IFR FaD Nuclear

The IFR vs the LFTR: An Exchange of Emails

With regards to Generation IV nuclear fission technology, most of the attention on BNC has been on the Integral Fast Reactor (IFR), for reasons explained in this post, which I quote:

The focus of this series (IFR FaD) is aimed squarely at the Integral Fast Reactor (IFR) rather than other Gen IV designs, such as the Liquid Fluoride Thorium Reactor (LFTR) or Advanced High Temperature Reactor (AHTR). The reason for this is two fold: (i) I’m more familiar with the IFR technology (and I am in regular email exchange with the world experts on this technology, via SCGI and other links), and (ii) LFTR has a strong and welcoming advocacy group elsewhere, and I’d encourage people to go there to ask more questions about that technology … However, I should make it quite clear that I’m not “for IFR and against LFTR” — both 4th generation nuclear designs hold great appeal to me, and I will sometimes consider IFR vs LFTR comparisons in the IFR FaD series, as a point of comparison or contrast.

I think we need to be pursuing the final stages of research, development and commercial-scale deployment of all of these next-generation fission technologies, since it would require such a trivial input compared to the huge investment that will be required anyway in energy infrastructure over the next few decades (>$26 trillion globally by 2030). However, it is nevertheless useful to consider the relative merits of the individual technologies, and I hope to look at this from a number of angles in blog posts during 2012.

For some initial ideas and to initiate discussion, below I reproduce an email exchange on this matter, including aspects of commercial readiness,  that was recently posted on the Science Council for Global Initiatives website. The conversation is from three highly experienced nuclear physicists/engineers, Dr George Stanford, Dr Dan Meneley, and Prof. Per Peterson. I’m sure this will stir some debate! (And, as I said, I will have more to post on this in the new year).

I have also added a few hyperlinks to clarify terms that may be unfamiliar to the general reader; please note that the links and pictures were added by me (Barry Brook), not the original correspondents.


G. Stanford wrote (11-29-10):

We’ll see what others on this list have to say, but in my opinion, Carlsen’s enthusiasm for thorium is premature, to say the least.  The ONLY significant advantage a thorium cycle would have over fast reactors with metallic fuel (IFR/PRISM) is its lower requirement for start up fissile.  That advantage is offset by the fact that the thorium reactor is at a stage of development roughly equivalent to where the IFR was in 1975 — a promising idea with a lot of R&D needed to before it’s ready for a commercial demonstration — which puts its deployment about 20 years behind what could be the IFR’s schedule.  The thorium community has not yet even agreed on what will be the optimum thorium technology to pursue.

I think that thorium should indeed be investigated as a possible future competitor for the IFR.   But what would be gained by putting off demonstrating the IFR/PRISM technology while waiting to see if thorium really lives up to its promise?  Nothing would be lost by getting a fleet of IFRs up and running.  They could be breeding fissile for decades while a possible thorium fleet gets up and running, and the IFR-bred fissile — several times more than was started with — could be used for expanding the hypothetical thorium fleet at the end of the IFRs’ lifetimes.

If the current perceived urgency is to sequester plutonium to put it out of the reach of proliferators, that can be done much faster with early deployment of IFRs rather than by later deployment of thorium reactors — and each IFR will sequester 8 – 10 times as much plutonium (Pu) per GWe as a thorium reactor.

— George S. Stanford 11-29-10

D. Meneley wrote:

On the matter of thorium, George and others have repeated a . . . realistic picture.

[Boosting thorium] will do no good. This is another idealist’s dream, like large-scale wind energy. They only want to save the world and are not interested in practical details.

If you’ve tried to do control, fuel cycle, and safety system design on a thorium reactor you’ll not be so enthusiastic. The flux shape is a strong function of the past flux shape — because of the protactinium. After you shut the thing down you must account for the later reactivity increase. And then there’s the detail of not having any fissile isotope to start up in the first place.

If you’re using thoria fuel, how are you going to extract the U233 economically?

And so on.

Thorium if absolutely necessary, but absolutely no thorium if not necessary.

–Dan Meneley 11/27/10

P. Peterson wrote:


Your assessment on the relative technical maturity of LFTR versus IFR is correct.

But there are other substantive technical differences besides the lower fissile start up requirement for thorium reactors.

Thorium reactors operate in with a thermal spectrum, which allows them to use graphite as the primary structural material in the reactor core.  Graphite can be heated to very high temperatures without losing structural integrity. Combined with the very high boiling temperature of the fluoride-salt coolant (> 1400°C), thorium reactors can deliver heat at substantially higher temperature (between 600 and 700°C with current primary pressure boundary structural materials) than IFR (between 370 and 510°C with current fuel cladding materials).  This is a sufficiently higher temperature that several options exist for gas-Brayton power conversion, while at the IFR temperatures steam Rankine is likely to remain the most practical option.

There are a number of substantive theoretical advantages to gas Brayton power conversion (this is the reason Brayton cycles are now used universally with natural gas), but essentially all of the existing turbine and compressor technologies optimized to open combustion cycles and thus substantive development is needed to adapt it to nuclear power conversion.  Once successfully developed, though, one would expect substantial commercial pressure to move from steam Rankine to gas Brayton cycles as the dominant approach for nuclear power conversion (as has already happened with natural gas).

