Alvin Weinberg was one of the pioneers of nuclear energy. He was a participant on the Manhattan Project, he holds the patent on the light-water reactor, and he was the director of Oak Ridge National Lab from 1950 to 1970. He gave his rationale for the development of a “molten-salt breeder reactor” in the opening pages of ORNL-TM-1851 in 1967.
WHY DEVELOP MOLTEN-SALT BREEDERS?
Nuclear power, based on light-water-moderated converter reactors, seems to be an assured commercial success. This circumstance has placed upon the Atomic Energy Commission the burden of forestalling any serious rise in the cost of nuclear power once our country has been fully committed to this source of energy. It is for this reason that the development of an economical breeder, at one time viewed as a long-range goal, has emerged as the central task of the atomic energy enterprise. Moreover, as our country commits itself more and more heavily to nuclear power, the stake in developing the breeder rises—breeder development simply must not fail. All plausible paths to a successful breeder must therefore be examined carefully.
To be successful a breeder must meet three requirements. First, the breeder must be technically feasible. Second, the cost of power from the breeder must be low; and third, the breeder should utilize fuel so efficiently that a full-fledged-energy economy based on the breeder could be established without using high-cost ores. The molten-salt breeder appears to meet these criteria as well as, and in some respects better than, any other reactor system. Moreover, since the technology of molten-salt breeders hardly overlaps the technology of the solid-fueled fast reactor, its development provides the world with an alternate path to long-term cheap nuclear energy that is not affected by any obstacles that may crop up in the development of the fast breeder.
The molten-salt breeder, though seeming to be a by-way in reactor development, in fact represents the culmination of more than 17 years of research and development. The incentive to develop a reactor based on fluid fuels has been strong ever since the early days of the Metallurgical Laboratory. In 1958 the most prominent fluid-fuel projects were the liquid bismuth reactor, the aqueous homogeneous reactor, and the molten-salt reactor. In 1959 the AEC assembled a task force to evaluate the three concepts. The principal conclusion of their report was that the “molten-salt reactor has the highest probability of achieving technical feasibility.”
This verdict of the 1959 task force appears to be confirmed by the operation of the Molten-Salt Reactor Experiment. To those who have followed the molten-salt project closely, this success is hardly surprising. The essential technical feasibility of the molten-salt system is based on certain thermodynamic realities first pointed out by the late R.C. Briant, who directed the ANP project at ORNL. Briant pointed out that molten fluorides are thermodynamically stable against reduction by nickel-based structural materials; that, being ionic, they should suffer no radiation damage in the liquid state; and that, having low vapor pressure and being relatively inert in contact with air, reactors based on them should be safe. The experience at ORNL with molten salts during the intervening years has confirmed Briant’s chemical intuition. Though some technical uncertainties remain, particularly those connected with the graphite moderator, the path to a successful molten-salt breeder appears to be well defined.
We estimate that a 1000 MWe molten-salt breeder should cost $115 per kilowatt (electric) and that the fuel cycle cost ought to be in the range of 0.3 to 0.4 mill/kWh. The overall cost of power from a privately owned, 1000-MWe Molten-Salt Breeder Reactor should come to around 2.6 mills/kWh. In contrast to the fast-breeder, the extremely low cost of the MSBR fuel cycle hardly depends upon sale of byproduct fissile material. Rather, it depends upon certain advances in the chemical processing of molten fluoride salts that have been demonstrated either in pilot plants or laboratories: fluoride volatility to recover uranium, vacuum distillation to rid the salt of fission products, and for highest performance, but with somewhat less assurance, removal of protactinium by liquid-liquid extraction or absorption.
The molten-salt breeder, operating in the thermal Th-233U cycle, is characterized by a low breeding ratio: the maximum breeding ratio consistent with low fuel-cycle costs is estimated to be about 1.07. This low breeding ratio is compensated by the low specific inventory* of the MSBR. Whereas the specific inventory of the fast reactor ranges between 2.5 to 5 kg/MWe the specific inventory of the molten-salt breeder ranges between 0.4 to 1.0 kg/MWe. The estimated fuel doubling time for the MSBR therefore falls in the range of 8 to 50 years. This is comparable to estimates of doubling times of 7 to 30 years given in fast-breeder reactor design studies.
From the point of view of long-term conservation of resources, low specific inventory in itself confers an advantage upon the thermal breeder. If the amount of nuclear power grows linearly, the doubling time and the specific inventory enter symmetrically in determining the maximum amount of raw material that must be mined in order to inventory the whole nuclear system. Thus, low specific inventory is an essential criterion of merit for a breeder, and the detailed comparisons in the next section show that a good thermal breeder with low specific inventory could, in spite of its low breeding gain, make better use of our nuclear resources than a good fast breeder with high specific inventory and high breeding gain.
