Simple Molten Salt Reactors: A Time for Courageous Impatience (Opinion Editorial)
Preliminary Draft for Discussion Only
Center for Tankship Excellence, USA
After World War II, there was a big debate between the Argonne Lab led by Enrico Fermi and the Oak Ridge Lab led by Alvin Weinberg on how best to use nuclear power in peace. Both men were thoughtful geniuses. At the time, both thought that the world’s supply of uranium was very limited, and both knew that only 0.7% of this uranium, the isotope U235 was fissile, that is, could be made to fission. This meant that for every ton of U235 produced about 140 tons of uranium were required; and, in order to be useful in a reactor, that uranium had to be put through an expensive enrichment process in which the U235 was separated from the much more common, but not fissile isotope U238.
In the early 1950’s, the only non-experimental power reactor was the Navy’s pressurized water reactor, which had been invented by Weinberg, and developed in a crash program for submarine propulsion. This reactor used enriched U235 in solid form. The working fluid was water at very high pressure (160 bar) but at a rather low temperature (330C).1 Both Fermi and Weinberg believed that, if we tried to use this system for civilian power, we would very quickly run out of uranium.
Fermi argued for sticking with solid fuel based on U235, and the enrichment technology developed during the war. His solution to the fuel problem was to bombard non-fissile U238, which made up more than 99% of naturally occurring uranium with high energy neutrons in the reactors. This converts some of the U238 to plutonium which then can be fissioned. This became known as the fast breeder reactor. Most of the designs based on this concept use a liquid metal — often sodium — as the coolant.
Weinberg argued for a completely different approach. His idea was a liquid fuel reactor based on converting thorium to the fissile uranium isotope, U233. Thorium is 500 times more abundant than U235, much more easily mined, and requires no enrichment. The reactor is made up of a core of molten salt in which the U233 is dissolved, surrounded by a blanket, also a molten salt in which the thorium is dissolved. The blanket gets bombarded with some of the core neutrons converting the thorium to U233. Both fluids are continuously circulated. The core salt is run thru a heat exchanger, some processing to remove fission products and add some new U233 from the blanket. The blanket fluid is processed to remove the U233 which is sent to the core, and replaced with new thorium.
It turned out that there was far more uranium on the planet than Fermi or Weinberg thought. The industry fastened on the pressurized water reactor (PWR) that the Navy had developed. It was the quickest way to deploy civilian reactors. The PWR took advantage of the very expensive and difficult enrichment process developed for the bomb. The same companies who built the Navy’s reactors could use almost exactly the same skills and knowledge to build the civilian reactors.2 And the PWR had an attractive business model. Once you sold a reactor, the customer had to come to you for the very specialized fuel rods for the life of the reactor. In some cases, companies were willing to take a loss on the construction in order to obtain the fuel element cash flow.
Weinberg kept the liquid fuel concept alive at ORNL including building a 8 MW test reactor that ran successfully for four years, 1965 to 1969. But in 1972, Weinberg was fired at least in part because of his complaints about PWR safety The decision was made to focus all the nation’s reactor research effort on the fast breeder. In 1976, the Nixon administration shut down the molten salt reactor program. A few years later the fast breeder program, which was experiencing skyrocketing costs was shut down as well.
2 Comparison of Molten Salt with Pressurized Water
Forgetting about thorium for the moment, why did Weinberg, the man who invented the Pressurized Water Reactor, think that the Molten Salt Reactor (MSR) was a far better technology? The reasons are astonishingly obvious:
Excellent control characteristics
A molten salt reactor has a strongly negative feedback. As the core heats up, the salt expands decreasing reactivity and vice versa. The feedback is so strong the reactor is auto load following. Extract more heat, salt temp goes down, power output goes up and vice versa. There is no need for control rods which can accidentally be removed. Inherent safety.
Higher thermal efficiency
A molten salt reactor operates at around 700C. This translates into thermal efficiencies in the mid to high 40’s. A PWR reactor operates at about 330C. This implies an efficiency of about 33%.3
The MSR’s operating temperature is limited only by material considerations. As materials improve, the temperature can be raised, and the thermal efficiency still further improved. At 850C, we can disassociate hydrogen from water efficiently and produce hydrogen based fuels.
Ambient pressure operation
A PWR operates at about 160 bar. High operating pressure means 9 inch thick reactor vessels and massive piping. Some of these castings can only be done by a few specialized foundries, none of which are in the USA. Worse, if we have a big piping failure, the pressurized water explodes into steam spraying radioactivity all over the place. We need a very large and strong containment structure. In the event of a loss of coolant,
this enormous structure must somehow be kept cool despite all the decay heat in the core.