The other major differences arise from the different thermophysical properties of the two coolants.  The fluoride salts have volumetric heat capacity slightly larger than water and about 4.5 time larger than sodium.  So the primary systems for thorium reactors are physically much smaller than for IFRs, or alternatively, a primary system of the same physical size can produce substantially more power (factor of a 2 to 4).  Thorium reactors have no sources of stored energy that can pressurize containment, so they also can use a compact, low-pressure containment structure and thus a correspondingly smaller reactor building.

These are substantive technical differences that are likely to affect the relative levelized cost of electricity (LCOE) produced by the two systems.   But one of the major issues with LFTR is that one must overcome multiple, substantive technology development problems simultaneously (gas-Brayton power conversion, qualification of materials for corrosion resistance, on-line fuel processing, licensing for fluid-fuel reactors). This creates a significant activation energy problem, even if the final LFTR technology would have desirable LCOE and sustainability characteristics.    One of the reasons that we’ve been working on solid-fuel variants at UC Berkeley is to see if one can reduce the activation energy barrier by capturing most of the LCOE benefits (which come primarily from improved power conversion efficiency and reduced capital cost relative to advanced light water reactors [ALWRs]) while keeping the licensing approach much closer to that used for passive ALWRs and not taking on the technical issues for fluid fuel.

In the end, LCOE will be a dominant consideration in commercial decisions to deploy nuclear power. In the near term the best opportunities involve further improvement to ALWR technology and construction methods (with AP-1000 providing the best role model to date).  In the longer term some mix of uranium fast spectrum and thorium thermal spectrum reactors is likely to emerge as optimal.

-Per F. Peterson 11/29/10

G. Stanford wrote:


Thanks much for the additional information, clarifying the technical challenges and strengthening the case that thorium power is worth pursuing and might well have an important role down the line.

While the LCOE will undoubtedly be an important consideration, it seems to me that breeding potential also is destined to be important if we are to have abundant clean energy.  It also seems likely that thermal efficiency per se will not ba a major issue, in view of the very low cost of fuel for breeders (or “isobreeders” like the LFTR) — but you point out that the Brayton cycle potentially offers significant additional advantages.

I gather that you do not take issue with the proposition that it would behoove us now to complete the development of what is currently closest to commercial readiness with the characteristics needed for an assured indigenous energy supply — namely LMFBRs with metallic fuel and pyroprocessing.  At present, that appears to be a U.S.-developed technology that we have abandoned, bequeathing it to other countries for exploitation.


—  George S. Stanford 11-29-10

P. Peterson wrote:


For LFTRs, the breeding potential may not be particularly important, as long as they can achieve isobreeding.  Uranium from seawater provides a backstop technology that sets the maximum cost of fissile material, much as coal-to-liquids provides a backstop for the cost of oil (absent a price on carbon dioxide emissions).  The startup of an isobreeding LFTR requires about 1/4 to 1/2 the fissile needed to start up an LWR, and uranium from seawater will have a cost around 4 times greater than current uranium prices.  Thus the capital cost for the fissile to start up isobreeding LFTRs will be comparable to the current cost for the initial core loading for LWRs, which constitutes a modest fraction of the total capital cost of current LWRs.

Our experience to date is that “backstop” energy technologies never emerge to be economically competitive, because lower cost alternatives tend to be developed instead (at the scale that we use energy, the economic incentives are very large).

So I would be very surprised that the cost of fissile will ever rise to the point where one would actually begin commercial efforts to recover uranium from seawater (although there is always some slim probability that the government might in the future enact a “seawater uranium portfolio standard,” to create an assured market for seawater uranium, so the technology will be brought to commercial readiness regardless of cost).  Absent such government intervention, the cost of fissile to start up LFTRs will likely remain lower than the cost of fissile to start up current LWRs, in perpetuity.

My expectation is that the LCOE for electricity from ALWRs will drop well below the LCOE for new pulverized coal plants before the end of this decade, as Westinghouse’s costs to build AP-1000’s and enhancements to the AP-1000 drop and as competing LWR technologies for the AP-1000 emerge, and as construction methods improve further. Financing nuclear construction will likely remain a challenge, although SMRs may prove to be helpful in this respect.

But we need an aggressive effort to develop multiple technologies that can improve upon and ultimately replace ALWRs.  Fast-spectrum reactors clearly have advantages, along with thorium cycles, from the perspective of fuel cycle.  IFR metal fuels are vastly better than conventional oxide fuels from the perspective of affordable and secure fuel recycle. LFTR is also a potentially attractive technology, but clearly has substantial technology risk.  So yes, I strongly support demonstration of IFR technology.  The key issue is that IFR needs to remain a part of a portfolio of technologies the federal government invests in, and that IFR demonstration needs to sustain discipline to assure that federal investment is likely to result in commercial success

A simple type of evidence, which Congress has required for the next-generation nuclear plant (NGNP) project, would be 50% cost sharing by commercial interests.    I think that this approach is too simplistic, since it does not recognize how risk changes during design, licensing, and construction of a demonstration reactor.  The best approach is to require very small or zero commercial investment at the stage of conceptual design and NRC pre-application review, moderate commercial investment during detailed engineering and NRC licensing, and substantive commercial investment for the construction of a prototype unit (where the intellectual property and up-side commercial potential ends up being owned by the commercial entities who invest).

This sort of decision framework is also easier to implement in statute, since one can authorize the needed expenditures, but the actual appropriations can depend upon progress being made and commercial investment materializing.

