The two-fluid MSBR is graphite-moderated and -reflected, with a 7LiF-BeF2-UF4 fuel salt circulated through the core and a 7LiF-ThF4-BeF2 blanket salt circulated through separate flow channels distributed throughout the core, as well as in a surrounding under-moderated region. The fissions raise the temperature of the fuel salt to about 1300°F and that of the blanket salt to about 1250°F. Heat is removed from the salts in shell-and-tube heat exchangers to raise the temperature of a circulating NaBF4-NaF coolant salt to about 1150°F. The coolant salt transports the heat to steam generators and reheaters to provide 3500-psia 1000°F/1000°F steam for a conventional turbine-generator.

The conceptual design was based on use of four reactors and the associated heat transfer systems in a so-called modular arrangement to supply steam to a single turbine-generator. This made it practical to consider replacement of an entire reactor vessel assembly after the core graphite received its allowable exposure to neutrons. The total fluence at which it was thought that additional graphite dimensional changes would become excessive was taken as 3x10²² neutrons/cm² (E > 50 keV), or about eight years of full-power operation.

All portions of the systems in contact with the fluoride or fluoroborate salts would be fabricated of Hastelloy N that has a small amount of titanium added to improve the resistance to radiation damage. The graphite would be a specially coated grade having low gas permeability to xenon and better resistance to radiation damage than conventional material. The two-fluid concept involves joining graphite core elements to Hastelloy N tubing using a brazing process developed at ORNL.

The reactors and associated systems would be housed in concrete cells to provide biological shielding and double containment of all radioactive materials.

Plant flowsheets and layouts were developed sufficiently during the study to give an indication of feasibility and to give a basis for cost estimates, but no optimization studies were made. Safety aspects were considered throughout the design effort, but no formal safety analysis was completed.

Fuel and blanket salts would be continuously processed in a nearby cell to remove fission products and to recover the bred product. The processing rate would correspond to removal of uranium and protactinium from the blanket on a 3-day cycle and rare-earth fission products from the core on a 60-day cycle. Since no conceptual designs for the chemical plant were completed, cost estimates could not be on a definitive basis. The tentatively estimated fuel cycle cost is about 0.5 mill/kWhr, which includes the fixed charges and operating costs for the processing equipment, the fuel inventory charge, and the credit for bred fuel. Graphite replacement costs, which are not included, would add about 0.2 mill/kWhr.

The tentatively estimated total construction cost of a 1000-MWe MSBR station, based on the early 1968 value of the dollar, is about $141 per kilowatt. The power production cost for a privately owned station, based on fixed charges of 13.7% and 80% plant factor, is about 4 mills/kWhr. The net thermal efficiency of the plant would be about 44.9%.

The off-gas, fuel processing, afterheat removal, and maintenance systems needed further investigation at the time the study was suspended, and the limited performance of the graphite undoubtedly restricts the design and imposes a maintenance penalty, but the study did not disclose any aspects which indicated that major technological discoveries would be required to design a two-fluid molten-salt reactor power station. The major concern was whether mechanical failure of graphite tubes in the reactor core would cause the effective lifetime of the core to be significantly less than the eight years imposed by the effects of irradiation on the graphite.

The basic objective of the Molten-Salt Reactor Program is to develop the technology for economical nuclear power reactors that make use of fluid fuels which are solutions of fissile and fertile materials in suitable carrier salts. A major goal is to achieve a thermal breeder reactor based on the thorium-233U fuel cycle that will produce power at low cost while conserving and extending the nation's fuel resources.

Conceptual design studies of a variety of molten-salt breeder reactors for large plants are an important part of this program. In August 1966 we published a survey report, ORNL-3996, in which we described briefly the status of molten-salt reactor technology and the designs of reactors and fuel processing facilities for 1000-MWe power stations. This survey led us to conclude that the two-fluid reactor which separates the fuel and blanket salts held the most promise for development as a breeder reactor. The modular version, consisting of four reactor modules and associated intermediate systems supplying steam to one turbine-generator, was selected for more detailed analysis.