The molten salt approach to a breeder promises to satisfy the three criteria of technical feasibility, very low power cost, and good fuel utilization. Its development as a uniquely promising competitor to the fast breeder is, we believe, in the national interest.
It is our purpose in the remainder of this report to outline the current status of the technology, and to estimate what is required to develop and demonstrate the technology for a full-scale thermal breeder based on molten fluorides.
Hydrogen can be produced at high efficiency (~50%) from nuclear reactors that can produce high-temperature heat, such as the liquid-fluoride thorium reactor. Typically the hydrogen is generated from water using thermochemical processes and catalysts such as sulfur and iodine. With water as the feedstock, no CO2 or greenhouse gases are released during the hydrogen generation.
My friend Charles Forsberg of the Oak Ridge National Lab sent me a paper he recently wrote on the uses of nuclear-generated hydrogen. I was especially interested in the potential uses of nuclear hydrogen to generate liquid hydrocarbon fuels from carbon feedstocks such as biomass and coal.
An excerpt from his paper reads:
if economic hydrogen is available from non-greenhouse-emitting sources and the energy for the fuel processing does not release greenhouse gases to the atmosphere, the atmospheric carbon CO2 emissions from liquid-fuel production per vehicle mile (unit of liquid fuel) can be lower than that available today from light crude oil. With nuclear hydrogen, this conversion process can become the following:
Carbon-based feedstock + Water + Nuclear energy ? Liquid fuels (2)
When nuclear energy is used (reaction 2), no CO2 is released from the fuel production process. All the carbon is incorporated into the fuel, and the carbon in the feedstock is not used as an energy source in the liquid-fuel production process. Carbon dioxide is thus released only from the burning of the liquid fuels.
Enjoy the paper!
On a number of occasions in this weblog I have talked about the central importance of the temperature coefficient of reactivity in the safety of nuclear reactors, because the temperature coefficient governs how a reactor responds to changes, transients, accidents, etc. It must ALWAYS be negative, and strongly negative temperature coefficients are even better.
While reviewing some of the old ORNL documents I was able to compare and contrast the difference in temperature coefficients between the one-fluid design and the two-fluid design. The information and relevant sections are reproduced here, and the parent documents are also available on this site.
From pages 68-70 of ORNL-4528: Two-Fluid Molten-Salt Breeder Reactor Design Study (PDF, 6.9M)
6.5 Temperature Coefficients of Reactivity
In analyzing power transients in the two-fluid MSBR, one must be able to determine the reactivity effects of temperature changes in the fuel salt, the fertile salt, and the graphite moderator. Since the fuel is also the coolant and since only small fractions of the total heat are generated in the fertile salt and in the moderator, one expects very much smaller temperature changes in the latter components than in the fuel during a power transient. Expansion of the fuel salt, which removes fuel from the active core, is thus the principal inherent mechanism for compensating any reactivity additions.
We accordingly calculated the magnitudes of the temperature coefficients of reactivity separately for the fuel salt, the fertile salt, and the graphite over the range of temperatures from 800 to 1000K. The results of these calculations, as shown in Fig. 6.10a, illustrate the change in multiplication factor vs moderator temperature (with ?k arbitrarily set equal to zero at 900K). Similar curves of ?k vs temperature for fuel and fertile salts are shown in Figs. 6.10b and 6.10c, and the combined effects are shown in Fig. 6.10d. All these curves are nearly linear, the slopes being the temperature coefficients of reactivity. The magnitudes of the coefficients at 900K are shown in Table 6.8.
The moderator coefficient comes almost entirely from changes in the spectrum-averaged cross sections. It is particularly worthy of note that the moderator coefficient appears to be quite insensitive to uncertainties in the energy dependence of the 233U cross-sections in the energy range below 1 ev. This is to say that reasonable choices of cross-sections based on available experimental data yield essentially the same coefficient.
The fertile salt reactivity coefficient comprises a strong positive component due to salt expansion (and hence reduction in the number of fertile atoms per unit core volume) and an appreciable negative component due to temperature dependence of the effective resonance absorption cross sections, so that the overall coefficient, though positive, is less than half as large as that due to salt expansion alone.
The fuel salt coefficient is due mainly to expansion of the salt, which of course reduces the average density of fuel in the core. Even if all core components were to undergo equal temperature changes, the fuel salt coefficient would dominate. In transients in which the fuel temperature change is far larger than that of the other components, the fuel coefficient is even more controlling.