What commercial interests will tell you is that it is much easier to make a decision to make a substantial investment if they have an NRC construction license to build a reactor, while it is almost impossible if the reactor is just a concept that needs a lot of detailed engineering work.  But in the end, the commercial entities that perform this reactor development work are also in the best position to assess its commercial potential–so a lack of willingness to place some commercial money at risk (less earlier and more later) should be viewed as evidence that the concept needs more R&D, not accelerated demonstration.

For IFR, though, the availability of affordable fuel is a big issue.  It requires the capacity to recycle used LWR and IFR fuel, as well as to test and qualify recycled fuels for use in IFRs.  This is a problem that the commercial sector is not going to be willing to take on, and thus it requires purely federal effort.

-Per F. Peterson 11/29/10

George Stanford wrote:


Thanks for the further elucidation.

To get quantitative about LFTRs, suppose the world were to want 50,000 GWe of isobreeding LFTRs by 2100, primed with 10% EU at 1 tonne of U-235 per GWe.  That U-235 would be contained in 10 tonnes of EU, which would come from ~200 tonnes of Unat.  Thus the amount of uranium to be mined would be ~ 50,000 x 200 = 10 million tonnes of Unat — which is well within the realm of the possible.

However, a downside would be the perpetuation and global expansion of uranium enrichment infrastructure, with its proliferation implications.  Also, left over would be some 9.5 million tonnes of orphaned depleted uranium containing 9.5 million GWe-years of unavailable (and unwanted) energy.

—  George S. Stanford 11/30/10

CANDU reactors in Qinshan, China

D. Meneley wrote:

George, Per:

You guys seem to be intent to ignore the only fully developed, in service, economically competitive, high conversion ratio, and safe medium size reactor system on earth [namely, the CANDU-type heavy-water reactor]. Perhaps you could explain why.

Dan Meneley 11/30/10

G. Stanford wrote:


Sorry if I slighted the CANDU. It’s a fine reactor design, struggling to acquire a bigger share of the international market. To believers (like you and me) in the importance of conserving fissile material, its high conversion ratio is an important asset. But – apart from the fact that it doesn’t need enriched uranium — to a first approximation it’s just another thermal reactor. Since most of the generalities about LWRs apply also to CANDUs, much of the time it’s convenient to use the term “LWR” as shorthand for “uranium-based thermal reactor.”

While I can’t speak for Per, of course, I suspect that he would ascribe less importance to a high conversion ratio than I do.

George S. Stanford 12/01/10

By Barry Brook

Barry Brook is an ARC Laureate Fellow and Chair of Environmental Sustainability at the University of Tasmania. He researches global change, ecology and energy.

137 replies on “The IFR vs the LFTR: An Exchange of Emails”

The emphasis seems to be on MSRs. It appears to be largely backed by the Cheks, given the preponderence of Chek presenters. I became awate of it through the media. The registrations have closed now, but I was wondering if anyone else had arranged to attend.


Here’s something I wrote about a year ago:

Here are some of the advantages/disadvantages relative to IFR (Integral Fast Reactor).

LFTR is better than IFR because:
* better coolant:
* chemically stable liquid salt instead of liquid sodium which reacts violently with air or water
* higher heat capacity
* 1/5th the fissile load per megawatt
* Liquid fuel
* fuel integrity cannot be damaged by radiation and not subject to fatigue or pressure failure
* fuel allows continuous removal of Xenon so no startup transient poison
* suitable for continuous online reprocessing for fission product removal
* safer
* minimal geometric configuration, so cannot become super critical through an accidental reconfiguration
* already fully moderated, so cannot become super critical through accidental moderation (like core becoming physically close to materials containing hydrogen such
as concrete or water during an overheat meltdown accident)
* can be designed with no excess reactivity and continuous online refueling
* thermal spectrum operation makes it much easier to control as all operation is below nuclear resonances
* burning Th232 produces ~1% of the long-lived higher actinide waste compared to burning U238 (the fertile fuel starts with 6 fewer heavy nucleotides)
* lower breeding ratio so easier to control the proliferation of reactor operators by controlling access to startup fissile material

IFR is better than LFTR because:
* has been much more thoroughly studied and funded
* burns U238 which is very widely stockpiled
* burns spent PWR fuel which would be nice to get rid of
* high theoretical breeding ratio (1.8 vs 1.3) so more reactors can be started up more quickly — this was true, but now the world has so much stockpiled bomb plutonium that this may no longer be a practical limitation
* U238 is more available from seawater than Th232

Ultimately I think a mix of a few fast-spectrum reactors (IFR or chloride salt fast MSRs?) and a much larger number of thermal spectrum thorium burners with breeding ratios just under 1.0 are probably the safest and easiest to control. But I think that starting with the harder fast spectrum design is more expensive and less safe. But of course, I’d rather build IFRs than just sit around while we wreck the atmosphere.

