operating cost associated with graphite replace- ment would amount to ~0.2 mill/kWh for a two- year life, or 0.10 mill/kWh for a four-year life. (Capital investment would also be required for the replacement equipment and is included in the capital cost estimates of Bettis and Robertson.”) Thus, the cost penalty associated with graphite replacement is not a crippling one, although it is large enough to merit considerable effort on graphite improvement. Assuming pyrolytic impregnation, as discussed by McCoy etal.,’® successfully excludes xenon from the graphite, three significant material requirements need to be met: the graphite must be isotropic, have entrance pore diameters <1 u, and have a radiation stability at least as good as the Gilso-graphite. All of these require- ments can be met with existing graphite tech- nology, although all have not been met in a single graphite of the dimensions required. Unquestion- ably, there will be production difficulties in initially producing such a graphite, but the prob- lems will be in process control rather than in basic technology. NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 1970 GRAPHITE AND XENON BEHAVIOR 0 / 0 1 2 3 4 TIME AT 80% PLANT FACTOR (year) AXIAL DISTORTION (%) ' N—_— | Fig. 5. Axial distortion of the central graphite prism as a function of time for the reference design. INTERNAL STRESSES IN THE GRAPHITE In the preceding section we have tacitly as- sumed that the thermal and radiation-induced stresses developed during the lifetime of the graphite are not limiting. We turn now to validate this assumption. We again consider the central prism as in the preceding section, and have for the constitutive equations, € 2% [6; - H(ff,' + 0,)] + k¢[0i '%(01' + Ok)] + &, where (10) e; = strain in the ¢’th direction o, = stress E = Young’s modulus u = Poisson’s ratio B = secondary creep constant ¢ = damaging flux dots = time derivatives g = dG/dt = damage rate function defined from Eq. (2b). In cylindrical coordinates, ¢ (r, 6,2). In the absence of externally applied stresses and be- cause of the vanishing of ¢ at the ends of the core prism, the z component of strain will not be 9 function of #» and # (plane strain). We shall assume E is a function of T, but will not let 2 or ¢ vary with » and 6. Then Eq. (10) can be inte- grated in closed form. The resulting dimensional changes in the prism are Az _ Aa Z a b (11) 183 Scott and Eatherly GRAPHITE AND XENON BEHAVIOR 0 . ; l \ /\48 mos. O, _)kl(b (Ago-BAg1)+ %Aglt . (13) 6 mos. \‘\ // It can also be shown that g = g(7) with an error = \ - not exceeding 2% for the parameter values of = " \ N \ interest. ;C:’ \ \\ 42 mos. » . The thermal stresses have the same interre- S \ lationships as the radiation stresses, and are = \\ \( 36 mos. . given by g‘ -2 \Q\ L/ — 0-00) = 0, 0gd) = 00) & ;; 18 mos. and 24 mos. £ 30 mos. 0z0 = T S (T -T()] et . (14) -3 We may substitute this result into Eq. (12) to give 0 0102 03 04 05 0.6 0.7 0.8 0.9 1.0 the maximum Stress o, (b) throughout the entire FRACTIONAL HEIGHT (z/1) life of the central prism. For the reference design concept, the maxi- mum stresses also occur at the point z/I = 0.55 Fig. 6. Radial distortion of the central graphite prism and are shown in Fig. 7. The stresses reach a as a function of vertical height for the refer- maximum at the end of life and amount to only ence design. Times are calculated at 80% plant 490 psi. Since the isotropic graphite which is factor, presumably to be used in the reference design would have a tensile strength in the range 4000 to where G is the average value of G over the cylin- drical cross section. Hence the prism behaves locally as though it were at the average radiation distortion G. 500 L The tensile stresses are maximum at the out- side surface of the cylinder, and are given by 0,(0) =0, 0y(d) = 0,(b) 400 and / 0. = 7o e [Jlg gt et + 00e™ (12) / 300 with g = Ek ¢ 2(1 - p) - 200 , , We may consider the initial stress o,, to be the thermal stresses introduced as the reactor is brought to power, and these anneal out exponen- tially with time because of the radiation-induced 100 creep. We are interested only in large times ¢, / and if we set Ag =g -g0) . / and remember g is a linear function of time g=27=57<1‘7>=g°+g1t’ -60 0 1 2 3 4 then TIME AT 80% PLANT FACTOR (year) SURFACE STRESSo OR o, (psi) Ag = A + Ag,t & &o &1 Fig. 7. Axial or tangential surface stress at z/I = 0.55 and Eq. (12) thus takes the asymptotic form as a function of time, 184 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 1970 5000 psi, there appears to be no reasonable proba- bility that these stresses can cause graphite failure. XENON-135 BEHAVIOR Because of its high-neutron-absorption cross section, it is important to keep xenon out of the eraphite