Part 8 – The Molten Salt Reactor concept

Schematic layout of the proposed Molten Salt Reactor concept (from Wikipedia)

8.1       Introduction

Some nuclear energy supporters would say that the problems I’ve highlighted so far are there because we stick to a policy of using nuclear fuel in its solid form. If we were dealing with any other sort of chemical process, we’d liquefy it and run it around a chemical plant. This in essence is the idea behind the Molten Salt Reactor.

In a molten salt reactor, the fuel is dissolved within a fluoride salt mixture (producing either Uranium Fluoride or Thorium Fluoride) and circulated around a graphite moderated core. While deliberately allowing a nuclear reactor’s fuel to melt down might, from a safety point of view seem slightly counter intuitive, actually it does present certain advantages. Obviously a “melt down” accident is no longer a worry (its already melted down!). UF4 and TF4 both have a negative temperature co-efficient, thus when the temperature drops, the reactivity increases, when it rises up, the reactivity decreases. Also, as the fuel requires slow neutrons to maintain chain reactions, removing the fuel from the core, should shut down the reactions. Many Molten Salt designs have a series of emergency dump tanks at the base (thus gravity fed) into which the fuel can be dumped in an emergency. As a further backup measure the valve controlling the tanks can be a freeze plug of solid salt. If the temperature in the core exceeds some threshold level, or the cooling loop/blower maintain the freeze plug is turned off, the fuel escapes out of the core and into the dump tanks.

Thus a MSR type reactor offers several passive safety advantages over conventional reactor types. Another advantage is the ability to process the fuel as the reactor runs by passing it through a chemical plant, this removes the various isotopes that would otherwise “poison” the nuclear reactions and force a shutdown and fuel replacement (as is regularly the case in a conventional reactor). This would also reduce the nuclear waste volumes produced by these reactors. MSR’s also operates at a low vapour pressure, which means they don’t need the high pressure forged parts of other reactor types (but it still presents some material challenges, as I’ll discuss later). In theory the core should also be relatively small and compact (indeed the initial attempts to build one were focused on the idea of a reactor that could be carried inside a plane, see ARE) which of course has numerous advantages. It may also be possible, as with many Generation IV reactors, to use MSR’s to produce hydrogen via the Sulfur-Iodine process and utilise the more energy efficient Brayton cycle.

The MSR and the LFTR (or lifter) reactor has something of a cult following on line. Unfortunately, this has led, through a process of what amounts to internet Chinese whispers to a lot of myths and miss-conceptions about the LFTR building up. Up to the point where its started to take on aspects of a Scientific Cargo Cult. As I go along with the technical analysis of this reactor concept, I’ll be taking the opportunity to debunk a couple of these, as they do a disservice to the design, and get in the way of the genuine science.

 8.2       The MSRE experiment

Firstly, there is a view that the LFTR concept has been “proven” already via the Molten-Salt Reactor Experiment (MSRE) project in the 1960’s. While it is certainly true that such a reactor ran successfully for 4 years and that this project proved that some of the ideas behind the MSR have merit, there are a couple of key things it didn’t do. Notably, it never generated a single watt of electricity. As I’ve mentioned previously the turbo generator systems for high temperature reactors is technically challenging, especially for the LFTR as the molten salt presents a number of design challenges.

The ORNL Molten Salt reactor experiment of the 1960’s

That said, the goal of the MSR experiment was to prove the reactor concept, not develop turbo machinery kit, which would have been a serious (and costly) distraction. The molten salts at the MSRE were passed through a cooling loop and fans used to blow the pipe work cool again. Stories of said pipe work glowing red (see below) are worrying, as it indicates they were operating well within the thermal creep zone. At the time very little was known about thermal creep, in particular the delirious effects of neutron bombardment on exacerbating the problem. Consequently, its unlikely one could utilise the same design spec today for a commercial plant. Indeed, reports of distortions in the graphite moderator after just a few years exposure and worse  inter-granular cracking (a corrosion related failure phenomenon usually caused by excessively high temperatures) of some metal components exposed to the molten salt, suggest it was operating well outside the limits of what would count as a reasonable safe technical envelope (at least for a commercial reactor with a long operating life). As I will detail later this has significant design implications. The reactor also spent a good deal of its time down for maintenance.

The cooling circuit of the MSRE glows red hot due to its high operating temperature

Also, the MSRE never included the more tricky Chemical Processing Plant. One was designed by ORNL but never installed. Aside from using a chemical spray technique to separate out the more nasty neutron “poisons”, such as Xenon-135, much of the remaining “chemical plant” functions of this reactor design have never been tested. While the MSRE did run once on U-233, this was generated off site, not by the reactor itself. Finally, as I hinted earlier, 40 years is a long time. Very little of the technical side of building this reactor would be relevant today given how much technology, especially material science has changed. Many of the scientists who worked on it are either dead or retired. While one won’t be starting off with a blank sheet of paper, you probably won’t find yourself far removed from that.

8.3       Thorium Cycle questions and problems

Questions have also been raised by some nuclear scientists about the Thorium cycle, in particular the proposed one that the LFTR would use. I’m not a nuclear physicist so I’ll merely forward you on to the relevant paper here, and a rebuttal here. The crux of the argument seems to be the proliferation risk (I’ll come back to that one later), the fact that a number of its spend fuel outputs (such as Technetium-99) are “nasty stuff” with a long half life and the fact we’ll still need supplies of Uranium to get Thorium reactors going again whenever we have to turn it off (which will happen at least once a year or so during its annual maintenance shutdown). They also highlight a number of technical issues, which I discussed in the chapter on HTGR’s.