Some would argue that we can only afford the resources to develop one technology branch. I disagree. The amount of money required to develop a technology is quite small compared with the cost of 20–50 years of the world’s energy consumption. The real problem is an under-investment in technology development combined with the stifling effect of draconian regulation. The main advantage (from my perspective) of different nations pursuing independent nuclear technology development programs is that some of those countries are relatively free from excess regulation. Others might argue that this lack of regulation will result in radiation spills or nuclear weapons being used and that is even more unacceptable than systemic climate change. In my opinion, those dangers are overstated and those priorities are inverted. Let’s build all of these:
1. more Gen III LWR like AP-1000
2. Advanced high temperature, salt cooled reactors like Per’s AHTR
3. thermal spectrum, liquid fluoride fuel Th-232/U-233 reactors like Kirk’s LFTR
4. fast spectrum, liquid chloride fuel U-238/Pu-239 reactors
5. and even fast spectrum, liquid metal cooled, solid metal fuel U-238/Pu-239 reactors with integrated reprocessing like IFR.
Then we can stop worrying about energy costs and damaging the environment. We can do all this for far less than 1% of the cost of the energy we will consume over the next 20 years.
Chris – please be aware that BNC requires substantiation of personal opinion by the use of refs/links etc. Please supply these.


Chris Uhlik, what does “safer” mean? Is there really a need for more technical safety beyond a certain standard?

Cultural or operational safety is just as significant, IMHO. A workforce which manages complexity or the unexpected safely, which adapts to change safely and does the mundane safely is capable of achieving far higher standards of safety than simply relying on engineered safety systems.

I am not saying this to discredit the principle of continuously improving safety programs, only to suggest that the law of diminishing returns suggests that once the engineering design is appropriately safe, the biggest issues relating to safety will not be improved by further design.

On re-reading what I have just written, I realise that it is a bit clumsy, so I’ll try another tack.

If a technology is safe, and by this I mean really safe, as in fit for purpose, it matters not to me what it is called – IFR or LFTR or whatever – it is adequately safe. More layers of redundant designed-in safety systems suggest to me overdesign and unjustified waste, the cost of which would probably be spent much better elsewhere in the community, perhaps on mental health initiatives or eliminating malaria or something else which has statistically significant personal cost. I’m loathe to put a price on human life, but expenditure must be justified by anticipated returns or we will all end up joining the Helen Caldicott Party.


+John Bennetts, I meant “safer” in the sense that traveling on the ground is safer than flying. It might not be true for a particular situation (e.g. flying a Quantas 747 may be safer than walking through a jungle) but in general flying involves high kinetic energy and weather which can cause dramatic problems. A low pressure, inert coolant has fewer ways to go wrong and will likely go wrong in a less dramatic fashion than a high pressure steam system or a liquid sodium metal system. That said, any of those systems can be engineered to any properly specified level of safety at some cost. Just as the modern jet transport is highly refined and quite safe. But it takes many years of experience to get there, and more years are required for systems that are intrinsically more prone to energetic phenomena.


John Bennetts — Well stated as I have come to expect from you. Current Gen II NPPs are about as safe as eating peanut butter, so additional safety is less an issue that simplicity of desing which ordinarily lowers cost. For example the NuScale Gen III unit, soon forthcoming, requires less concrete and less steel per kilowatt; lower capital cost per kilowatt. But because of its convective cooling desing (which I find quite elegant) it is also about 10–100 times safer than eating peanut butter; fine, a useful sales point.


> BNC requires substantiation of personal opinion by the use of refs/links etc. Please supply these.

I said many things that I think are self evident. Please ask a more specific question and I’ll back up that statement.

Perhaps some confusion is due to the posting being stripped of leading whitespace. This makes my outline look more like a list.


I regret that I did not answer George Stanford’s message of 12/01/2010, earlier. NO, the CANDU is not “just another thermal reactor”. The difference lies in its high internal conversion ratio, and its ability to operate primarily on a mainly-thorium cycle. China and India both see this difference very clearly.

It is obvious that the world must build many, many thermal reactors very soon, to overcome the serious consequences of our over-dependence on coal and oil. Then, we will need to pay attention to the shortage of available fissile materials. This leads us directly to the need to build thermal reactors with high internal conversion ratio, in order to reduce the demand on enrichment services as the world nuclear fleet grows into thousands of units. (See “Transition to Large Scale Nuclear Energy Supply”, at , under Members Views).

There is only one fully developed nuclear energy system now in existence that can satisfy this need for a thermal reactor that can prepare us for the coming transition to a sustainable long term energy system based on uranium and thorium. It is the CANDU system.


I’m planning to go to the Canberra symposium next week. Looks like the initiative came from one of the miners, but there are some pretty heavy actors on the program, including the minister, the deputy head of the department, and the head of ANSTO. It will be interesting to hear what the Czechs are up to.


@ P. Peterson 11/29/10:

This paragraph caught my eye

My expectation is that the LCOE for electricity from ALWRs will drop well below the LCOE for new pulverized coal plants before the end of this decade, as Westinghouse’s costs to build AP-1000′s and enhancements to the AP-1000 drop and as competing LWR technologies for the AP-1000 emerge, and as construction methods improve further. Financing nuclear construction will likely remain a challenge, although SMRs may prove to be helpful in this respect.

Could you provide any links about the cost projections.

@ John Bennetts, on 17 November 2011 at 3:24 PM

Here, here to that comment.


I see how sodium coolant is always listed as a disadvantage of the IFR system. I can see that point. If not threatening to the reactor itself, sodium fires have been a nuisance at fast reactor prototypes all around the world, leading to shutdowns and increased cost of operation.

Why not switch to lead? Lead has already been used in fast reactors in the Soviet Union and it doesn’t react violently with water or air.