The study of the modular design of a 1000-MWe plant was begun in the fall of 1966, and some of the results were published in the MSRP progress reports, ORNL-4037, ORNL-4119, and ORNL-4191. Much of the effort was spent on designs for the core and in exploring the effects of radiation-induced damage to graphite on the core designs. The plant layout, the cell designs, the drain tank systems, the nuclear characteristics, the maintenance, and the cost estimates were also examined in more detail than had been possible in the earlier survey.

Considerable progress had been made in these studies when, in August 1967, encouraging information obtained from research on the processing of molten-salt fuels indicated that protactinium and some fission products could. be separated from the uranium-and thorium-containing fuel salt of a one-fluid reactor by reductive extraction into liquid bismuth. At about this same time, nuclear calculations indicated that a conversion ratio greater than 1 could be achieved in a one-fluid reactor of acceptable dimensions by increasing the fuel-salt-to-graphite ratio in the outer regions of the core. The one-fluid breeder is mechanically simpler than the two-fluid breeder because it involves only one salt stream, which contains both the fissile (233U) and the fertile (thorium) constituents. Also, the one-fluid breeder is a direct descendant of the one-fluid Molten-Salt Reactor Experiment, which has operated well at Oak Ridge National Laboratory. The attractive possibility of being able to progress in a direct path from the MSRE to large thermal breeder reactors of similar design led us to set aside the studies of two-fluid breeders to examine one-fluid breeder reactors in equal detail. The studies of the one-fluid breeders were begun in September 1967 and are continuing.

Although the one-fluid breeder has the desirable features mentioned above, the fact remains that the two-fluid MSBR is inherently capable of achieving a significantly higher breeding performance. This feature alone will sustain interest in the two fluid system. It is thus important to document the progress made in the two-fluid breeder study before it was set aside. Presenting this information adequately is difficult, because several months of studies of the one-fluid reactor have changed some of our ideas about MSBR design, and new data relevant to the two-fluid reactor have continued to come from the research and development program. For example, the physical properties of the salts have a profound influence on the design, yet many of these properties are under continuous study and adjustment. Some of the new information will be mentioned briefly, but the reader should understand that this report does not fully represent current ideas and that some designs and conceptual drawings presented here would be considerably altered if they were to be reexamined on the basis of today's knowledge.

The studies upon which this report is based involved personnel from almost all the divisions of ORNL, but particularly those from the Reactor Division, Reactor Chemistry Division, Chemical Technology Division, the Metals and Ceramics Division, and the General Engineering Division. A group composed of members of these divisions, under the leadership of E. S. Bettis, provided the conceptual designs and data which are basic to the report.

Several basic considerations influenced our choice of a two-fluid MSBR concept and many of the details of the plant design. They are reviewed here to provide the reader with a better understanding of the design that evolved.

A simplified diagram of a two-fluid breeder reactor is shown in Fig. 2.1. The core of the reactor consists of an array of tubular graphite elements in the center of the reactor vessel. A molten fuel salt is recirculated through the graphite elements and through a shell-and-tube heat exchanger by means of a centrifugal pump. A molten blanket salt is similarly recirculated through the space around and between the graphite pieces in the reactor vessel and through an external heat transport circuit. Heat generated in the reactor is transferred from the fuel and blanket salts to a coolant salt in the heat exchangers. The coolant salt is recirculated through steam generators where the energy is used to convert the feedwater into superheated steam that drives a conventional turbine-generator to produce electricity.

The MSBR is a thermal breeder reactor that is intended to attain the highest breeding performance consistent with producing power at low cost. Our past studies have indicated that a good measure of the performance of a breeder system is the total quantity of fissionable material that must be mined in order to provide the fissile inventory for a large nuclear power system. This total ore requirement should be low. The terms that describe the performance vary with the assumed growth rate of the nuclear electrical industry and thee types of reactors that precede and accompany the breeders, but in the range of interest the performance of a breeder is approximately proportional to the product of the breeding gain G and the reciprocal of the square of the specific inventory, 1/S² . The "conservation coefficient" G/S² for MSBR's can be expected to be in the range of 0.02 to 0.10, where the specific inventory has units of kilograms of fissionable material per megawatt of electricity and the breeding gain is dimensionless.