and also out of the fuel salt. The **Xe comes from two sources, directly from the fission of uranium and indirectly from the g decay of the fission products tellurium and iodine. Tellurium probably exists as a free metal which is believed to be insoluble in molten salt and tends to migrate to the available surfaces such as the core vessel walls, the heat exchanger, the graphite moderator, and even any circulating gas voids which may be present. However, “°Te has a relatively short half-life, <3 min, and it is conservative for our purposes to assume that most of it decays into iodine before it can leave the fuel system. The iodine forms stable iodides in the molten-salt fuel and will remain with the fuel unless steps are taken to remove it. Since the half-life of **° is 6.7 h, it appears possible to process the salt stream at a relatively slow rate to remove a large portion of the iodine before it decays into xenon. The equivalent yield of **Xe in the processed salt can then be repre- sented by U _ ,U U _ 1 YXe yxe+ V1 <1 Tp >\I+ 1) ’ where y)l(Je= direct yield of xenon from uranium of mass U yU = chain yield of T from uranium of mass U decay constant of '*°1 Al T, = time required to process the entire salt inventory for the removal of iodine. It is apparent that the effective yield cannot be reduced by iodine processing alone to below ~1.1% for 233U fission.! As shown in Fig. 8, a relatively high processing rate of 62 gal/min for 1500 £t * of fuel only reduces the effective yield of 133%e to ~0.0225. One scheme for processing a side stream uses HF for converting I” to I, and then removing the HF and I, by purging with helium or H,. However, it would still be necessary to re- move additional xenon, and so stripping the fuel salt with helium is the preferred process for re- moving “°Xe. Xenon is very insoluble’ in molten lithium- beryllium fluoride and obeys Henry’s law for gases; at 650°C the pressure coefficient of solu- pility is 3.3 X 107° moles(Xe)/[ecm® (salt) atm]. NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 Scott and Eatherly GRAPHITE AND XENON BEHAVIOR 6 235y yye = 0.0024 y1 = 0.0617 233U yye = 0.0111 yy = 0.0505 5 4 — // 233 ’,/::::::: 3 L/ / 0 200 400 600 800 TIME TO PROCESS COMPLETE FUEL INVENTORY (min) EQUIVALENT YIELD OF '33Xe (%) 62 gal/min FOR 1500 ft® FUEL INVENTORY 1000 Fig. 8. Effect of iodine removal on 135%e yield. Because of the relatively high xenon partial pres- sure at very low concentration, it will tend to leave the salt through any free surface available before the concentration becomes high enough to make a significant contribution to the ***Xe poi- soning. In the MSBR, the free surfaces of impor- tance are those associated with the voids in the graphite and the entrained helium bubbles. The graphite planned for MSBR use, but without sur- face impregnation, has a bulk void volume avail- able to xenon of ~10% and could contain a large inventory of xenon if there is an umimpeded flow from the salt to these voids. Almost all the xenon poisoning in the MSBR will result from the neu- tron absorptions in the '*°Xe within the graphite, and so the graphite should have those properties which keep xenon concentration in graphite low. The concentration of xenon in the graphite is controlled by the concentration in the salt, the mass transfer coefficient from the salt to the graphite, the total surface of graphite exposed to the salt, the diffusion coefficient of xenon in the graphite, and the void fraction available to xenon. The mass transfer coefficient from the salt to the graphite is very strongly influenced by the char- acteristics of the salt flow boundary film which, together with the area of graphite exposed to the FEBRUARY 1970 185 Scott and Eatherly salt, are determined by the heat transfer condi- tions required to cool the graphite. However, the effective diffusivity and available volume of the graphite can be controlled during manufacture by impregnating the surface with a thin layer of material having low permeability. The base stock graphite which would be used in the core will probably have properties which are dictated by radiation damage and lifetime considerations and not those which affect xenon concentration in the voids. However, as will be illustrated below, a sealed surface will act as an effective barrier to the transfer of xenon into the graphite interior if the value for the diffusivity of the sealed sur- face can be controlled to <~0.25 x 10~ ¢m®/sec, and the associated void volume is 1%. As dis- cussed by McCoy et al,” experiments have been performed which indicate that the application of such a sealant is feasible. With the above condi- tions of permeability and available void volume met, the xenon poisoning in the core becomes controlled by its concentration