Certainly the fission products from a Thorium reactor are a worry, Technetium-99 has a half life of 220,000 years, uranium-232 produces thallium-208 (a nasty wee gamma emitter), Selenium-79 (a beta emitter with a 327,000 year half-life), even Thorium-232 is a problem with its half life of 14 Billion years (and while the T-232 isn’t a major worry its only mildly radioactive, all the time during this 14 Billion years it will be decaying and producing stuff that is!).

The UK based NNL (National Nuclear Laboratories) also pour cold water on the idea of Thorium fuelled reactors (see here). While the report is low on detail (they seem to be saying “trust us we’re scientists who work with nuke stuff… and we smoke pipes!”) they do highlight the major time delays it would take to establish and get working a Thorium fuel cycle (10-15 years with existing reactors, 30 with more advanced options), point out that under present market conditions its unlikely to be economically viable and will (as the points above raise) offer only a modest reduction in nuclear wastes.

MIT recently undertook a study of future nuclear fuel supplies. The Thorium cycle barely gets a mention, and even then its usually in relation to Fast Reactor programs (of which the US currently has none) and modifed LWR systems, rather than the MSR.

Obviously, once we exhaust the world’s U-235 stockpiles, LFTR’s and any other Thorium fuelled reactors will cease to function. Indeed long before then the spike in Uranium prices will have rendered MSR’s (and all other nuclear plants) uneconomically viable (of course there’s plenty who’d say that’s already the case!). The LFTR fans usually groan at this point and state that “all we need is a little plutonium”. Now while I’m quite sure that in the fantasy world which the LFTR fans inhabit Plutonium is available in any good hardware store but back in the real world, it’s a little harder to come by! As with the HTGR’s using Thorium (if its possible) would certainly help stretch things out….a bit! But not by nearly as much as the supporters of Thorium reactors would have you believe.

8.4       The Chemical Seperation Plant and waste output

One other misconception on the internet is the view that a LFTR reactor will produce almost no nuclear waste, as the following You-tube video implies  (or see this “activists” banner here). This is not the case. All the while during the plant’s operating life that chemical plant will be producing nuclear waste material, and as discussed earlier some of that is pretty “nasty stuff”. Not a lot of it per day, but it all adds up! Also the supporters of the LFTR seem to assume that this CPP can operate with 100% efficiency (i.e remove all the radioactive poisons). This would be very technically challenging, especially in the LFTR case given the importance about separating out of U-232 (and its Thallium-208 payload) from U-233 or indeed removal of protactinium-233 as well as a host of other nuclear “poisons” discussed. Build up of these in the core both leads to increased irradiation of the core as well as the eventual shutdown of the nuclear reaction process altogether.

An CPP facility capable of that level of operating efficiency would likely be physically very large. Given that it will be working with radioactive materials, and the real radiological hazard is a pipe burst (an all too common occurrence and any chemical plant, and especially likely at these sort of working temperatures and radiation levels), we would thus need to put the CPP underneath our concrete containment dome. Obviously a large CPP will not only be expensive to build and maintain but greatly increase the size of this containment structure, further increasing reactor construction costs as well as increasing construction time (and reducing the number of said reactors we comission in any given time period).

And of course the supporters of the LF reactor concept have yet to come up with a functional design of an CPP. I’ve seen various dusty line drawings of the 1970’s ORNL proposal, you can see them yourself here (more info from ORNL about MSBR’s can be found here) but that’s it. I would firstly note that materials science and chemical processing technology has moved on hugely in the last 40 years, so I doubt it would be sensible to build an CPP as shown in these plans. A new one would have to be redesigned (all but) from scratch.

The LFTR supporters have tried to counter this by coming up with designs of their own, but I’ve yet to see an actual working schematic, one that specifically discusses cycle efficiencies and above all else ENERGY INPUTS! The designers of this reactor seem to be assuming that this CPP, which will involve various stages of pumping, sparging, vacuum processing and filtering of the working fluid, often at a variety of set temperatures or pressures will operate with no net energy input and achieve 100% separation efficiency! In science we have a technical term for such a belief.

As the working fluid will be coming off the exhaust from the heat exchange cycle it will be relatively cool (in the MSRE it was at around 570 °C) yet some of these processing stages will require the fluid to be heated back up to 1,600 °C. Where’s that energy going to come from? We could use pinch technology and “pinch” (if you’ll pardon the pun!) some of that heat from our heat exchanger, but that has the disadvantage of yet more piping (and more safety risks!) and a reduction of heat exchanger efficiency. And this isn’t going to solve the issue of all the other kit I mentioned. Notably, the LFTR supporters have suggested (see here) using electrolysis to help improve the filtering efficiency of their plant. An excellent idea, it would solve a number of problems, but unfortunately electrolysis systems practically eat electricity! Where’s all that electricity going to come from?

Fortunately its probable some form of balancing point can be achieved in which we compromise the standards of our CPP, accept one that is smaller and less efficient (and thus our reactor burns much more fuel and produces more waste) but is sufficiently efficient to give us a decent fuel burnup rate without being over complex (or large), nor energy hungry. Of course exactly where this “balancing point” lies is the question and it would take some degree of research and experimentation to find out. Inevitably from time to time (probably at least once a year or so) we’d likely need to dump the entire core’s contents and replace it with fresh fuel. The “dumped” contents being added to the global nuclear waste stockpile.

Of course the wider economic problem with this CPP is the fact that we need to install and pay for a chemical plant right next door to our reactor, as well as pay staff to run it, buy in chemical feedstock and accept the fact that some of our valuable electricity coming out of the power station gets consumed on site. By contrast alternative record designs do not come with any of these costs, nor of course do fossil fuel plants nor renewable facilities. While nuclear waste output will be reduced, as with the GcFR (see chapter 7) we have to question whether the increased costs imposed by this CPP make it worthwhile, and whether we’d be better sticking with deep geological storage of waste.