A molten salt reactor operates at near ambient pressure, less than 1 bar gage. There is no need to pressurize the system since the salts have very high (1400C) boiling points at ambient pressure. This means much thinner, cheaper, easier to fabricate, and easier to inspect piping. The primary loop is kept at the
lowest pressure, so any leaks are inwards, the opposite of a PWR.
The containment structure just needs to be an air-tight radiation barrier, and just big enough to house
the reactor core and the primary loop.4 A 1000 MWe MSR can easily be put underground. Far more complete fuel burn up
Perhaps the biggest drawback of solid fuel reactors is that the fuel cannot be completely consumed. As U235 fissions, it converts some U238 to plutonium and a whole bunch of fission products. The problem is that these fission products are trapped in the solid fuel. Some of these products are strong neutron absorbers, known in the trade as poisons. The build up of these poisons will eventually shut the reactor down. So the fuel must be removed and replaced long before all the fissile material is consumed. At that point, the 35 tons of enriched uranium introduced to a gigawatt reactor will contain about 33.4 tons of U238, 0.3 ton of U235, 0.3 tons of long-lived plutonium and 1.0 ton of other fission products.
One of the most troublesome of the poisons is Xenon-135. Xe135, a gas, is an extremely strong neutron absorber with a half-life of about 9 hours. Xenon is the the reason solid fuel reactors cannot be safely restarted for several hours after a shut down. If a PWR is shut down, Xe135 builds up. If the control rods are removed before the xenon decays, the reactor burns off the excess Xe135, and reactivity increases generating an unstable situation. Failure to manage this xenon transient properly was a necessary link in the Chernobyl cause chain.
In a liquid fuel reactor, xenon is almost a non-factor. It simply bubbles out of the molten salt as it is formed, extracted into the off-gas system, and allowed to decay. The other fission products can be continuously removed allowing the fissile portion of the fuel to be almost completely consumed.
The combination of the increased thermal efficiency and the much more complete fuel burn results in a fuel requirement that is one-third that of a PWR. And there is no need to shut the reactor down to refuel.
5000 times less long-lived radioactive waste
More complete fuel burn up not only means we need less fuel; but much more importantly, it means much less radioactive waste to dispose of, and that waste has a far lower half-life. When it comes to long-lived waste, the key is the very heavy elements, the transuranics, most importantly plutonium. All the lighter fission products decay to background levels within 300 years, most within 30 years. But unlike a PWR, in a molten salt reactor, almost all the transuranics stay in the fuel until they are “burned” (i.e. fissioned). The leading design for a molten salt reactor envisions producing 30 grams of transuranic waste per GW-year. A PWR produces about 200,000 grams of transuranics per GW-year. A liquid fuel, molten salt reactor generates 5000 times less long-lived waste than a pressurized water reactor.
Inherently far, far safer
With a liquid fuel, molten salt reactor, the phrase “core meltdown” is irrelevant. The core is already melted down. One by-product of this is totally passive shutdown in the event of power failure. The core vessel is fitted with a drain plug of frozen salt cooled by a fan. If the reactor loses power, the fan stops, the plug heats up, melts and the core salt and fuel drain to underground tanks. If the core somehow overheats, ditto. If there is a big leak in the primary loop, the core is drained to the same tanks.5 See Figure 2.
The decay heat is transferred to the drain tanks which are fitted with sufficient passive cooling to keep the salt from over-heating with no operator intervention. There is no emergency cooling system to fail or screw up a la TMI.
Fission products either quickly form stable fluorides that will stay within the salt during a leak or are volatile, such as xenon, and continuously removed. There will be nil release of radioactive gases to the environment, even if all containment levels fail and the core salt somehow fails to drain to the drain tanks. The core salt is highly radioactive and would emit radiation in the event of failure to flow to the drain tanks, but that radiation would be confined to the immediate vicinity of the salt pool.
There is no water anywhere in the primary or secondary system, that could lead to steam explosions or hydrogen production. The hydrogen explosion danger, a key player at TMI and Fukushima, is non-existent. Of course, if Three Mile Island and Fukushima had been molten salt reactors, these names would have no meaning to us. In both cases, the freeze plug would have melted, and the core salt dumped into the drain tanks.6
3 The Molten Salt Reactor Experiment
The keystone of the ORNL work on liquid fuel reactors was the Molten Salt Reactor Experiment (MSRE). The MSRE was an 8 MW thermal reactor designed to test the whole liquid fuel concept.