in the salt and therefore by our ability to process the salt stream for xenon removal. The salt will be processed for xenon removal by injecting small bubbles of helium into the fuel salt stream and then removing them after they have taken up a portion of the xenon present. The rate of transfer of xenon to the bubbles is affected by the concentration of xenon in the salt, the total surface area of bubbles, the mass transfer coef- ficient of the dissolved xenon from the salt to the entrained bubbles, and the residence time of the bubble in contact with the salt. We can control all of these except the mass transfer coefficient to the bubbles, and a good value for the coefficient is difficult to determine. A review® made of existing data on rates of mass transfer to gas bubbles points to reliable equations for predicting mass transfer under certain simple conditions involving stationary liquids and rising bubbles. It was also found by analysis that the mass transfer to bubbles associated with turbulent liquid in a pipe could vary by as much as a factor of 6 de- pending upon the characteristics assigned to the bubble-salt interface. An experiment is under way to evaluate this effect; at present a conservative value of 17 X 10~ cm/sec at the lower end of the range for the mass transfer coefficient under turbulent conditions is being used in the calcula- tions to estimate the xenon poisoning in the MSBR. The surface area of a given volume of en- trained bubbles varies as the inverse of the spherical diameter and therefore the void volume needed in the circulating fuel can be kept small if the bubbles are small. Methods of injecting the bubbles into the salt are under study and one in particular appears promising.’ In this latter 186 NUCLEAR APPLICATIONS & TECHNOLOGY GRAPHITE AND XENON BEHAVIOR method, the gas is injected into the throat of an inverted venturi which is formed in the annulus between a teardrop shape (installed on the pipe axis) and the pipe wall. When tested with water, bubbles of ~0.5 mm were produced, and we ex- pect similar results when used with molten salt. The bubbles in the salt are allowed to recirculate through the system several times before removal. The solubility of xenon in molten salt is such that with helium-to-salt volume fractions as small as 7%, the bubbles can be recirculated ~10 times be- fore the xenon concentration in the bubbles gets high enough to significantly affect xenon poisoning. The final step in the xenon processing of the salt is the removal of the bubbles, and this can be done with a centrifugal gas separator of a type which was developed for another circulating fuel reactor.”® Kedl has proposed'’’*? a model for de- scribing the migration of the noble gases in a molten-salt reactor and for estimating the poison- ing resulting from this migration. This model was evaluated in a summary experiment using a gas containing a “’Kr tracer on the MSRE during prenuclear operation; calculated results were in good agreement with experiment. After the MSRE had been operating at power, the calculated re- sults for **Xe poison were not in complete agree- ment with experiment results; in general the poisoning was less than expected for the case where the circulating bubble fraction was very small and about as expected for the larger bubble fractions. Studies are under way to determine the source of the difference. Figure 9 summarizes the predicted effect on the poison fraction of varying the volume flow of helium into the salt stream of the MSBR for two different graphites in the core. Poison fraction is the neutron absorptions in Xe relative to fissile fuel absorptions. The values of the physi- cal constants used are considered the most likely 135 > [MOLTEN-SALT BREEDER REACTOR- ., \SINGLE FLUID 1000 Mw(e) x4 MASS TRANSFER COEFFICIENT T0 = BUBBLES, 17 X 10~® cm/sec = 3 Xe DIFFUSION COEFFICIENT IN S \ GRAPHITE, AS SHOWN e N FRACTION OF BUBBLES REMOVED - 2 S PER CYCLE, 10% 3 \\ BUBBLE SIZE, 0.5 mm AVERAGE S 1 = D >0.25 x 10~% cm?/sec P — — 0 D = 0.25 x 10"® cm?/sec o 1 2 3 4 5 6 7 8 9 10 He INJECTION RATE (scfm) Fig. 9. Effect of helium injection rate on 13Xe poison fraction, VOL. 8 FEBRUARY 1970 for the MSBR. The upper curve is for an uncoated base graphite where the main resistance to xenon transfer is in the fluid flow boundary and not in the graphite. The lower curve represents the ad- vantage gained by surface impregnation of the base graphite to a xenon permeability of ~0.25 X 10~ cm’/sec. One important difference between the two curves is in the volume of helium which must be handled to get the desired results. The conditions of the lower curve are such that only low helium gas flows are needed to reach the target of 0.5% Xe poison fraction. However, it should be noted that