 8.5       Graphite core and Fire Risk

Another issue is that graphite core. As I detailed previously with regard to the HTGR (part 6.4.3) it’s a potential fire hazard. Thus we would need to put the MSR within a containment dome of sorts. Again, as with the HTGR, this dome need not be built to the same exacting standards of a LWR dome as we are merely trying to contain a graphite fire, not an out of control reactor. We would need an effective on plant fire control team and some form of fire detection and suppression system, within the containment dome and all the necessary gear that this entails. Again, I refer you the relevant section of the HTGR anaylsis, but needless to say such an arrangement would involve certain costs.

One solution would be to remove the graphite core and seperate the reactor into two layers (see here) with an outer breeding blanket feeding on the neutrons (that would otherwise be absorbed by the graphite) produced by the active core. This would make the operation of the CPP a bit easier and thus more economically viable. While such an arrangement gets around the fire risk issue, it would be technically harder to build and complicate an already narrow material choice, largely due to the higher neutron flux parts of the core would now be exposed to, as well as well as the generally more complex nature of such a core arrangement (i.e. harder to manufacture, assemble and maintain).

8.6        Why air cooling a LFTR would be a very bad idea

Another misconception is that LFTR’s can be air-cooled (here and here) rather than being dependant on the water cooling process we utilise in most other power stations. I’m assuming this rumour got going as a result of the fact that the MSRE was air-cooled. While this is true, you could air cool any power station (indeed many small diesel fired units are typically air-cooled), it’s just there are a host of good reasons not to!

Firstly, fire safety, air is an oxidising substance. Fires start all the time at power stations (fossil fuel fired and nuclear ones), especially in the turbine halls and the last thing we want in an emergency is a load of big cooling fans blasting in air and literally fanning the flames! In this scenario we’d face the dilemma between stopping the fans and cutting of the source of cooling (forcing us to SCRAM the reactor to prevent a LOCA scenario) or risk the fire spreading out of control, possibly to the point where it compromises the reactor’s safety. This was of course very similar to the dilemma faced during the Windscale fire, which was air cooled (although in this case directly, rather than indirectly as we currently discussing). And on the subject of Windscale, you will recall what I said earlier about fires and that Graphite core, so we’d be opening a very serious potential safety loophole.

Cooling fans also aren’t terribly reliable, which is why the MSRE was down for several months due to a cooling fan failure. Air based cooling is also very weather dependant, indeed I note that the fans at the MSRE seems to have failed in the summer, when they would have likely been struggling to cope with higher daytime temperatures.

Thirdly, it’s the matter of thermal efficiency. Air based cooling is not very efficient, largely because air has such a low heat capacity compared to water (1.15 against 4.2 J/kg K). A typical COP (Co-efficient of Performance) for fans would be of the order of 2 – 3.7, while you can get 5 – 7.5 with water based cooling. Assuming a COP of 3 (it would be more like 2.5 at the temperatures in question, but bear with me!) and assuming a 1,000 MWth LFTR with a thermal efficiency of 50% (to keep my numbers easy!) = 500 MWe. Our cooling fans, in order to dispose of that 500 MW’s of excess heat, would be consuming 166.67 MW of electricity, dropping our effective plant efficiency down to 33%, barely Rankine cycle levels! This is why we use water in most power stations for cooling.

Also this air based cooling argument strikes me as a bit of a red herring, LFTR fans essentially inventing reasons why their “precious” is better than anything else. With the exception of a few geothermal power stations in arid areas (or hydroelectric plants!), I’m unaware of any major power project that was derailed for lack of cooling water. Either you can use cooling towers (forced draught or natural convection types) and minimise water losses to an acceptable level or simply move the plant next to a ready water source and transmit the power to where it is needed. Many desert countries operate large thermal power stations from around the coasts and several such as Iran, UAE and Libya are even planning to build nuclear stations too. So I fail to see how “air based” cooling offers any real benefits.

One option for MSR’s is so called “dry cooling” using condensers (see here). This relies on the high thermal mass of a liquid working fluid to take away heat as its passed through the condenser matrix (important note, it does not rely on the evaporative cooling effect as the two previously mentioned cooling methods do). However, obviously it requires water on site (which has various design implications for our thermal plant) although the water usage levels are low. Such arrays can be bulky (compared to forced draft or cooling fans, though smaller than Hyperbolic cooling towers) , less energy efficient (though better than direct air cooling, i.e. fans) and thus will consume some part (maybe as much as 5-8%) of the power stations electrical load. But again, whether you utlise such a system or not really depends on the circumstances where it is built.

 8.7       Why power cycling a LFTR would be an even worse idea!

Some LFTR supporters seem to be of the opinion that LFTR’s can be used for more than just baseload power, with them being used to meet the needs of variable electrical loads  (which is currently performed by fossil fuel plants and hydroelectric systems). I think this one got going over confusion regarding the MSR’s aforementioned negative thermal co-efficient, which some mistakenly assumed implies a variable load supporting capability. In this video here the speaker seems to imply this (but he specifically didn’t say it). It is worth noting the differences between the MSRE (a small test reactor) and a much physically large production reactor, with inevitably a slower rate of reaction.

