Let’s Burn Rocks to Produce Energy



A nuclear reactor can “burn rocks” to release vast amounts energy.

This may not surprise you if you know a little about common nuclear reactors. But they can only use about 1% of their fuel before it has to be changed because it becomes so contaminated with the products of the reaction, that those products soak up too many of the neutrons from fission (atom splitting) flying around which would allow cause another atom to split, sustaining a chain reaction. The fission products “poison” the reactor fuel.

A conventional and common boiling or pressurised water reactor has to be shut down to change fuel in its core.

To understand the magnitude of the problem, imagine filling up your car with 50 litres of fuel in your car, but after having used only 1 litre of the fuel, you have to stop the car because the fuel has become too contaminated, drain the fuel tank and pay for a tank of fresh fuel. That’s low fuel efficiency.

Continuous Refuelling

Over the decades scientists and Engineers have contrived different means by which fuel in the reactor can be continuously cleansed of the fission by-products that diminish operational efficiency.

One method, devised in the late 1950’s is the pebble-bed reactor in which the reactor core is filled with spherical pebbles that are made up of a tiny core of fissile material (which has atoms that can split when struck by an energetic particle) and several layers of encapsulation for strength, and thermal stability; and a moderator to slow down the neutrons passing into and out of the pebble. All of that in a the size of a billiard ball.

Each pebble is, by itself, incapable of sustaining a chain reaction. The reactor core consists of a heap of pebbles (Kugelhaufen in German) that establishes the critical mass required for a sustained chain reaction which releases the heat to drive generating equipment. The heap is cooled by an inert gas, usually Helium, heated to temperatures around 750ºC.

Pebbles are cycled through the core,. Each is examined for damage and residual fissile quality. If it doesn’t come up to scratch, it’s replaced with a fresh one. The next step is where it gets complex and expensive, the sub-par pebble is then put into fuel reprocessing because the vast majority of the fissile material in the pebble is still good.

The pebble has to be crushed, its components separated and re-refined with the good stuff going into new pebbles and quite a lot of what remains going into long-term storage until another use can be found for it. (“Waste” is a misnomer.)

A Solution

Another method devised in the early days of the nuclear age was the molten-salt reactor, where the fuel components are kept in a solution of molten salts. The liquid form allow for a circulation of the fuel through the reactor, easy separation of gaseous fission products and chemical separation of others dissolved in the molten salts; all on a continuous or quasi-continuous basis.

The high operating temperature (over 700ºC) of the molten salts has the potential for high thermal efficiency and, as the core is already molten, precludes a melt-down from being a bad situation. Molten salts are kept in the reactor core by means of a freeze plug which is a section of pipe cooled externally so that it remains solid under nominal conditions. If the external cooling ceases or the reactor temperature rises high enough, then the plug melts, shutting down the reactor. The molten salts drain into tanks with a large surface area so that decay heat can dissipate safely to the surrounding structure and naturally convecting fluids or gases.

A reactor of that type was operated at ORNL for about a decade. Not just a plain one; but one that could be extended to breed its own fuel in the long run. One that can convert fairly ordinary “rocks” into fuel.

That had a moderator of solid graphite which introduced its own problems because it was in contact with the molten salts and the fission products could react with and diffuse into the graphite. So why have a moderator?

Everything in Moderation

The nucleus of an atom is easier to “hit” with a slow neutron than a fast one. A higher fuel efficiency is achieved in a reactor when the neutrons from a source are slowed from being fast (>1 MeV) to a thermal velocity (<0.025 eV). You simply don’t need anywhere near as much fuel to have a sustained reaction if you slow the neutrons to thermal velocity (“thermalize”).

Slowing  is done by introducing a moderator into the neutron stream. The neutron collides with the nuclei of the moderator and is either scattered (deflected) or absorbed. In each collision, the neutron loses some of its kinetic energy. It is best for the neutron to lose as much energy as possible in each collision and for it not likely to be absorbed, especially as the neutron slows.

Choosing the material and mechanism for the moderator is crucial to keeping the reactor compact, reliable, economic and productive.

Chicken and Egg

Yes. I mentioned “breed” in the previous section. The Goldilocks nuclear fission reactor. Every resource in the universe is limited. (Except stupidity; Einstein said so.) A breeder reactor produces fuel as a by-product of releasing energy for generating electricity and other processes that can use the high-quality heat available from a nuclear reactor.

A breeder reactor that produces as much fuel as it uses isn’t a perpetual motion machine. It consumes fresh “rocks”. The initial energy comes from ordinary fissile material that releases lots of spare neutrons as it splits in a chain reaction. The spare neutrons are caught by the fertile material, which absorbs the neutron. Under the right conditions, that absorption results in a transmutations and decays into another elements into something that can eventually be used as a fuel.

The favourite materials of the people at ORNL was 232-Thorium which is slightly radioactive, fertile but not fissile; and 233-Uranium which is a highly-radioactive, synthetic isotope. When 232-Th absorbs a neutron to become an unstable 233-Th, this, most of the time, decays to 233-Protactinium which after a while (27 days) decays to become 233-U. i.e. the fissile fuel.