The core, Figure 1, was a 1.4 m diam by 1.6 m high cylinder made out of a high nickel alloy called Hastelloy. Like a standard PWR, the MSRE was a thermal reactor, that is, the neutrons had to be slowed down in order for fission to occur. The material that accomplishes this is called the moderator. In the MSRE, the moderator was graphite. The core was filled with vertical graphite bars 50 mm square. Channels were machined into the faces of these bars. The core salt flowed upward thru these channels.
The core vessel was housed in a air tight, inerted concrete enclosure, Figure 2. Drain tanks were housed in a lower enclosure. The core drain line was fitted with a freeze plug, a fan cooled plug of solid salt. If the fan was turned off, or the system lost power, the core salt flowed to the drain tanks. The drain tanks were fitted with cooling systems capable of handling the decay heat of the radioactive salt. In the event of a primary loop rupture, the salt would also drain to the drain tanks. With no moderator, there was no chance of the core salt going critical in the drain tanks.
The primary loop was entirely within the air-tight containment. It exchanged heat with a secondary loop which was led outside the containment, and for the purposes of the tests cooled by radiators fitted with big fans. The secondary loop was maintained at a higher pressure than the primary, so any leak between the loops would be into the primary.
The MSRE operated for four years, and performed closely to ORNL’s calculations. The xenon and other noble gases bubbled out a bit better than expected. The uranium and plutonium and most of the fission products stayed dissolved in the salt. There was some plate out of the noble metal fission products but they caused no problems.7 The reactor was self-regulating, responding well to all sorts of upsets including a catastrophic radiator fan failure. Shut-down was as simple as turning off the power to the freeze plug fan.
The MSRE also demonstrated on-line refueling, removal of uranium from the salt, and removal of fission products.
A key concern was how the piping materials would stand up to the temperature, the salt, and the radiation. The MSRE plumbing, after 13,172 full power hours, 21,788 circulating hours in the primary loop, and 26,076 circulating hours in the secondary loop showed no signs of salt corrosion. The Hastelloy was in excellent condition visually.
There was a minor amount of chromium transfer. Chromium leaches out (is oxidized) from the Hastelloy in the high temp core, and is deposited (reduced) in the low temp portion of the loop. They found that this could be controlled by modifying the UF3/UF4 ratio, If this ratio is high enough, there’s nil free fluorine around and the salt is never oxidizing. UF3 was generated by periodically sticking a beryllium rod into the circuit, which sucked up some of the fluorine.
Lab tests showed some superficial embrittlement and microscopic cracking of the Hastelloy. This was traced to diffusion by helium and tellurium into the metal.8 They discovered that they could handle these problems by adding 1 to 2% niobium to the alloy, which became known as Hastelloy-N, and by keeping the UF3/UF4 ratio above 0.005.9
As far as ORNL was concerned, at the end of the experiment, the key remaining issue was tritium. Tritium, a radioactive gas, is produced by neutron reactions with the lithium in the core salt. Tritium can worm its way through heat exchanger metal and eventually get into the turbine working fluid loop, and escape the system. There are three possible fixes to this problem:
- Using a salt in the secondary loop which captures the tritium, and then separating out the tritium from the salt. Possible but expensive. This was ORNL’s preferred solution.
- Using a core salt that does not contain lithium. ORNL resisted this because they were focused on breeding ratio, and the non-lithium salts are not as efficient neutronically. In an environment where fuel economy is not the key issue, this is probably the way to go. Also saves money on salt.
- Capture the tritium in the helium in the Brayton gas turbine cycle, and separate the tritium from the helium. This process is now well-established in high temp gas reactors. It was not available to ORNL in the 70’s when their thinking was restricted to steam turbines.
Overall, the MSRE was an extremely successful experiment answering many questions, and raising almost no new ones. Indeed the MSRE was a remarkable technical triumph. What is even more remarkable, this triumph received no publicity.
4 The Molten Salt Thorium Breeder Reactor
One reason for the lack of publicity is that ORNL wasn’t that interested in the MSRE. They saw the MSRE as simply a step toward their real goal a thorium fueled reactor based on converting thorium to U233. This was their Holy Grail. Buoyed by the success of the MSRE, ORNL immediately moved on the thorium breeder. This was a terrible mistake, but an understandable one.
Thorium is 10 ppm of the earth’s crust. Uranium is 2.5 ppm of which only 0.7% is U235. There is 500 times as much thorium around as U235. Thorium is easily mined and requires no enrichment. High quality thorium ore can be scooped off Indian beaches. If they could pull off a thorium based reactor, they would have solved the fuel issue for a millennium, and reduced fuel costs to a few mils per kWh.