Barry, George, Dan, Per,

I have little understanding of the alternative Gen IV’s and it is not my focus. However, I would like to see a simple table with the most likely Gen IV contenders compared on the most important parameters. Some important parameters that come to mind are:

1. A realistic estimate of LCOE (by a defined date).

2. Most likely date by which their LCOE could be competitive with Gen III+, recognising that the LCOE of Gen III will decrease over time

3. Construction duration

4. Plant life

5. Suitability for use in under-developed and early-stage developing countries

Any other parameters that are relevant regarding commercialisation and cost. I intentionally did not include “safety” as a key parameter. As far as I am concerned, any Gen IV that gets through licensing will be safe enough, although I accept it will have accidents just like any other industry. So the safety is actually included in the LCOE.

Can either IFR or LFTR do all we need alone or will we need both?


@Max: I can just imagine the outcry from the anti-nuke campaigners the first time there’s a molten lead leak… (never mind the phenomenal amounts of lead emissions from the various smelters around the world!)

I thought I remembered that there are chemical problems with using molten lead as a coolant (affecting steel alloy strength), but Wikipedia tells me it’s been done since the early 70s, and commercial designs are on the drawing board. I must have been thinking of some other coolant…


One of the questions about IFR technology is their likely breeding ratios. Oral claims of 1.65 breeding ratios have been made, but this how much R&D is required to get there is unknown, and there are grounds for doubting the safety of a high breeding ration IFR.

A better approach might be a lower breeding ratio IFR which offers both Uranium and thorium fuel cycles.


Considering thorium’s breeding process as — Th 232 +n -> Th 233 -> Pa 233 -> U 233

The long half life of Pa 233 (27 days ) must put a significant constraint on the design options for a thorium reactor. In order not to lose too much Pa 233 to neutron absorption as U 234, the probability of neutron absorption by Pa 233 has to be less than the probability of its beta decay to U 233. That could be arranged, either by having a low neutron flux across a large core, or by having a long resting time for fuel removed from the core.

At first glance, it would seem that thorium reactors have to be big to be self-sustaining. However, neutron economy could be augmented by the addition of recycled plutonium , which would be one way to burn the stuff without breeding more plutonium. It might also allow the design of small “thorium” reactors.


An IFR could be lead-cooled as well. Lead coolant would be more acceptable to the public, since the fire hazard of sodium is often the No.1 argument brought forward against the deployment of fast reactors.

Or is there something about lead coolant which makes it less desirable? Is it more difficult to achieve high breeding ratios with lead or why are there so many sodium-cooled designs?

It can’t be material issues, since lead-cooled systems have already been successfully demonstrated as mobile power reactors on Soviet submarines.


I found something:

Click to access SodiumCoolant_NRCpresentation.pdf

Sodium vs LBE vs lead vs helium

Outlines why sodium is the preferred coolant.

I definitely don’t want to see helium-cooled fast reactors. Sounds like a recipe for disaster. Apparently they aren’t even sure whether natural helium circulation would be able to remove the decay heat of a fast reactor in shutdown when all power to the pumps is lost. They would have no safety advantage over present-day light water reactors.


With the best will in the world, chaps, this is a misplaced, and essentially academic argument. We have very much more urgent needs.

The argument is one of priorities, and of what is needed for the sector at this moment.

Right now is probably (Fukushima aside) probably as favourable a moment as nuclear has or will see for years. Interest rates are low, the decarbonisation agenda is prominent, uranium supplies are plentiful, and we have on the stocks a number of designs that have the potential to be built quickly and consistently. I’m not willing to bet things will look half as favourable in a decade or two.

Counter to that we have to acknowledge that (from an investors perspective) that outside of France and Asia, the nuclear industry’s track record on delivery stinks. Plants arrive years late and heavily over budget, and even when eventually delivered take years to get to a point where they perform to design specs (look at US capacity factors from the 1980s and the present day).

A big contributor to that, in the US and most of the world, has been the complete failure to settle on standardised designs, and then “productionise” them – think six different BWR generations with God knows how many variations within each, for example. Think of the plethora of demonstrators and prototypes.

That is NOT the hallmark of a mature industry.

And we should be a mature industry. It’s 55 years since Calder Hall and its Russian equivalent started supplying power to the grid, 53 years since Shippingport started up. In the same timescale aviation went from the Wright Flyer to the Boeing 707.

To use that analogy to the aviation industry, it’s as if after building the first dozen 707s, Boeing radically revised the design producing something else with no common parts, different flight controls etc. Oh, and to make it worse, that redesign necessitates a whole new regulatory licensing process.

That – in the eyes of the people who take the decisions to buy and fund new plants – is roughly where we are. Worse, unlike the aviation industry, we’ve no track record of eventually getting the situation stabilised then being able to give reliable dates, prices, performance and cost figures and so on. Instead, we’ve got the Olilkuoto/Flammanville debacle setting us back even further.

That’s why the US industry has needed a loan guarantee programme. Loan guarantees don’t change the long term economics of plant, but what they do is protect lenders and investors against the risk of companies buying the plant going into default because they’ve gone bust – as a result of delays and over-runs. It lowers borrowing costs by removing the risk premiums. But that’s largely failed, because the calculation of the “insurance premium” based on past experience has been horrible (the fact that the same logic hasn’t been applied to renewables isn’t relevant).