A practical thermal breeder reactor can only be fueled on the thorium-233U cycle, and it has a small potential breeding gain. Typically, η for an MSBR is 2.22 neutrons produced per neutron absorbed in fissile material that is an equilibrium mixture of 233U and 235U. Absorption of one neutron in fissile material and one in fertile material leaves 0.22 of a neutron for losses to moderator, carrier salt, leakage, higher isotopes, protactinium, fission products, and structural materials and for absorption in thorium to produce the gain in 233U.

Achieving high performance in a breeder depends on keeping the parasitic absorption of neutrons and the specific inventory of fissile material low. Losses to carrier salt, moderator, and structural materials and the rate and cost of processing to keep the fission product losses low all decrease with increasing concentration of uranium in the fuel salt and increasing inventory in the reactor core. The specific inventory, however, includes the inventory in the heat transfer equipment external to the reactor vessel, in storage, and in the fuel processing plants, so that the specific inventory and the total inventory cost increase rapidly with increasing concentration of uranium in the fuel salt. The breeding gain and specific inventory must be balanced to obtain the highest breeding performance (large G/S²) that is consistent with producing power at low cost.

The favored fuel salt contains about 0.2 mole % UF4, of which about 70% is 233U and 235U, 23% is 234U, and 7% is 236U. The uranium fluoride is dissolved in a 7LiF-BeF2 (67-33 mole %) carrier salt. As shown in Table 3.1, this salt has a liquidus temperature of about 840°F and good flow and heat transfer properties at the working temperatures. It also has excellent thermal and radiation stability and, with the use of 7Li, a low cross section for the parasitic absorption of neutrons. A ThF4-7LiF-BeF2 salt (27-71-2 mole %), which melts at about 1040°F, is a good choice for the blanket salt. The physical properties of this salt are also shown in Table 3.1.

Although lithium and beryllium nuclei are good moderators for neutrons, the moderating properties of the fluoride salts are not sufficiently good, when compared with their neutron absorbing properties, to build a thermal breeder without the use of other moderator. Graphite is the best material for this purpose, because it has good moderation properties, a low neutron absorption cross section, and good structural properties at high temperature and can be used in direct contact with molten fluoride salts.

The design and performance of the reactor depend considerably on the effects of fast neutrons on the graphite. Neutron irradiation causes graphite to change dimensions and its physical properties to deteriorate. The life of the graphite is expected to be limited to some total exposure to fast neutrons and therefore to vary inversely with the maximum power density in the core. Selection of a design power density for the core must be based on a balance between the costs of fuel inventory, periodic replacement of the graphite, and other factors that reflect on the net cost of the electricity produced.

In order for the graphite to have an acceptable radiation life, we estimate that the maximum power density should not exceed about 100 kW per liter of core volume. With this limit on power density, the core of a central-station power reactor would have a volume of several hundred cubic feet. This size is too large for the core to consist of graphite bars and highly-enriched fuel salt contained in a thin metal shell and surrounded by a region of blanket salt. The critical concentration of 233U in the fuel salt would be so low that the absorptions in the carrier salt and the graphite would be excessive. Absorption of neutrons by the shell would further degrade the performance.

The concentration of 233U in the fuel salt can be raised to the desired level by dispersing blanket salt throughout the core. This is accomplished by making the graphite moderator in the form of tubular elements and flowing the fuel salt through the elements and the blanket salt around the elements. The core composition is obtained by optimizing the relative volumes of fuel salt, blanket salt, and graphite within bounds imposed by limits on the concentration of thorium in the blanket salt and by the engineering of the core.

Results of many calculations have shown that the combined neutron losses to fuel and blanket carrier salts, the graphite moderator, and higher isotopes will be near 0.11 in an optimized reactor, leaving 0.11 for other losses and the breeding gain. Leakage losses are reduced to a small amount by a thorium blanket of reasonable thickness around the core. The losses due to protactinium are kept small by keeping its concentration in the blanket salt low. This is accomplished by having a blanket of large volume at low neutron flux or by removing the protactinium from the blanket salt on a few-day cycle and allowing it to decay in the processing plant. Xenon-135 must be removed from the fuel salt on a few-second cycle, or the surfaces of the graphite elements must be sealed to greatly reduce the rate of diffusion of xenon into the pores. Most of the other fission products must be removed by processing the fuel salt on a 30- to 50-day cycle. Limiting the total of the above losses to 0.03 to 0.07 appears to be reasonable; this leaves a potential breeding gain of 0.04 to 0.08.