some helium processing is required even if a good coating is obtained. OFF-GAS SYSTEM The removal of **°Xe from the circulating fuel by sparging with helium will also remove many other fission products including the other noble gases together with some of the relatively noble metal fission products. Altogether these repre- sent a significant radiation and heat source. It is the purpose of the off-gas removal and disposal system to safely collect and store these materials to permit recycling the helium. The present ver- sion of the off-gas system removes the helium and certain fission products from the circulating salt stream, and de-entrains any salt which may be carried along; separates the noble gas from the helium, and holds the '**Xe outside the reactor for a total of 48 h, and then reinjects a Scott and Eatherly GRAPHITE AND XENON BEHAVIOR major portion of the gas stream into the fuel salt for recycle. The remainder of the stream is fur- ther processed for the removal of the longer- lived fission products so that the clean helium can be used in the helium purge stream (i.e., along the fuel pump shaft). This system is described below. Figure 10 is a flow diagram which illustrates the considerations involved in the off-gas system design. Ten percent of the total salt flow from the primary salt pump is bypassed through the 133Xe removal system, which includes the gas separator and the bubble generator. About 9 scfm of helium with fission products is removed from the salt as it passes through the bubble separator. This gas, together with some salt, goes to the entrainment separator where the salt is removed and returned to the primary circulating system through the pump expansion tank. The remaining gas, together with ~2 scim of purge gas from the pump expansion tank, then goes to the particle trap and 1-h decay tank. The pipe lines from the bubble separator to and including the entrainment separator are cooled by entrained primary salt; in addition a secondary coolant salt flowing in an annulus around the section of gas line from the entrainment separator to the particle trap will remove the heat released to the pipe wall. At the particle trap and 1-h decay tank the entrained fission-product particles are removed and allow the short-lived xenon and Kkrypton isotopes to decay for 1h before entering the absorption charcoal beds where the xenon is held up for 47 h. 2 scfm_ CLEAN PURGE He PRIMARY SALT PUMP 11 scfm_ | ENTRAINMENT SEPARATOR He SUPPLY 18 MW ACCUMULATOR 3.2 MW, - 1 | PRIMARY 'finguée | - SALT N SEPARATOR CHARCOAL | REACTOR BUBBLE BED l, | VESSEL { GENERATOR | COMPRESSOR l 9 scfm 0.27 | 85 vy 90 day (Y1 TRI'%UQND HOLDUP ' REMOVAL CHARCOAL SED SYSTEM Fig. 10. Molten-Salt Breeder Reactor Off-Gas System. NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 1970 | 187 Scott and Eatherly The decay of the noble gases in the 1-h volume delay tank reduces the heat load on the head end of the charcoal bed by a factor of 6, making it easier to keep the temperature of the charcoal low enough to maintain an acceptable adsorption ef- ficiency. In the decay tank the fission-product particles will be removed from the gas stream by washing with a liquid coolant which will also cool the gas and all surfaces of the tank walls. This coolant stream will be pumped through a heat exchanger where the heat will be transferred to a closed steam system which in turn dumps its heat to a natural-draft air-cooled condenser. About 18 MW of energy is released in the decay tank. One of the primary objectives in the design of this system is to make the heat-rejection sys- tem independent of external power needs such that a power failure will not jeopardize the system. The coolant stream will be processed to keep the concentration of fission products below the satu- ration level to prevent deposition on the walls. The gas goes from the decay tank to the char- coal adsorption beds where the xenon is held up for 47 h and the krypton is held for 4 h. This holdup is by adsorption of these heavy gases on the charcoal which does not delay the helium carrier gas. The total delay of 48 h in the two holdup sections is sufficient to reduce the “*°Xe concentration returning to the circulating salt system to 23% of its value when it left the salt. About 3.2 MW are released in this adsorption holdup system. After leaving the charcoal bed the flow splits into two streams. One system of ~2 scfm goes to the low-flow charcoal beds where the xenon is held for 90 days, after Wthh most of the activity has decayed. However, ®*Kr (10. 