So, what at this stage would be a smart response? Is it to launch on yet another round of demonstrators, prototypes and radical new technologies, which will inevitably in their turn hit delays and over-runs (think B787 and Airbus 380) and worsen an already bad reputational problem?

And especially is it smart for the enthusiasts for some of the new technologies to argue that the current choices are fundamentally misconceived, to imply that they’re inherently unsafe or to argue on the basis of fuel economy – when there’s no realistic prospect of uranium shortages for a half-century or more?

After all, no-one realistically regards any of the GIII+ designs as anything other than extremely safe; and they can all happily run conventionally recovered MOX should those putuative uranium shortages manifest themselves.

Or is the smart choice to knuckle down and focus the limited available resources and capital on showing that the current designs can be delivered reliably? Even better, to concentrate on the learning curve, economies of scale and incremental improvements to those designs, and deliver them cheaper, faster and with better performance?

To put it another way – to establish investor confidence we need a dozen AP1000s consistently delivered on plan, and getting down to 36 months construction time a lot more than we need an IFR prototype that might be commerically competitive in 30 years, if uranium prices skyrocket. We need eight or ten EPRs up and running, and maybe testing burnable poison fuel (so they can run for four years without refuelling) more than we need yet another an HTGR demonstrator.

Let’s return to the Boeing analogy. Imagine they’ve built the first dozen 707s – they’ve been late, expensive and it’s taken several years in service to get each one to be reasonably performant. However, most of the bugs have been worked through , and we’ve got a manufacturable and operable version just about ready to go. Boeing wouldn’t now turn around to Government, customers and investors and say, “we’ve thought of something much better – a blended wing design. So we’re going to focus our R&D effort on that on that. Oh, you can forget all the investment on the 707, that’s old hat”. I’d expect the CEO – rightly – to last about 5 minutes….

The sane thing to do instead to do what the GIV consortium is actually doing. Focus on lab scale R&D and paper studies with the possible deployment of pilot plants in the 2030s, at which point there MAY be some indications of the fuel issues that would make thorium or fast reactors attractive.

In the meantime, concentrate on getting the GIII+ designs really working, and then (only then) exploit them more. Get build times down to a consistent 36 months. Demonstrate availabilities of 95%+ within two years of start up, and show economically attractive load-following. Then we might see nuclear utilities with the same cost of capital as non-nuclear ones – that alone would make nuclear hugely more competitive. At that point, think about a stretch the AP1000 to 1400MW and the EPR to 1800 or 2000MW for the same unit prices as they cost at the moment, or less, or extending refuelling intervals.

Once that’s been done, and only then should the industry will be in a place to look for funding for radical alternatives
This comment is off topic on the discussion of IFR vs LTFR. A general opinion on the state of the industry,financial aspects etc should be on an appropriate/or Open Thread. Further violations may be deleted.


I believe Andy Dawson is right when he says the industry needs to concentrate on delivering a reproducible design. That said, I believe the problem (delays and cost overruns) is rarely due to design issues, but rather regulatory issues. Not licensing issues, but delays due to inspections, changing requirements (regulatory ratcheting), protest movements, non-cooperative state governors, political maneuvering, etc. The aircraft industry has enjoyed the ability to make continued progress while routinely killing hundreds of people while the nuclear industry sneezes a puff of tritium laced steam and the whole world freaks out. To stretch the analogy, suppose a 707 knocked out a runway edge light during a crosswind landing; the fleet gets grounded while the regulatory agency demands that the airplane be redesigned to eliminate the possibility of landing more than 1/2 centimeter from the runway center line. Then, perhaps worse, the industry responds with an improved (but more expensive and time-consuming) new design that actually results in more accurate landings, but at twice the cost and all planes under construction are torn down and rebuilt to incorporate the new features.

This is the aircraft industry’s equivalent to Three Mile Island

A significant system failure where nobody died and the passengers cheering the pilots. The A380 still flies every day yet undamaged reactors in Japan remain shut down and almost nobody is cheering the Japanese for their magnificent and largely successful effort to safe their reactors following the largest tidal wave in 100 years — perhaps roughly equivalent to landing a plane with half a wing blown off without killing any passengers, but severely damaging some airport buildings and leaving the plane a wreck.

“Currently, 43 of Japan’s 54 nuclear reactors are shut down, either because of mechanical problems or routine inspections, which must be conducted every 13 months. Local approval is required to restart nuclear power plants, even after routine inspections, and local leaders fearing repercussions at the polls have been loath to provide it.” — These are not new designs. These are existing plants that were operating safely, but are not being allowed to be restarted out of fear. The nuclear industry is hobbled by fear, not by bad design decisions.
See my comment to Andy.


Here in one the west coast of the USA, our utility PG&E gets 10.5% guaranteed rate of return(*) on their capital investments. That sounds pretty sweet doesn’t it? How much would you like to spend on your next powerplant if you were guaranteed 10.5% ROI for 30 years? Answer: as much as I can get away with. So who cares how much it costs to ensure the safety of the people living near the powerplant. The utilities have a strong incentive to add every conceivable safety feature at any allowable cost. Think of the children!