Everything looked good with one crucial exception: how to keep the core salt and the blanket salt separated while at the same time properly bombarding the thorium in the blanket with neutrons. Their first thought was to alternate core salt and blanket salt channels in the graphite moderator. Very neat, the moderator and the barrier are one and the same. The neutron calculations showed it would work fine.
The problem is that when graphite is bombarded with neutrons it first shrinks and then swells as the carbon atoms get shoved around in the lattice. Eventually, we will have a leak. This is not much of an issue if there is only one fluid in the core, but a big problem for a two fluid system. And we have a large number of problematic graphite to Hastelloy joints. The ORNL guys who were very smart and very motivated worked on this problem for ten years and never solved it.
Eventually they went to a single salt concept in which the thorium was dissolved in the core salt, along with the fissile U233. The problem with this system is that you now have to remove the fission products without removing the thorium, and thorium chemically is very similar to some of the more important fission products.10
ORNL semi-convinced themselves that they had a solution to this processing issue, but admittedly it was far more complex than the simple MSRE processing system associated with the two salt concept. It has never been tested outside the lab. At a minimum, it was going to take a lot of R&D. To an outsider, it looks very unattractive. About that time the molten salt reactor program was shut down.
5 The Denatured Molten Salt Reactor
The ORNL people had no money, but they weren’t quite ready to give up. By the 1980’s it was clear we were not going to run out of uranium for a while and issues such as proliferation resistance were becoming much more important than they had been. ORNL finally realized the country didn’t need a thorium breeder. The country only needed vastly improved safety and vastly decreased long-lived waste. The country needed a full scale MSRE.
The remains of the ORNL group went back to the drawing board and designed a reactor that required nil on-line salt processing. They called it the Denatured Molten Salt Reactor to emphasize its proliferation resistance. Denatured refers to uranium whose U235 content is too low to be useful in a bomb. A better name would have been the Simple Molten Salt Reactor (SMSR).
The concept was a 1000 MWe liquid fuel, molten salt reactor that would be fueled once in 30 years.
- The initial fuel load would be low enriched uranium and thorium. Over time the thorium would beconverted to U233 and burned.
- Other than bubbling off the xenon, there would be no salt processing, no removal of the fission products,just adding small amounts of U238 annually to keep the uranium denatured.
- A low power density yielded 30 year lifetime for the graphite moderator, but resulted in a fairly largecore (8 m in diameter, and 8 m high).
Building on the MSRE experience, the group, some of the best reactor designers in the world, crunched
the numbers and produced a 162 page report, which lays out the reactor design in some detail.11 They showed us that the concept worked. In other words, they gave us the design of a reactor which would be smaller, cheaper, far safer, and produce 5000 times less long lived waste than a PWR. A reactor that could be built with no new technological advances.
The report has been ignored.12
6 It’s all Alvin’s fault
How did we make such a bad choice? The short answer: Rickover. Admiral Rickover was the acerbic, dictatorial head of the Navy’s nuclear sub program. Once Weinberg pointed Rickover toward the PWR, the ascendancy of the PWR was almost inevitable. Rickover had originally intended to build a high temperature, sodium cooled, graphite moderated reactor. He was turned off by the PWR’s poor thermal efficiency and high pressure. But the reactor had to fit within a 28 foot wide submarine hull. Weinberg pointed out that an PWR could be shoe-horned into a smaller space even after allowing for the extra thermal output required. Besides high pressure steam piping was something the Navy had lots of experience with. They knew next to nothing about handling sodium, except that, if sodium comes into contact with water, you have an explosion. This was in the late 40’s, shortly before the molten salt reactor was invented.
Rickover with essentially unlimited funds and a maniacal drive fueled by the Cold War and the threat of extinction made the PWR, with all its problems, work.13 In 1948, he started with little more than a sketch and some back of the envelope calculations. At the time there was no such thing as a pressurized water reactor at any scale. Nobody knew how to make control rods or fuel element cladding or bearings that could handle the PWR conditions. Rickover did not build a lab scale reactor. He went straight to a full scale on-land prototype inside a mock-up of a submarine immersed in a giant swimming pool. He ordered the ship before this prototype, the first PWR ever built, even went critical. In 1954, the first nuclear powered submarine, the Nautilus, was launched. The Nautilus immediately undertook a string of high publicized exploits. Rickover had created the PWR from scratch and he did it in six years.
When Eisenhower started the Atoms for Peace Program in 1953 he wanted results now. Rickover, who was a master at manipulating congress, grabbed control of the Atoms for Peace program, over the objections of nuclear scientists, and others who knew that Rickover would not consider anything but a PWR. Rickover wrote a mocking parody praising the wonders of the other concepts but among the wonders was “not available”. In fact, at this time, some of the other concepts were much further along than the PWR was when Rickover committed to his full scale prototype.