Now from pages 63-64 of ORNL-4548: Molten-Salt Reactor Program: Semiannual Progress Report for Period Ending February 28, 1970 (PDF, 57.0M)
6.13 Reactivity Coefficients
A number of isothermal temperature coefficients of reactivity were calculated for the single-fluid MSBR, using the reference reactor geometry shown in a previous progress report. These calculations were performed with a detailed two-dimensional representation of the reactor in R-Z geometry, using the diffusion code CITATIONs with nine neutron energy groups. Both forward and adjoint fluxes were calculated, and the effects of various changes in microscopic cross sections or in material densities were calculated by first-order perturbation theory. The cross sections themselves were obtained from a series of calculations, using the code XSDRN in which group-average cross sections were calculated for each major region of the reactor for each of three different temperatures (800, 900, and 1000K) and for various combinations of material densities. In this way the effects of temperature-dependent changes in microscopic cross sections can be calculated separately from those of temperature-dependent changes in density.
The calculated reactivity coefficients are summarized in Table 6.3. The Doppler coefficient is primarily that of thorium. The graphite thermal base coefficient and the salt thermal base coefficient, that is, the effects of microscopic cross-section changes caused by changing the temperatures of the graphite and the salt, respectively, are positive because of the competition between thermal captures in fuel, which decrease less rapidly than those of a l/v absorber, and thermal captures in thorium, which decrease nearly as l/v, with increasing temperature. The salt density component represents all effects of salt expansion including the decreasing salt density.
The graphite density component includes both changing graphite density and displacement of graphite surfaces. In calculating the displacements it was assumed that the graphite-vessel interface did not move, that is, that the vessel temperature did not change. For short-term reactivity effects, this is the most reasonable assumption, since inlet salt bathes the vessel’s inner face. These dimensional changes in the graphite without a concomitant expansion of the vessel produce a significant change in the thickness of the salt annulus between the core and the reflector. The reactivity effect of this change is not readily calculated by perturbation theory and was therefore obtained by comparison of two conventional criticality calculations with different thicknesses of the salt annulus and with appropriately differing core density. In any case, it should be noted that the graphite density coefficient is a small and essentially negligible component.
From Table 6.3 it is seen that the total core coefficient is negative. But more important, the total salt coefficient, which is prompt and largely controls the fast transient response of the system, is a relatively large negative coefficient and affords adequate reactor stability and controllability.
The salt density coefficient is particularly important with regard to bubbles in the core salt. It is expected that the salt will contain about 1% helium bubbles. Under certain circumstances the bubbles might expand or collapse without change in core temperature and hence without invoking the total sa
lt temperature coeficient. Since the salt density component is positive, bubble expansion would produce a positive reactivity effect. Using a salt expansion coefficient ?V/V = 2.1 x 10-4/°C, an increase in core bubble fraction from, say, 0.01 to 0.02 would yield a reactivity change of ?k/k = +0.00039. This is approximately one-fourth the worth of the delayed neutrons in the core. Analogously, complete collapse of a 0.01 bubble fraction would yield a reactivity change of ?k/k = -0.00039.
Finally, the fuel concentration coefficient, (?k/k)/(?n/n), where n is atomic density, was calculated to be 0.42 for 233U and 0.027 for 235U. The large difference between these two numbers is primarily a result of the substantial difference in concentrations (i.e., n23 ? 11 x n25), so that a given fractional increase in 235U concentration produces a far smaller reactivity effect than does the same fractional increase in 233U concentration.
As can be seen from the data, the isothermal temperature coefficient for the two-fluid reactor is roughly five times as strong as the isothermal temperature coefficient for the one-fluid reactor (-4.34 vs -0.87). The isothermal temperature coefficient refers to the situation where a temperature change has made its way evenly throughout the reactor. Since most temperature disturbances originate in the fuel, the fuel temperature coefficient is even more important, and in this respect the 2-fluid is over twice as strong as the 1-fluid (-8.05 vs -3.22).
There is even recent work that casts doubt on whether a one-fluid LFR has a negative temperature coefficient at all!
These strong differences between the magnitudes of the temperature coefficients show how the two-fluid LFR design has the potential for greater safety and margin than the one-fluid design later favored by ORNL.
Several people have pointed me to this recent piece of news:
Unfortunately, I’ve never heard of this company (Northamerican Energy Group) before. Their website seems to indicate that they are involved in oil and gas development. As far as their plans to use thorium, the article states:
As part of this agreement Bayport has also identified, and has options on, two potential sites for constructing new Thorium generating plants in Idaho, and Utah, and according to Bayport has received favorable Senatorial support in those efforts.
In addition, it is estimated that 225 of the 444 commercial nuclear power plants in operation worldwide are suitable candidates for conversion to Thorium/uranium fuel and those possibilities to build new, or convert existing, power plants in foreign countries such as Africa, South America, China and other Asian countries will also be explored as part of potential business opportunities.