4-year half-life) and tritium (12-year half life) remain in the gas stream and are removed by appropriate collector beds using hot titanium beds for tritium removal and molecular sieves for °°Kr removal. The resulting clean helium is then compressed into the accumulator for reuse in the cover-gas sys- tem and the pump shaft purge. The other stream of ~9 scfm is simply com- pressed and returned to the bubble generator for recycle to the "°’Xe removal system. SUMMARY AND CONCLUSIONS The previous sections indicate that although radiation damage to the graphite and “**Xe poison- ing of the core introduce design complexities into the reactor plant, accommodation of these factors requires neither significant new technology nor excessive cost penalties. We specifically con- clude the following. 188 NUCLEAR APPLICATIONS & TECHNOLOGY GRAPHITE AND XENON BEHAVIOR 1. Radiation damage in graphite will limit core life, but existent materials are adequate to provide a core life of the order of four to five years without excessively penalizing reactor per- formance. Although a specific grade of graphite that fulfills all our requirements is not currently available commercially, only minor extensions of existing technology, including the experience factor, are required to produce such a material. 2. Under the conditions existing in the central region of the core, stresses induced internally in the graphite are small and do not limit the per- missible graphite exposure based on the present design. 3. Although not specifically discussed here, we have had suff1c1ent experience with pyrolytlc carbon impregnation® of graphite to feel confident the process can be developed for use with MSBR graphite. The times and process conditions used to impregnate small graphite samples are practi- cal for scale-up, and in at least one case the per- meability of the graphite was not significantly affected by neutron irradiation. 4. Xenon-135 can be removed successfully from the circulating salt by sparging with helium bubbles. However, several parameters such as salt-to-gas mass transfer coefficients are only approximately known. Nevertheless, it appears that the parameter values are in a range where the sparging will be effective. 5. Helium bubble insertion and extraction have been demonstrated successfully in an air/water system. Although previous experience with mol- ten-salt systems has confirmed the ability to transform fluid flow results on water to a molten- salt system, it remains to be demonstrated that the helium-bubble/salt system performance can be predicted using results from the air/water system. 6. An off-gas system has been conceptually designed which permits removal of the fission products from the helium, recirculation of the cleaned helium, and control of the fission-product heat. Such a system appears practical, based on the capabilities of present technology. ACKNOWLEDGMENTS This research was sponsored by the U.S. Atomic Energy Commission under contract with the Union Carbide Corporation. Oak Ridge National Laboratory is operated by the Union Carbide Corporation Nuclear Di- vision for the U.S. Atomic Energy Commission. VOL. 8 FEBRUARY 1970 REFERENCES 1. H. S. CARSLAW and J. C. JAEGER, Conduction of Heat in Solids, 2nd ed., Oxford, Clarendon Press (1959). 2. W. H. McADAMS, Heat Tvransmission, 3rd ed., McGraw-Hill, New York (1954). 3. E. S. BETTIS and R. C. ROBERTSON, ‘‘The Design and Performance Features of a Single-Fluid Molten-Salt Breeder Reactor,”’ Nucl. Appl. Tech., 8, 190 (1970), 4, A, M. PERRY, ‘‘Reactor Physics and Fuel Cycle Analyses,’’ Nucl. Appl. Tech., 8, 208 (1970). 5. H. C. McCOY, W. H. COOCK, R. E. GEHLBACK, J. R. WEIR, C. R. KENNEDY, C. E. SESSIONS, R. L. BEATTY, A. P. LITMAN, and J. W, KOGER, ‘*Materials for Molten Salt Reactors,”’ Nucl. Appl. Tech., 8, 156 (1970). 6. C. B. BIGHAM, A. OKAZAKI, and W. H. WALKER, ““The Direct Yield of Xe-135 in the Fission of U-233, U-235, Pu-239, and Pu-241, Tvans. Am. Nucl. Soc., 8, 11 (1965). NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 Scott and Eatherly GRAPHITE AND XENON BEHAVIOR 7. G. M. WATSON, R. B. EVANS, III, W. R. GRIMES, and N. V. SMITH, ‘‘Solubility of Noble Gases in Molten Fluorides,”’ J. Chem. Eng. Data, 1, 2, 285 (1962). 8. F. N. PEEBLES, ‘“Removal of Xenon-135 from Circulating Fuel Salt of the MSBR by Mass Transfer to Helium Bubbles,”” ORNL-TM-2245, Oak Ridge National Laboratory (1968). 9. R. J. KEDL, ‘‘Bubble Generator,”” MSR Program Semiannual Progress Report, February 28, 1969, ORNL- 4396, Oak Ridge National Laboratory (1969). 10. R. H. CHAPMAN, ‘‘HRE-2 Design Manual,”” ORNL- TM-348, p. 112, Oak Ridge National Laboratory (1964). 11. R. J. KEDL and A. HOUTZEEL, ‘‘Development of a Model for Computing ¥Xe Migration in the MSRE Graphite,”’ ORNL-TM-4069, Oak Ridge National Labora- tory (1967). 12. R. J. KEDL, “A Model for Computing the Migration of Very Short-Lived Noble Gases into MSRE Graphite,”’ ORNL-TM-1810, Oak Ridge National Laboratory (1967). FEBRUARY 1970 189