In any event, Rickover made sure that the first US civilian reactors were PWR’s. The first couple were essentially given to the utilities. Rickover’s naval contractors were only too happy to sell their skills to the civilian side. Once they committed to the civilian PWR that became just about the sole focus of their work.
Meanwhile, Fermi and Weinberg were doing experiments at a far more leisurely pace. But Fermi’s defi- nition of leisurely was a lot more expensive than Weinberg’s. He was spending hundreds of million dollars per year on the fast breeder and that spending was spread over many congressional districts.14 And the fast breeder program was getting plenty of publicity.
Weinberg was spending 2 to 4 million dollars per year on the MSR and almost all that money was concentrated in one small town in Tennessee. The ORNL culture made matters worse. As you read the meticulous ORNL reports, you are struck by the complete lack of salesmanship. The reports focus almost entirely on problems and screw ups, potential problems, errors and possible errors in the calculations, and jobs undone. Successes and accomplishments are mentioned briefly if at all. Good engineering and very slow reading. No where in the voluminous ORNL library could I find the vision articulated in a manner understandable to outsiders.
But the thorium vision was there, and it messed them up. They spent way too much effort and time failing to solve the core/blanket barrier problem, when they should have been moving ahead with a scaled up MSRE.
In any event, almost no one outside of Oak Ridge understood the potential of the molten salt reactor. Few had even heard of it. There was zero political push behind the project. When Weinberg fell out with the AEC and the AEC’s congressional backers on reactor safety, the MSR project was doomed.
There have been sporadic, weak attempts to revive the MSR; but they have gone nowhere. Part of the reason is antipathy to nuclear power, any kind of nuclear. Part of the reason is economics. During the 80’s and 90’s, fossil fuel was cheaper than nuclear and few were worried about CO2. Part of the reason is simple ignorance. When asked about the MSR at his 2009 confirmation hearings, Secretary Chu, a Nobel Laureate physicist, replied “One significant drawback of the MSR technology is the corrosive effects of the molten salts on the structural material used in the reactor vessel and heat exchangers.” Dr. Chu may be the one scientist that President Obama really trusts; yet he is unaware that ORNL had solved the corrosion problem.
Very recently, there has been a flurry of interest in molten salt in the guise of the thorium breeder. But the proponents of the thorium breeder are selling something that does not exist and requires a major technological breakthrough which may never occur. They are making the same mistake that ORNL made 40 years ago.
7 A Time for “courageous impatience”
“Good ideas are not adopted automatically. They must be driven into practice with couragous impatience.”, Admiral Hyman Rickover. We need to start building a full scale Simple Molten Salt Reactor tomorrow. Thanks to ORNL, we are ready to go to design and engineering. We are far further along than Rickover was in 1948. The prototype plant would put the SMSR on the map, and make manifest the multi-order of magnitude improvements in long-lived waste and safety. And it will pin down the economics.
The whole design concept should be based on what we already know, just a scaled up MSRE. The operating temperature should be a conservative 700C. The design should be based on a super-critical steam turbine which uses standard modern coal plant technology. It should be installed underground, Figure 3. The plant should be sized so that the reactor vessel is road transportable. This will be about 500 MW thermal or about 200 MW electric. The plant will be much more than a demonstration. It will be a prototype for assembly line core construction. As we gain experience, we will almost certainly be able to increase the power density.
The project must have the enthusiastic support of the President, who can sell it as essentially a new form of energy, as different from current nuclear as a modern computer is from a mechanical calculating machine. See Table 1 for the talking points.
The project must be run outside the NRC licensing process, or it will be doomed from the start. Even Rickover could not have built the PWR under current regulatory procedures. The prototype plant experience could provide the basis for a completely new set of regulations tailored to the MSR.
The schedule should be tight, allowing for no research. A plant up and running in four years.
We should be working hard on the closed Brayton cycle and switch over to the smaller, more efficient gas turbine as soon as its available. The demo plant can be designed with this switch in mind, but it should not depend on it.
We should be working hard on the materials side to attempt to push future molten salt operating temps up to 850C at which point hydrogen generation becomes a realistic alternative. But there is no need to wait for this advance in materials. 700C is plenty, and far better than the PWR.
We should be working hard on the core salt/blanket salt barrier problem. If we solve this problem, the second or third generation MSR’s can be based on cleaner, more abundant thorium, and fuel costs pretty much disappear. But we don’t need this break through and we can’t wait for it.
All we need is the 5000-fold reduction in long-lived waste and the vastly improved safety of the SMSR. And we can have that in a very few years if we want.