Firstly, it is worth noting that existing nuclear stations are capable of some level of power cycling anyway, just not much! The truth is that the LFTR is as constrained in it power output capabilities as other reactors, possibly more constrained in fact. You will recall what I said in the materials section about thermal creep. You will also recall me stating the need to avoid thermal cycles (i.e regular increases and decreases in temperature). Power cycling a LFTR would necessitate such cycles, worsening our already narrow materials choice and requiring a much more heavily constructed reactor. Indeed the MSR suffered a number of problems likely related to (or made worse by) the excessive thermal cycling of its core (which went through nowhere near the sort of paces it would be if the LFTR fans proposal was adopted). Notably the aformentioned distortion of of graphite core elements under irradiation (discussed here and here) and the failure of its freeze valve due to “thermal fatigue” (pg 39 of this report).

A typical load daily load cycle graph for the United states in summer weather

Furthermore there are a host of other good practical reasons why we would wish to avoid any power cycling. That CPP, which requires the input MS/fuel mixture at a series of given temperatures, won’t work terribly well (i.e less efficiently) if the temperature falls much below a certain threshold (i.e. we’ll be “pinching” more heat from the heat exchanger or having to install supplementary heating elements to bring up the temperature). Indeed the chemical plant might even be damaged by such an event (requiring a reactor trip on safety grounds).

Also a MSR with dissolved Thorium (or Uranium) present has the problem that as it cools back down towards its solidification temperature, the Thorium (or Uranium) below a certain threshold of temperature and pressure will begin to solidify with small flakes of solidfied fuel forming within the mixture. This can cause all sorts of problems, with fuel channels, valves and pumps being potentially clogged. Several meltdowns of liquid metal cooled reactors have resulted from such clogging incidents. The usual cop out for the MSR (dump the fuel & salt) won’t work if channels are blocked by solidified fuel. Also corrosion, abrasion and expansion related damage becomes a danger. Most pumps are designed to either handle a gas or liquid. They tend to be intolerant of a mixture of both ( as would be the case if we pushed the temperature too high, towards the vapourisation temperature). Passing a mixture containing partially solidified fuel through them would be the equivalent of taking a sandblaster to the pumps. While the above happening occasionally under timid operating conditions, as it inevitably will during reactor shutdown or start up scenarios, is nothing to worry about. However, using a MSR to perform short aggressively power cycling on a daily basis will very quickly invalidate your warranty and endanger safety. So no, you can’t use a MSR for anything other than baseload power, or industrial heat provision.

Sodium Secretions within the Dounreay PFR demonstrate the hazards of channels being clogged by solidified fuel

8.8      Thermal windows and material choices

The  problems above demonstrates that the MSR has a relatively narrow thermal window. Its filtering plant will not work if the temperature of the fluid drops much below a certain threshold and the danger of fuel solidification raises the risk of the reactor being damaged. With UF4 the solidification temperature is 1,036 °C and its vapourisation temperature is 1,417 °C at atmospheric pressure. We can move this temperature “window” by altering the working pressure (the original MSRE operated at a work pressure of around 0.3-0.4 bar and as I recall operated within a thermal window of 700-560 °C) but this comes at the disadvantage of increased working stress (the pressure outside is trying to squeeze the reactor walls inwards, although the effect is generally minor) and lower thermal efficiency (more on that later) or by increasing the salt concentrations (lowers power density, may increase corrosive effects, thought that depends on a number of other factors). With TF4 our “window” is 1110 – 1,680 °C, but again we can potentially move this by lowering the pressure (or raising it if we want to go the other way…not that we do!). A low vapour pressure also creates a few potential problems in terms of keeping the reactor sealed (air is more likely to leak in if the pressure inside is less than atmospheric…possibly starting a fire!) and maintaining a good flow rate from our pumps.

Either way we are likely to be operating a MSR well outside of the operating temperatures of any stainless steel alloy and likely forced to rely on using Nickel Alloys such as Hastalloy instead. This will of course be expensive. For anyone thinking, given the temperatures posted above (a recap the operating temperature window for Hastalloy is 1050 – 1200 °C), that I’m sailing a little close to the wind here (as in have a forgotten about thermal creep). I would remind you that assuming we are not actually power cycle these reactors (that would be silly!) and that by maintaining a low pressure cycle (and again lowering pressure below atmospheric lowers the working temperatures, again it was 700-560 °C for the MSR) that should drop us to within the window of probability that we can get by with.

That said the issue of Inter-granular cracking would need to be investigated, as it could prove to be a show stopper for Hastalloy usage. Also the MSRE suffered, as noted, a very worrying thermal creep failure of part of its drain valve (mentioned on pg 39 of this report), probably due to excessive thermal cycling (again this is why power cycling the reactor would be a bad idea!). There were also problems with Grain-boundary embrittlement, and a decrease in creep ductility. Again I would include the caveat that these issues would also need to be addressed in detail before one could proceed with Hastalloy use. Many of the attempts at Fast Breeder Reactor development in the early days were held up by a failure to get to grips with material science issues from day one. Consequently if the MSR researchers neglect to answer these questions above they could find history repeating with a series of costly and time consuming delays.

With the LFTR however, weighting up all the evidence, I doubt you could operate one made out of Hastalloy, contrary to everything said on the internet. Bare in mind I’m thinking in terms of a good lengthy service life with a sensible factor of safety, not a flimsy test reactor in a lab (with a 100 mile exclusion zone!). While a LFTR/MSR might be less vulnerable to thermal creep than other reactors (thought that depends on a host of specific design issues), it certainly isn’t immune. Also, as a minimum standard of safety it would need to be easily able to withstand the vaporisation temperature of TF4 at atmospheric pressure. Failing to do so would undermine our entire concept of passive safety, and like the LWR we’d have to start adding this and that to the reactor and soon it will be the size of an aircraft carrier! Our LFUR (might!) just about limbo under the bar on this point (if we can address those material science and corrosion issues detailed earlier), but I seriously doubt the LFTR could do it. At the very least a good deal of material science would need to be done to prove the issue either way.