A reactor with a breeding gain in this range and a specific inventory of 1.5 kg/MWe or less will have good breeding performance. In order to have this low a specific inventory, the amount of 233U external to the reactor core must be kept to a minimum. The heat transfer circuit of the reactor must be closely coupled to the reactor vessel, and it must have high performance. The fissile inventory in the blanket systems must be kept small by extracting the bred 233U from the blanket salt on a few-day cycle and making it available for adding to the fuel salt to compensate for burnup. Processing the fuel and blanket salts at the reactor site is necessary to avoid inventory in transport and storage, and short cooling time is important in reducing the inventory in processing. The processes must be simple and involve few changes in the physical or chemical nature of the salts if they are to be carried out rapidly and inexpensively. Fluorination to remove the uranium as the volatile UF6 followed by vacuum distillation to separate the carrier salt from the rare-earth fission products satisfies these requirements for processing the fuel salt. Fluorination to remove the uranium or extraction of protactinium and uranium into molten bismuth can satisfy the requirements for the blanket.

With thorium blanket salt dispersed throughout the core, the breeding gain is largely independent of the size of the core, but this arrangement imposes several conditions on the design. The first of these is that graphite elements must be joined to metal-piping in the reactor vessel. A perfect separation between the fuel and blanket salts is not essential to the safety of the operation, but the leakage must not be so great as to put an excessive burden on the processing facilities. Processing considerations lead to a preference for any leakage to be blanket salt into fuel salt, and the leakage must be kept below about 1 ft³/day in a 1000-MWe plant. Such a plant would have several hundred graphite-to-metal joints. Our experience led us to choose graphite-to-metal brazing as the method for obtaining adequate leak-tightness.

The graphite elements for the core must be of a size and shape that are within the capability of manufacturers to make and inspect for reasonable cost and with good quality control. Isotropic material appears desirable and may be essential from the standpoint of irradiation effects. Thicknesses of sections must be limited so that the temperature rise due to heating in the graphite is not large. Effects of irradiation increase with temperature, and stresses increase with temperature difference, so a large rise in internal temperature could result in a large decrease in service life of the core elements. Graphite tubes 6 in. or less in diameter and with a wall 3/4 in. or less in thickness appear to fulfill all these requirements.

Neutron irradiation produces substantial changes in length of the graphite elements, and the difference in expansion of the graphite and the metal parts of the reactor vessel with temperature changes can also be large. These effects must be accommodated without overstressing the graphite. We propose to accomplish this by making the graphite elements in the form of concentric tubes connected to the reactor vessel at only one end in order to provide freedom for axial expansion and contraction. The fuel salt would flow in and out at the same end of the elements, and the connections would be to tube sheets at the bottom of the reactor vessel to allow the salt to drain completely.

Because of the irradiation effects, the graphite tubes will have to be replaced periodically. Also, one could expect an occasional failure of a graphite element or a graphite-to-metal joint from other causes. The reactor vessel and internals will be highly radioactive after a short time at high power, and with the graphite elements brazed to a tube sheet in the bottom of the reactor vessel, individual tubes could not be readily inspected or replaced. We concluded that the most practical way to renew the graphite in the core would be to replace the entire reactor vessel and its contents. Suitable provisions would be required for remotely operated tools and viewing equipment to cut, weld, and inspect joints in the piping system. Provisions for handling and disposing of spent reactor vessels would have to be included in the plant.

The high melting temperatures of the salts make it necessary to preheat the reactor equipment to high temperature before introducing the salts and to maintain the temperature when they are present. The special problems of maintenance and inspection of the reactor equipment after it has become radioactive led to our proposals to install the reactor systems in heated cells, which are comparable to large furnaces, rather than to apply heaters and insulation to the vessels and piping.