(*) They asked for 13.1% and were allowed _only_ 10.5%

Click to access UE-88-Remand_Exhibit_6600_Blaydon.pdf


At best the LFTR seems many years away. However within just a few years Australia will be producing more thorium than uranium from monazite sands and rare earth by-production. This suggests using CANDUs as a ready-to-go application for thorium. What I don’t know is how much more energy can be extracted from thorium using LFTRs as opposed to CANDUs and reprocessing if that is applicable. My questions are
1) are LFTRs worth the wait?
2) what do we do with all that thorium?


Max — The Russians have a ‘working’ lead or lead-bismuth cooled design. The Hyperion will be similar, just much smaller.

I suppose these are therefore Gen 4 although I fail to understand just why.


LFTR is a good design for the long run, but it is the wrong design for build out in the next 50 years. The first generation MSR should be the simplest possible uranium fueled reactor. It has several advantages.

1… No continuous online reprocessing. Just replace graphite and clean the salt at 30 year intervals.

2… Uses 1/4 the uranium of conventional reactor. Fuel cost is negligible for the foreseeable future. Next generation reactors do not need to be breeders.

3… Shortest development time and cost in a well funded engineering environment unconstrained by political and emotional constraints. It would be a scaled up version of the technology demonstrated at Oak Ridge in the 60’s, using similar materials running a steam cycle at relatively low temperature.

4… Easiest and lowest cost to mass produce due to compact size and lack of complex safety systems.

5… Highest level of safety without complex active safety systems. Volatile fission products are extracted as they are produced and converted to safe chemical forms that can easily be stored and passively cooled without risk of meltdown or criticality.

With molten salt there is no large bolus of volatile fission products trapped in fuel pellets waiting to be released in an accident with fuel melting; for example, cesium atoms quickly react with fluorine ions to form cesium fluoride, boiling temp 1251 C vs. 671 C for elemental cesium.

Machines designed for high temperature operation are easier to passively cool than machines containing low melting temperature materials like solid metal fuel.

Liquid fuel can be redistributed into a geometry that is critically safe and easy to passively cool while solid fuel reactors must be protected in their existing geometry.

Click to access TEAC3_LeBlanc_David.pdf

This technology can produce abundant supplies of reliable dispatchable energy at a much lower cost than burning fossil fuel in the shortest time. KWh’s from simple MSR reactors will be available sooner and at lower cost than kWh’s from LFTR or LMFBR.


Bill Hannahan, on 18 November 2011 at 9:19 AM said:

3… Shortest development time and cost in a well funded engineering environment unconstrained by political and emotional constraints.

How can you have “well funded” and “unconstrained by political and emotional constraints”.

Wasn’t that supposed to be the case for EBR-II, and what happened to it?


It doesn’t matter much which nuclear fuel we burn for the next several hundred years. What matters is how hot it burns. Simple, set and forget hot heat will enable nuclear to have sufficient mass-market penetration to make a meaningful dent in Global Warming – the goal of the BraveNewClimate web site.

All conventional LWR reactors run about 550F. The IFR comes in at about 900F. But it, in turn, is outclassed by the 1,300F LFTR.

Hotter is better when it comes to heat. The 910F BN-800 IFR is an excellent example of what happens to an amply large reactor – with the most excellent of development pedigree – that still is neither fish nor fowl in the temperature world.

There are excellent reasons steam temperatures are what they are. Once over 1,100F, you are in a position to make world-standard 1,005F superheated steam, once over 1,200F, supercritical and ultra-critical steam. At 1,300F, you are in Stirling air turbine territory. These applications, in the 1,200 largest of the world’s 30,000+ fossil fuel power plants are where more than 30% of ALL Global Warming is coming from now.

Today’s nuclear reactors have become huge temples of technology moving at all deliberate glacial speed. The IFR is likely to be a continuation of that paradigm. Not a good omen for quickly reining in Global Warming.

A pot, a pump, and pipe was someone’s quip about the omnivorous basic converter type of thorium-fueled molten salt reactor they experimented with at Oak Ridge Laboratories.

I think after what we’ve been through, much simpler, much hotter, and much quicker is what the World needs. Let’s take the LFTR back off the shelf.


Jim: Hotter is better? No, that is wrong. Cheaper is better, any day. Pressing materials into service at their extreme limit of tolerance is not only unsafe, but it usually is a poor economic alternative. Cheap fuel — natural uranium — at a reasonable temperature is a much more attractive choice.


Andy Dawson @ 3:46: You certainly make many good points, but allow me to comment on a couple. You write: “After all, no-one realistically regards any of the GIII+ designs as anything other than extremely safe; and they can all happily run conventionally recovered MOX should those putuative uranium shortages manifest themselves.” MOX is a non-answer. It increases uranium utilization from 0.6% to 0.8% (whoopee!), costs huge amounts to build and operate the plants translating into fuel vastly more expensive than fuel made from virgin uranium, and can be recycled once, leaving you with virtually the same spent fuel problems we face now (which, admittedly, aren’t that big a problem except in the eyes of the uninformed public and policymakers).

You also write: “… to establish investor confidence we need a dozen AP1000s consistently delivered on plan, and getting down to 36 months construction time a lot more than we need an IFR prototype that might be commerically competitive in 30 years…” The very first two ABWRs were built in Japan in 36 and 39 months and worked fine. This won’t be a problem. Your contention that IFRs “mmight be commercially competitive in 30 years” is a baseless assertion. And good luck getting ten EPRs up and running. The AP-1000 and ESBWR are going to eat AREVA’s lunch.