Thus the pressure vessel of any LFTR would likely have to be formed out of either an alloy of Tantalum or Niobium, or possibly a Nickel alloy with a coating of either of the latter two. Both of these options would “complicate” the design, though not nearly as much as the other alternative, using Ceramics! (very expensive and difficult to form, especially given how critical getting an air tight seal is given the graphite core). For both the LFUR and LFTR key internal components will in any event likely have to be made out of Refractory metals, particularly certain high temperature parts of any CPP, given talk of operating temperatures in the range of 1600 °C (again I direct the reader to  chapter 3 on materials where this is discussed in more detail). The bulk of the rest of the power plant however would run at much lower temperatures (and neutron fluxes) and thus could indeed be made out of Nickel alloy (such as Hastalloy), but again only if this inter-granular cracking issue gets tackled successfully. As mentioned in the section 8.5 as regards the graphite core, a double blanket core (see here) would have even tighter material requirements due to the increased level of neutron bombardment and the thermal gradients imposed on the core by such a split pressure vessel arrangement.

So my instinct from a materials science point of view would be to drop the LFTR idea altogether and focus instead on a LFUR. While this isn’t able to use the Thorium cycle, the point was raised earlier about how the Thorium cycle might not be all it’s cracked up to be. Its going to be a lot easier to build a LFUR than a LFTR, cheaper (relatively speaking) and likely safer too. Of course it does come at the disadvantage of a slightly awkward acronym! but overall that would be my focus of attention, at least initially.

8.9       The Brayton cycle and MSR reactors

Another misconception is that an MSR can operate on an open cycle with a Brayton turbine. While true, it could be run this way, there are a host of practical reasons not to do it. Not least of them the fact that our turbine would have to be designed to withstand having a mixture of molten salt and fluorided fuel passed through it at very high temperatures. This would be tricky to say the least, likely requiring the use of those super expensive refractory metals, and while using such materials to make the odd turbine blade is one thing, an entire turbine casing is an entirely different matter. It would likely cost much more than the reactor itself!

Furthermore it won’t be terribly efficient, you will recall what was said earlier about our MSR having a very narrow thermal window. In most other situations (involving Gas turbines) we would get around this problem by using large pressure drops or a multi-stage turbine cycle to improve efficiency, but neither of these are options if running a MSR on an open cycle, at least so long as we want to run the reactor at a low vapour pressure. To be blunt if you managed to beat Otto cycle efficiencies (20-30%) with a MSR running on an open cycle (with a Brayton turbine), you’d be doing well! Also, as this would involve circulating radioactive material outside of the pressure vessel, the entire turbine set would need to be built under our concrete containment dome, further increasing costs.

It would make far more sense, from an energy efficiency, safety and economics point of view to simply split the cycle in two, as shown in my schematic previously (nicked off wikipedia), one half operating within the pressure vessel, passing through a heat exchanger (where we lose some energy) where heat is then transferred to the secondary loop and into the turbogenerator plant (or hydrogen plant), which runs on either steam or an inert gas.

Brayton cycle against Rankine cycle

I might also take the opportunity to correct another misconception I’ve seen several MSR bloggers make about the Brayton cycle. A number seem to be convinced that the key to the Brayton cycle is that its efficiency is dependant on Tmax = Constant, where as the Rankine is dependant on ΔT. That’s not entirely correct; a substantial temperature drop is still required in a Brayton cycle to achieve good efficiency. Think about it, if we keep ΔT at zero (i.e the fluid does not change temperature) then obviously the energy output from the process drops to zero and thus our efficiency is therefore zero! The “trick” with the recuperative multi-staged Brayton cycle (and we can’t do this with an open cycle) is to utilise large changes to ΔP (the pressure) and the gas laws  to produce large values of “Δs” the change in entropy. In essence what we’re trying to do is arrange such that the critical ΔT is the temperature difference between the exit temperature from our heat exchanger (to the turbine) and the temperature of the cooling water inlet (T3 and T1 respectively in the page here). With a Rankine cycle we are constrained to whatever temperature drop can be produced over the turbines, which is generally smaller than that achievable with a Brayton cycle.

So while its not as critical to get a nice large ΔT with a Brayton cycle as a Rankine, you still need to get some sort of a temperature drop. This is especially true for any MSR for as I’ve pointed out, its efficiency is ultimately going to be dependant on the temperature drop we can produce over the heat exchanger.

Again our narrow thermal window and low vapour pressure conspires against the MSR here. Given our limited ability to modify ΔT we must instead increase the mass flowrate through the heat exchanger to ensure good heat transfer. What this means in simple terms is having a pair of big high capacity pumps either side of it running at full speed. Unfortunately that low vapour pressure, means the volume of fluid we need to process from the reactor is quite high. And given what we’re pumping (molten salt at temperatures of 700 °C), this quickly leads us into needing very large and powerful pumps made out of exotic materials – which are of course expensive and energy hungry. That energy being charged against the power output of the reactor. Again, some sort a balance point will emerge between increasing pump power consumption (and capital costs) and optimum reactor efficiency. This also leads us to rule out MHD as proposed by some LFTR activists, either for pumping purposes or otherwise, as they are simply too energy hungry and expensive. Plus the radiation they’d be exposed to might well give them a short service life.

A set of large pumps also rules out the possibly of us using natural convection to maintain fluid flow (another misconception). Much as I am a fan of natural convection (did my PhD in it!) it just would not provide the necessary heat flows. This of course opens up a possible safety window in that our reactor will no longer be able to cool itself down via natural convection and thus becomes vulnerable to a LOCA scenario in certain situations (fairly rare situations mind, and even then were likely just talking about localised core damage rather than anything more serious).