And that is the Thorium Fuel Cycle.

Unlike fuel enrichment based on isotopes of the same element, the fluid containing mostly the fertile 232-TH and some 233-U can have the latter extracted by a chemical reaction because they are different elements with different chemical properties. It’s advantageous to separate out the 233-Pa which would linger in the solution until it decays, all the while potentially grabbing another neutron to become useless 234-U, 233-Pa being a much bigger target than 232-Th (40 barns vs 7.4).

Hitting the side of a barn

In the submicroscopic universe of nuclear physics, target sizes change depending on the speed at which the bullet moves and what you’re shooting at. Recall that nuclear physicists are inveterate jokesters so it should come as no surprise that the dimension that the chose to represent the size of a “reasonable” nuclear target to be a barn, equivalent to 10-28m2.

The objective for the fissile and fertile materials is to be big targets for absorption. Moderators on the other hand, should have large scattering cross-sections and teeny-weeny absorption cross-sections.

From CANDU Fundamentals, Table 8.1, Moderator Requirements Table 1; US-DOE-HDBK-1019/1-93 and Neutron Interactions with Matter by P.Rinard

Material Thermal Neutrons E = 0.025 eV
Fast Neutrons E = 1 MeV
σt (b) σa (b) Σt (cm-1) Σa (cm-1)
H2O 2.18
D2O 0.410
C 4.95
9Be 6.35
Nat Mg 3.47

Related Properties

Material Atomic/Molecular Weight ξ Average Collisions Thermalize Moderator Ratio
H2O 18 0.927 19 62
D2O 20 0.510 35 21000
C 12 0.158 114 216
9Be 9 0.207 86 126
Nat Mg 24.31

Moderator ratio = (ξ Σs) ⁄ Σa
ξ = average logarithmic energy decrement per collision ln(Ei/Ef)
Σs = macroscopic scattering cross-section
Σa = macroscopic absorption cross-section, a nonscalar value differing for different neutron energies
Σt = Σs + Σa
1 barn (b) is 10-28m2

A moderator is desirable as it reduces the radioactive inventory. A fast reactor requires a much thicker blanket and a larger fissile inventory. An “external” reflector tends not to contribute to natural, negative feedback within the reactor.

The mean free path for the neutron in a material is λ = 1/Σt

As an upper limit, the thickness of the moderator required to thermalize fast neutrons is given by the average free path length multiplied by the number of collisions required to thermalize. As the collisions result in a random change of path after each collision, the resulting “thickness” is the total free path length which zig-zags through the material and so the necessary thickness can be substantially less than the total free path length. Mathematical simulations are used to estimate the necessary thickness of moderator. Being random, it is thought that higher numbers of required collisions will result in worse estimates.

For pure carbon, the total free path length to thermalize is in the order of 4.6 m. Obviously, a great deal of zagging and zigging occurs as only a small proportion of that length is required as moderator thickness.

The use of neutron diffraction (surface scattering) to detect film thicknesses on colloidal graphite in alcohol and water opens up consideration that a colloidal carbon in a molten salt would be a more effective moderator than the proportion of carbon would predict. After all, the fast neutron is thermalized by loss of kinetic energy by collision and the more energy lost in collisions one gets without absorption, the better.

Core Thought: Three-Fluid Reactor

A “normal” molten-salt breeder reactor uses a solid moderator between the fissile core and the fertile blanket. This is simple to implement, but the moderator material is subjected to high-intensity neutron flux and in the classical implementation, in physical contact with both molten salts, allowing fission and decay products to both react with and diffuse into the moderator.

The estimated production life of such a moderator would be around 4 years of continuous operation.

One method to avoid down-time to rebuild a core to replace a contaminated moderator was conceived by this author; extending the paradigm of the working fluid to include the moderator. i.e. make the moderator a liquid as well, separated from the others by a thin, physical barrier.

3-Fluid CoreA colloidal carbon in molten salt would be easiest to implement on “small” scale, with concentric cylindrical shells, the inner containing the fissile fuel, the middle the colloidal carbon, and the outer shell the fertile blanket. Separating the moderator fluid at either side would be a thin metal, preferably less than 4 mm, so as to minimise the influence on neutron flux.

The fuel shell would be of the order of 70 cm in diameter and between 2.5 and 3 metres long. The moderator shell would be up to 30 cm larger in radius (guesstimate) and the blanket enclosure perhaps another 50 cm more in radius. This brings about a rather “dumpy” dimension of 2.3 m diameter and an overall height of about 4.6 metres.