This statement leads me to think that they have a goal in mind similar to that of Thorium Power, Inc., which itself was started by Alvin Radkowsky, who helped build the thorium core conversion for the Shippingport reactor. In 1978, Shippingport was operated as a “light-water breeder reactor” using thorium and uranium-233. Unfortunately, as WASH-1097 has pointed out, the benefits for using thorium in light-water reactors are not nearly as pronounced as the benefits of using thorium in liquid-fluoride reactors.
Nevertheless, I wish these folks the best.
We had another excellent meeting of our “Ohio thorium” group again at Ohio State University on Tuesday, August 8th, 2006.
Early that morning, Ray Beach, Al Juhasz, and I drove from Cleveland to Columbus. Chuck Alexander and Eugenio Villeseca from Cleveland State were on the road doing the same thing.
We arrived at OSU and met in the reactor building which was fairly close to campus. I had never been that close to an actual operating nuclear reactor before, which was very exciting. We were greeted by Dr. Tom Blue of OSU and some of his colleagues, including Dr. Tunc Aldemir, Dr. Richard Denning, and Dr. Richard Christensen. Several of Dr. Blue’s students joined us as well.
Ray Beach first discussed possible funding opportunities that may open up through NASA’s JANUS program, and there was great interest among all the participants in how to configure the research plan to support this goal.
I then spoke on thorium and thorium energy cycles in general, and then specifically on the liquid-fluoride reactor as an optimal thorium burner. One of the slides that I showed that caught the interest of the group showed the layout of the old MSRE reactor building. They saw how the reactor vessel drained into several “drain” tanks, and I told them stories that I had heard from ORNL personnel how they would drain the reactor on a Friday and then come in on a Monday, thaw it out, and start it up again. Such a capability would probably be particularly attractive in a university research reactor considering their limited budgets and personnel.
Dr. Greg Washington then joined us and Dr. Blue asked that we briefly rehearse our earlier talks, which we tried to do. Dr. Washington asked a number of good questions and seemed particularly interested in the possibility of building a small LFR as a research reactor at OSU.
We then took a tour of the OSU reactor. It is a pool-type reactor built in a large concrete vessel. The concrete is several feet thick at the base and is penetrated by a number of “beam ports” to enable experiments to be exposed to the internals of the reactor. Ascending a staircase, we were able to look down into the reactor from the top of the pool. The active core itself didn’t look to be much bigger than an egg crate. There were several control rod drives that penetrated the active core and reached up all the way to the top of the pool. The reactor was shut down at the time and the operators there explained how they would maneuver sampling systems over to areas near the core for testing. We also got to see pictures of the Cerenkov glow of the reactor when it is in operation.
After leaving the top of the pool, we were shown the small control room where the reactor was operated. I was very impressed with the built-in sequencing of the reactor operations, and how an inadvertent operation would trigger a scram and shutdown. I had thought how beneficial it would be to students to gain experience starting and operating an actual reactor.
After lunch, we heard talks from Dr. Al Juhasz on power conversion systems and we also heard from several of the OSU faculty on their current research.
I found a document (that can be purchased or read online) that talks about removing U-233 from the salts of the Molten-Salt Reactor Experiment, a liquid-fluoride reactor that was built at Oak Ridge National Lab and operated from 1965-1969.
I was aware of the issues that they had faced related to the disposition of the MSRE based on conversations I had had with some “old hands” at ORNL. Basically, it came down to the fact that when they shut down the MSRE in 1969, they thought they might be restarting it again at a future date, so they didn’t go through the full uranium decontamination process that they would have done for a complete shutdown. In that process, they would have fluorinated the salt to remove uranium as a gaseous hexafluoride, thus removing all the fissile material from the MSRE salt. They had already done this process previously when they removed the original uranium fuel (a mixture of U-238 and U-235) and replaced it with U-233 for experiments on that fuel.
But they didn’t fluorinate the salt when they shut down the MSRE. It just sat in the drain tanks. While the salt was liquid, any free fluorine that formed from radiolysis was reabsorbed into the salt as fluorides. But when the salt froze and fell below 150 C in temperature, then radiolysis led to the evolution of fluorine gas. That fluorine gas, in turn, acted like its own little fluorinator, liberating uranium as a gaseous hexafluoride. That UF6 then drifted through the pipes over the ensuing decades and became a concern–one that was not easily fixed by simply melting and fluorinating the original salt.
So things got complicated, and they’ve spent a lot of time remediating MSRE. A problem that would have been fixed it they had simply fluorinated the fuel at shutdown. But hindsight is always 20/20.
I’ve completed the conversion of WASH-1097 to a full-text PDF rather than just scanned page images. Here is the file.
WASH-1097: The Use of Thorium in Nuclear Power Reactors, PDF, 1.9 MB.