All in all my suspicion is that our heat exchanger would struggle to produce thermal efficiencies any greater than 70-75%. Assume a good high efficiency Gas driven Brayton cycle the other end (55-60%) so that yields us an overall efficiency of 38% – 45%, oh! but we almost forgot about that chemical processing plant, plus the cooling plant, and there net energy inputs, say we deduct 5-10% of reactor power output to account for running that, so overall between 29% – 40%, with a 35% overall efficiency being my best WEG. This is about comparable to current day power stations, but much less than what can be achieved with other nuclear reactor designs. While if we can get that CPP to work reasonably effectively, we could achieve a high fuel efficiency and lower levels of nuclear waste, they would not be not as much as the supporters say. And the danger is that if we were forced to ditch the CPP as uneconomic or not technically feasible, then we could end up with a MSR that burns more fuel and produces more waste, and yet still is less efficient that its principal competitors (namely those gas cooled reactors).

8.10  Piping, FMEA and leak prevention

As I mentioned before, the major risk to any MSR reactor is not a meltdown (its already melted down silly!) or a LOCA (there are some scenarios that could lead to major core damage but they are unlikely to release radiation into the environment) but either a fire effecting its graphite core (which for a LF reactors running at low vapour pressure is a greater risk than with any other graphite cored reactor) or more likely a burst pipe.

In the chemical industry pipes burst all the time. The consequences can vary from minor “oh! Fiddlesticks we appear to have spilt a load of sulphuric acid on the floor, quick! flush it down the drains before the EPA shows up” to utterly catastrophic, as seen in the Bhopal disaster. Thus the major danger with a LF plant is that somewhere in the lengthy network of pipes that it and its CPP consist of, something breaks. You may recall me expressing distaste for the mass of pipes that is a CANDU reactor; you may recall me going so far as to describe it as “a plumber’s nightmare”. Well in fairness, at least a CANDU is made out of easily welded materials! A LF reactor would likely be made out of materials that are either difficult to weld (nickel alloys) or (for ceramics and refrectory’s) down right impossible to weld! Again, putting such a large network of pipes together and getting it all signed off by the safety inspectors will be a difficult, lengthy and an expensive undertaking. Pipe bursts have already played a role in plaguing the nuclear reprocessing industry with problems, most famously the pipe burst in Thorp that led to 160 kg’s of Plutonium leaking into the basement, triggering a level 3 nuclear alert.

All this pipe work raises a whole host of design issues. As already pointed out we’d have to put much of the chemical works within the reactor containment dome, which would have to be hermetically sealed so that any gases can’t escaped (plus a gas treatment plant and a few emergency gas storage tanks on site). Also we would need to include a high-spec containment area in the basement of the reactor building to catch any escaping radioactive material, or chemical pollutants. But the individual components of the chemical plant/reactor would also need very careful design.

LFTR reactor and dump tank

Take for example the pipe at the base of the reactor that allows us to dump the core to the emergency dump tanks. This is a safety critical component – i.e it HAS to work in certain worst case scenarios. But suppose for example that it bursts during a dump scenario? Obviously we need a containment vessel around the pipe to catch any leaks. Also simply relying on gravity would be inadequate in certain scenarios, a pump on a separate stem (or a tank of inert high pressure gas connected up to the pressure vessel to “encourage” the fuel to drain away), would be necessary. But what if the trigger for the accident is a clogging of fuel channels (as discussed earlier) by solidified fuel? If we dump in that scenario we might cause the dump pipe to clog also, likely leading to a criticality incident or its failure and a breach. So we would need a thermal regulation system around the pipe to ensure it can be heated or cooled as necessary.

Also I don’t like the idea behind this “freeze plug”. I realise the passive safety benefits it brings, but it’s just going to be too slow to act in a real emergency and there’s too much that can go wrong with it. If I were an engineer at such a plant I’d want a big shiny red “dump core now!” panic button on my control panel. So we’d need to put a bypass valve in around the freeze plug, which could be activated in an emergency. In any event, as we’ll periodically have to inspect this pipe, a set of isolation valves would need to be installed around the freeze plug (or any pump) as a matter of routine anyway. Also relying on just one pipe, even an over designed one, doesn’t sound sensible. We have multiple dump tanks anyway, so why not have multiple feed pipes into them?

If you follow through the last paragraph you’ll see how in the process of getting one short section of pipe back to within a reasonable safety margin the result has been for it to balloon into a massively complex system in the space of 5 minutes. Obviously if we were to turn a similar critical gaze on many other parts of the LF reactor design (which of course we can’t as many of them haven’t been designed yet!), I suspect the same thing would happen again. Another common misconception of the supporters of the LF reactor is that it would be quicker to get a LFTR certified than a “solid fuelled” reactor, as the latter require being run through several fuel loads before going into commercial operation (which takes months or years).

However, the benefits of a Molten-Salt fuel system are outweighted by the lengthy inspection process of all that pipe work. All of those joints, valves, flanges and connectors would need to properly inspected (X-ray, ultra-sonic testing, etc.), something that would likely take months – and government inspectors aren’t exactly noted for their speed and efficiency! This is precisely why the ESBWR goes the other way and tries to eliminate complex pipe work. It would also be necessary to run the processing plant (under supervision) for a few months after inspection before the safety authorities signed off on it as ready for commercial operation.