In our studies of designs for molten-salt breeder reactors, we are concerned primarily with power stations having outputs of 1000 MWe or more. The capacities of salt circulation pumps, heat exchangers, steam generators, etc., needed for such plants are greater than could reasonably be designed into single units. In the 1000-MWe MSBR design described in ORNL-3996, we chose to connect four-primary heat removal circuits to one reactor vessel, to provide one coolant and steam generator circuit for each primary heat removal circuit, and to send the steam from all the steam generators in the plant to one turbine.

Since the two-fluid breeder has a blanket of low 233U and high thorium content around the core to capture the leakage neutrons, reactors of this type can have about the same breeding performance over a wide range of size if the maximum power density in the core is held constant. These facts, together with the special problems and time required to replace a reactor vessel, led us to consider a modular design for the two-fluid MSBR in which separate, but smaller, reactor vessels would be coupled to primary heat removal circuits to provide four autonomous reactor systems delivering steam to one turbine-generator. This modular plant would be slightly larger than the integral plant, since four small reactor vessels with associated control systems would be substituted for the single larger vessel. Otherwise the equipment in the plant would be the same. The advantage would be that the plant could continue to operate at part-load while one or two modules were down for maintenance. We were sufficiently impressed by this capability to make the modular concept the basis for the design studies described in later sections of this report. No analysis was made of the optimum size for a module. We simply decided for the purposes of this study to provide four modules in our 1000-MWe plant.

All our designs for MSBR plants have fuel and blanket circulation systems that are separated from the steam system by an intermediate coolant system. If the steam system were coupled directly to the fuel salt system by means of a steam generator, any leaks in the tubes of the steam generator would result in steam or water leaking into the fuel salt. Reactions between water and fuel salt would not be violent, but corrosive hydrogen fluoride would be generated, and uranium oxide would precipitate in the salt. Also, special provisions would have to be included in the design to prevent the fuel circulation system from being raised to the high pressure of the steam system. Molten sodium, helium, and other coolants have been considered for use in the coolant system, but we prefer a molten salt. Sodium reacts with the fuel salt to generate considerable heat, precipitate uranium, and raise the melting point of the salt. Helium does not react with the salt but must be used at high pressure in order to obtain a good heat transfer coefficient in the primary heat exchanger. At best the heat transfer coefficient with gas is considerably less than can be obtained with sodium or salt and results in an undesirably high inventory of fuel salt and fissionable uranium in the reactor system. The 7LiF-BeF2 coolant salt used in the MSRE is a good coolant, but it costs about $1400 per cubic foot, and its melting point is about 840°F. We would prefer to have a less expensive cooling salt with a lower melting point. The salt NaBF4-NaF (92-8 mole %) costs only about $60 per cubic foot, melts at 725°F, and is a favored candidate for use in the coolant system.

Minimum operating temperatures for the MSBR are set by the liquidus temperatures of the salts, and the materials of construction are governed by the operating temperatures and the properties of the salts. The reactor fuel and blanket systems must be operated at temperatures above about 1000°F, and the coolant system must be operated above about 750°F. High nickel alloys have good resistance to corrosion by fluoride salts at high temperature and good creep strength to about 1300°F. Since the temperature must be high and the materials are expensive, we believe it appropriate to couple the reactor plant to a steam cycle that is representative of the best current practice. The 3500-psia, 1000°F-throttle, 1000°F-reheat cycle that is presently being specified for most new large fossil-fueled plants was selected for use in our design studies largely on this basis. The supercritical-cycle has the added advantage that the feedwater to the steam generators could be preheated to 700°F without much loss in thermal efficiency by direct injection of superheated steam into the water. This procedure may be necessary if use of feedwater at a more common temperature creates problems in the steam generators by freezing coolant salt on the tubes. (At subcritical pressures the Loeffler cycle employing a steam circulator and mixing drum probably would have to be used to attain the requisite high-temperature entering stream.)

Finally, it is important to emphasize that the designs discussed here are based largely on current technology and developments that we believe to be readily achievable. The materials, processes, and performance factors are developed sufficiently that no major inventions appear to be required to solve the technological problems.