The density of the molten salt is about 3200 kg/m3, so the salt inventory, not including piping, heat exchangers, etc would be:

Component Volume (m3) Mass in tonnes Proportion
Fuel – Fissile 1.15 3.7 6%
Moderator 3.62 11.6 19%
Blanket – Fertile 14.3 45.9 75%
Total 19.1 61.2

Under nominal operation, no more than a thin metal barrier is structurally required as pressures can be equalized fairly easily by both flow rates and by using inter-connected, gas-filled pressure dampers to take care of surges. The outer concentric shell can be as thick as one would like. Convection of all the fluids tends to even out the temperature of all the shell, so thermal stresses can be minimised.

A liquid moderator brings advantages:

  • Physical separation from fissile and fertile materials as well as their decay products.
  • Ability to refresh the moderator while the reactor is online. This can be done by feeding in a fresh moderator of a volume corresponding to e.g. about 1% of total volume per week and bleeding off the “excess” to separate the components.
  • Ability to remove many “contaminants” from the moderator as gases while online.
  • An unstressed moderator.
  • A moderator than can also be used to extract heat from the core, including (most of) that from the kinetics of moderation.
  • As the liquid moderator is not in direct contact with either the fuel or the blanket material, it could technically be used to heat gas via a heat exchanger; to drive a gas turbine for power generation and/or circulation pumps.
  • The moderator may be diluted rapidly by bubbling an inert gas into the fluid as it’s fed into the reactor. Fewer neutrons would thermalize, reducing the rate of fission. The presence of gas bubble also enhances the negative void, with a hotter moderator more strongly reducing fission.

The disadvantages:

  • Metal between the moderator and the fissile and fertile materials. The walls between are subject to intense neutron flux so the materials must be stable under those harsh conditions. Alloys developed during and as result of ORNL MSBR research need to be verified.
  • Colloidal carbon concentration may be difficult to keep uniform. Carbon in other forms such as a (possibly fluoride-based) polymer which remains a stable liquid in the temperature range and has density similar to that of the salts dissolving U and Th may be feasible.

What can possibly go wrong? 😉

  • In the event of a moderator pump or other circulation failure, the moderator would overheat, causing its freeze-plug to melt, resulting in it draining off and fission effectively stopping.
    But: especially if the other reactor fluids also fail to drain, then the void previously filled by moderator would be stressed more than in normal operations. So the shell’s walls must be strong enough for that event.
  • In the event of a fuel circulation failure, the inner moderator shell might be similarly stressed; in compression. Compressive stress on the outer moderator shell increases.
  • In the event of a blanket circulation failure, the outer moderator shell may be similarly stressed; in tension. The tensile stress also increases on the inner moderator shell.
  • If both fuel and blanket drain, then the moderator fluid exerts increased compression on the inner, and increased tension on the outer, in a similar, but lesser way than the fuel and blanket losses by themselves.
  • Similarly for the losses of fuel and moderator; and moderator and blanket.
  • If at all feasible, the start of draining of one fluid should inherently result in the start of draining of the other two, at a rate so that the level (height) of the fluids remains approximately the same.

How to Douse the Reactor

Another means of bringing the reactor down safely, without undue mechanical stress, is to allow the draining action to siphon in flush salt that contains no Uranium, Thorium or even moderator. By sizing and arranging the drain receptacles accordingly, there would be little dilution of the working salts by the flush salt.

If only one working fluid drains from the reactor, its space will be filled with molten salt of approximately the same density and temperature; but inert, for purposes of the reactor. This removes the need to ensure a passive, hydrostatically synchronised drain of all 3 fluids.

In normal operation, the flush salts are ostensibly pumped through the primary heat exchanger, then to a secondary to e.g. work turbines, then go via a surge tank back to the suction side of the circulation pump. If that surge tank is elevated; maintaining a static head above the pump and the reactor vessel, then an inverted-U section connecting to the top of the reactor can isolate the secondary from the primary fluids with an inert gas “trap”, with bursting discs ass necessary. As the relative pressure in the reactor vessel falls, it’ll tend to draw in flush salt which has atmospheric pressure on the top of it. With sufficient difference, the discs will burst and the flush salt rush past the top of the bend, after which it’ll draw in flush salt by pure siphon action.

The flush salts could be siphoned from the secondary side of the primary heat exchanger, which in normal operations, separates the contained radioactive salts from those used to drive the machinery and processes outside the reactor containment. Of course, once the flush salts are drawn into the working salt piping, they cannot be immediately be reused for purposes outside of containment. One must assume contamination requiring containment as there is very likely to be some residual working salt in the reactor, pumps and piping.

The hydrostatic pressure from the elevated surge tank will push out fluid from the reactor more quickly than that fluid working under its own weight. The nominal flush volume is set by the depth at which the siphon is set in the surge tank. Once it draws air past the top of the salt remaining in the tank, the siphon effect breaks.

A variation of the siphoning is to include a “neutron poison” in the flush salt. Such could be valuable in destroying the fissile core material in case of a tangible threat of hostile action on the reactor building.

Updated 2011-12-14: Added tags to sections
Updated 2011-12-13: Expanded introductory part
Updated 2011-12-11: Moved to sub-page. Table tidied
Updated 2011-12-04: Added siphon flush and breed control via moderator concentration