There is also an issue regarding the fact that the Fluoride salts naturally produce hydrofluoric acid (when in contact with moisture) as well as decomposing into Flourine gas over time when cold. Both of these can lead to the release of toxic fumes, bearing in mind that any MSR reactor would inevitably have to have large stockpiles of salt stored on site or nearby, as well as some spent fuel. The MSRE already suffered a release of such gases from stored fuel tank and nearly suffered a criticality incident as a result. Radioactive or not, Fluorine gas is extremely toxic (several times more deadly than chlorine which was used as a chemical weapon in World War I) and anything that risks releases of that must be limited. Another misconception of the LFTR fans is that LFTR’s will not require the same large exclusion zones as other reactors. Actually, once the merest hint of this fluorine gas risk is raised, bearing in mind the consequences of incidents such as Bhopal, it would likely make getting a LFTR build anywhere near a large populated area, even harder than it would be for a conventional reactor.

A LFTR is essentially just a glorified chemical plant and getting chemical plants approved anywhere near a populated area is always difficult, doubly so I suspect if the locals learn it will be processing radioactive materials!

8.11     MSR’s, A proliferate problem?

One final point about MSR’s is the proliferation risk. As they can be used to breed U-233   (a potential nuclear bomb making material) this raises the risk of them being used by certain states as nuclear bomb factories. The supporters of MSR’s counter that the reactor’s CPP could be set up such as to make U-233 production impossible (the reactor would burn up all the U-233 in the core). The critics counter than a few screw turns from a suitably qualified engineer would undo such a setup.

My take on this is that it entirely depends on who you selling these reactors too. If the Russians are building MSR’s we have little to worry. Russia already has plutonium coming out of its ears and a multitude of means of making more of it. Yes, Russian engineers could easily modify a MSR to breed U-233 without the IAEA noticing (bear in mind North Korea and Israel managed to make Plutonium right under the IAEA’s noses), but I doubt they would do that. And would it make a huge difference to the geopolitical landscape if they did? On the other hand, if Iran suddenly announced plans to build a load of MSR’s……

Of course the LFTR fans will usually claim at this point that seperating out the various isotopes and chemical compounds would be very hard to do. An excellent point, but isn’t the whole point of the LFTR that such filtration can be done quite easily?…..obviously, both of these positions can’t be correct! You can’t have it both ways!

8.12     The Micro-Fuji MSR, proof of concept or white elephant in miniature?

The only proposed MSR concept currently being seriously pursued is the Fuji MSR, or more specifically the mini-Fuji MSR, which aims to build a 10 MWe prototype. Its design proves a lot of the points I’ve been making as I was going along. It is contained within a large concrete containment building (presumably to contain the fire risk I mentioned). It uses a secondary cooling loop, likely water cooled. There is little mention of any filtering plant (it is discussed as an option for the larger follow on plant), suggesting they’ve gone for the minimum possible filtration to keep the reactor going and will periodically flush and dump the fuel in the core. It also seems they plant to initially run on UF4, and will explore Thorium use later.

Also, there appears to be a funding issue, as very little new information about progress on this project has come available since 2008, suggesting the project has either stalled for some technical reason…. or because someone in authority questioned the logic of a R&D (plus design) budget of just $33 million (that won’t pay for the coffee!) for a reactor that will cost $300 million to construct!

The proposed Micro Fuji MSR

Even despite these delusions, given the micro-Fuji MSR’s tiny 10 MWe power output that works out at $30,000 per installed kW! or about 6 times the current install cost of PV and 25 times the installation cost of wind energy!

Now granted, it’s a prototype reactor, we’d expect costs to be high, but this is higher than the $24,000/kW price paid by the Japanese for the Monju Fast Reactor, which is generally seen as a white elephant! And of course, the Japanese originally said that Monju would cost a lot less than it did – anyone want to take bets on the micro-Fuji MSR coming in substantially over budget?….assuming of course it ever actually gets built!

Even if a production version of a LF reactor could be scaled up and built for ¼ of the installed kW costs of the Fuji MSR, LFTR’s would still be much more expensive than existing LWR’s (roughly 2 to 6 more expensive depending on who you ask…the fan boy’s would have you believe they would be half the cost of LWR’s…so they’re numbers are off by a factor of 4 to 12!). And indeed LTFR’s would still have installation costs 1.5 times more expensive than PV and over 6 times the price of wind energy, and of course we still have to decommission them, and the current decom costs for graphite cored reactors are high, between $2,600-6,800 per kW.

While the final installation costs of the other reactors we’ve reviewed are “up in the air” somewhat, I doubt any of them, with the possible exception of the GcFR would be this expensive.

8.13     LFTR, the Kool-Aid Fuelled Reactor?

Finally, the LFTR fanatics need to come off the Kool-Aid. I’ve gone to great lengths to debunk many of their crazy ideas because such cargo cult science as they are promoting does a great disservice to science, and gets in the way of more realistic and practical proposals. They also serve to confuse the public, and I mean even Wired News appears to have been taken in by this con, which makes the whole job of real scientists pursing real projects (such as Micro-Fuji above), all that harder. As I’ve shown many of the supposed advantages of the LFTR are simply figments of certain bloggers overactive imaginations. The fact that many of the LFTR supporters are Libertarians, individuals not entirely known for their grasp of basic physics, economics or social norms doesn’t help matters. This is how we see the ridiculous spectacle of some of these guys writing lengthy blogs about Obamacare (which is actually on course to reduce the deficit) and excessive government spending one minute, then turn around and post videos on You-tube  calling for some vast “big government” funded LFTR reactor program!

Of course, the LFTR mob just doesn’t get it. Many just can’t understand why governments won’t fund their baby. They’ve concoct various elaborate conspiracy theories, largely revolving around the fact the LWR’s were/are being preferred for reasons of Plutonium production. As I pointed out in the section on LWR’s a combination of hard headed economics and shear bone idle laziness on the part of the corporations involved were the more likely reasons. As far as the here and now, I can concoct another theory, along the lines of Ocums Razor.

The reason why the powers that be have resisted funding LFTR’s is that they’ve either met or been sent creepy letters (see here sent to Dave Cameron!) by some of these deranged LFTR fans. They have thus concluded that anyone mumbling about LFTR’s is either deluded, crazy or part of some elaborate 5th column scam being run by Greenpeace or the fossil fuel lobby (whose goal is to convince them that all nuclear energy supporters are stark raving mad) and thus make the job of responsible scientists trying to promote such ideas (or other alternatives to the LWR) all but impossible! So like I said, do everyone (aside from the fossil fuel lobby) a favour and stay off the Kool-Aid guys!

Indeed in the short time that this site has been up, various multiple strings of flaming trolls have appeared on pro-LFTR blogs, a good example can be found here, with my reply here …..I would note that the blogger in question sent in the link annoymously…let’s see symptoms of cargo cult….extreme hostility to criticism….paranoia…..wearing of tinfoil hats! The comments section here is a good example, take this guy who apparently seems to have convinced himself that I work for the IAEA! It would be funny if it weren’t so serious.

The ecologists magazine also made the mistake of criticising the MSR reactor. Inevitably an angry mob of LFTR fans soon arrived (see comments on the bottom of their page) and started screaming burn the heretics!

Again, this sort of carry on does the MSR design no favours, indeed its the sort of behaviour that will make most politicians and investors run clear across 6 states to get away from!

8.14     Summary

So in summary we can say that, while there is some promise here of “something better” than a LWR, were a long way from acheiving this.  Any MSR concepts are a long way from anything resembling a working commercial reactor. There are a whole bunch of technical challenges to overcome first. Notably a host of material science issues need to be resolved, top of my list being that issue with intergranular cracking. An efficiently designed and proven processing plant needs to be developed, tested and proven. Resolving those issues with graphite distortion in the core is another problem, as well as guaranteeing the Florine gas releases don’t become a serious issue, plus developing new fire fighting procedures for a high temperature graphite fire (although this last one is a problem for other reactor types too!) are all matters that need addressing. Our experience with large turbogenerator chains running on inert gases at high temperatures is limited, so this is another knowledge gap that needs to be closed (again an issue for several Gen IV reactors and indeed some future fossil fuel and geothermal plant proposals).

Even before the successful conclusion of this R&D, we can definitively say that any MSR would likely be substantially more expensive to build than any existing nuclear reactor (CPP, exotic material use, fire protection systems, containment dome, etc.). The construction rate would also inevitably be slower, due to the dependence on large amounts of exotic materials much of which needs to be welded or cast into odd shapes, joined together and rigorously inspected. MSR’s have some safety benefits, but it also opens up a number of safety loopholes. While probably safer than a LWR (if properly build and maintained) it also opens up a whole host of new safety problems.

Many of the other proposed reactor designs we’ve reviewed would be much safer and a lot easier to develop. While MSR’s might produce less nuclear waste and burn the fuel more effectively, this is entirely dependant on that  CPP working reasonably efficiently and even if it does there will still be a substantial waste footprint associated with MSR’s. Also the poorer efficiency of these reactors compared to the other proposed designs must be considered, as this will lead to increased fuel burn up (so if we can’t get the MSR to work then MSR’s generate more waste than LWR’s not less!). The fact that a substantial portion of the nuclear waste generated from these plants will be mixed in with Fluoride salts also complicates the spend fuel issue, probably resulting in relatively higher spent fuel storage costs, relative to other reactor designs.

Furthermore, one has to question whether, from a economics point of view, the MSR would be more of a hinderance than a help. As far as the power utilities are concerned (at the moment), cleaning up the nuclear waste mess, sourcing fuel for reactors and long term energy security are “externalities”, i.e. somone elses problem. They have thus very little economic incentive to install a CPP that will merely be a drain on their finances.

Decommissioning a MSR? That graphite core issues discussed in the chapter on HTGR’s and the problems mentioned above with Fluorine gas and Hydrofluoric acid all point to potentially much higher decommissioning costs than an existing LWR. The fuel decom costs of the MSRE (a small 8 MWth reactor) are estimated at $130 Million, although its probable the costs from a commercial verison of the plant would be much lower.  You’ll also recall my discussion in the materials section of how some alloying elements can become their highly radioactive “evil twins” after exposure to radiation. If we had not engaged in a lengthy and detailed program of material science evaluation beforehand, then its likely much of an MSR would have to be put into ILW storage, or for parts of the core possibly HLW storage post-decomissioning. So overall, decommissioning costs of an MSR would all likely be higher than with a LWR, although there is uncertainty on that issue.

If the Thorium cycle is your priority I’d ignore the MSR concept altogether as it works much better running on Uranium rather than Thorium, and trying to integrate this fuel into the design would complicate matters. The HTGR has a more proven track record in this regard and would be the more obvious alternative at the moment. While the MSR is scalable, practical matter would probably constrain them in size. Building big is difficult, as manufacturing critical parts out of the relevant materials would be tricky, while building them small would likely require eliminating the CPP and possibly dump tanks, removing two of the reactors unique selling points. This makes medium sized MSR’s the only plausible option (my guess 150 – 1,000 MWth), fixed in place! Mass producing a complex reactor such as this made out exotic materials? Difficult! This of course debunks the remaining myth about the LFTR, that they can be easily mass produced.

But what about Fusion power?

There’s been quite alot of traffic from LFTR fans to this page. The comments can be found at the base of the main page, just follow the link below. Also note that within the comments I’ve included links to various rebuttals from LFTR bloggers and my rebutal to their comments also.

Previous Chapter                                          Main Page                                         Next Chapter