Your “sane thing to do” reasoning that IFRs and other should be put off until we finally build pilot plants in the 2030s completely misses the point on many of the IFRs strengths: It will be small, mass-producible, easy and cheap to fuel (with the fuel itself being better than free aside from the easy and inexpensive fuel fabrication) and will operate as a non-pressurized system. Once the first pilot plant is completed they could be rolled out very quickly. The only limitation would be on fissile startup inventories.

The IFR is hardly a “radical alternative” except in the sense that it can revolutionize power production around the globe. It’s not like fast reactors haven’t been built and operated for decades. While I agree that Gen III+ reactors should be built, turning one’s back on the IFR just because you’re building III+ is like saying you can’t walk and chew gum at the same time.


To reinforce Tom Blees point, I have no issue with seeing a dozen AP1000s delivered on plan — indeed, that is happening in China now. If “we” is the US, and they cost $6 billion per AP1000, then that’s $72 billion for the AP1000s plus another say $10 billion to invest in getting BOTH the IFR and LFTR to commercial ready status (1/7 the cost of the 12 AP1000s). Putting that entire expense in the context of the estimated $26,000 billion required globally in energy investments by 2030, and we’re talking peanuts. That’s the point. Let’s do both, and stop pretending that crises can be solved by taking on only one issue at a time.


Oh, and following the moderator’s advice, let’s move further discussion of where nuclear investments ought to be directed to the Open Thread. This is a post about the IFR and LFTR comparisons. Not Gen IV vs Gen III.


Just a word about “hotter is better” and Per’s comments about Brayton vs Rankine cycles: The most recent developmental work with supercritical CO2 heat exchanger systems will very likely result in allowing the PRISM (or other systems) to increase its power output by about 50% and switch to the Brayton cycle. This has been worked on at Sandia and is pretty close to the point where these much smaller and efficient systems will be deployable—almost surely by the time the first PRISM is built. So you’ll be able to have a very efficient 550C system (and, by the way, eliminate the sodium heat-transfer loop, which I suppose will make some people more comfortable).


Moderation is not available 24/7 so please be aware of which thread you are commenting on and avoid being led by a commenter who posts off topic. It is important to keep on topic to avoid rambling discourse which deflects from the current discussion. Unfortunately BNC does not have the facility to re-post comments on other threads so deleted off topic comments must be re-posted by the individual.


The 1982 Nuclear Waste Policy Act is a problem for IFR and LFTR development. The Act prohibits the reprocessing of spent nuclear fuel (SNF) and the transport of SNF off site. This severely limits what private parties can do to further the technology of either system. Basically, no progress without being a government contractor and complying with stifling rules. This makes the technology development orders of magnitude more expensive than almost any other area. Smart, energetic engineers and scientists in the USA and many other countries look elsewhere for fulfillment.

In this regard, some of the technologies of LFTR (salts and coolants and chemical processes) look like they might be less prone to restrictive control than for example casting plutonium alloys.


+Tom Blees writes:

> The most recent developmental work with supercritical CO2 heat exchanger systems will very likely result in allowing the PRISM (or other systems) to increase its power output by about 50% and switch to the Brayton cycle.

Here’s a good source for supercritical CO2 as a turbine working fluid.

Click to access 2906955.pdf

CO2 has a critical point at 31.5 degrees C (88.7 degF).
There are two main advantages in using supercritical CO2 as a working fluid:

1. the fluid is relatively dense compared to steam, helium, nitrogen or air at feasible pressures. This minimizes the volume of the machine and allows high power density. A supercritical CO2 turbine will be tiny compared to an equivalent power condensing steam turbine in which the output is nearly a vacuum at ~70C.

2. when compressing the expanded CO2 back to high density, ***if the heat is taken out below the critical point***, very large density increases are available without corresponding increases in pressure. This greatly reduces compressor work compared to more ideal gases. Steam also enjoys this advantage, because the pressure increase (pumping) is done after condensation so the volume change is small. Helium, air, and other more-ideal gases without phase transitions are at a relative disadvantage here.

However, this compression advantage of supercritical CO2 only accrues if the compressor is multi-stage with intercoolers at every stage. These intercoolers need to deliver relatively cold CO2 to the next compressor stage. “Cold” means “below the 31.5C critical point” with 28C being typical of simulations; see the referenced paper. 28C is not easy to achieve on a warm day. Consider that many very high power industrial heat exchangers use a nominal temperature drop of 20–50C. The size of the heat exchanger increases rapidly as one tries to reduce this temperature gap. The super critical CO2 reject-heat heat-exchangers contain high pressure (~1500 psi) and will be very large and expensive. Perhaps the only way to get 28C compressor inlets in most parts of the world will be to use cold sea water on the cold side of the HX. If the sea water is ~8 degrees C, you might manage a reasonable 20 degree C HX drop. Note that the heat exchangers in one of the ORNL thorium MSR designs (ORNL-4528) are the largest parts of the reactor system and they allowed 190F (88C) and 125F (52C) drops. (pages 40 and 55)

In my opinion, the excitement about super critical CO2 turbines is missplaced. The physics of real world heat exchangers and the availability of low temperature heat sinks limit the utility of a working fluid with a 31.5C critical point. I’m saying it’s not gonna happen.

Bringing the subject back to IFR vs LFTR, the higher temperatures allowed by liquid salt coolants allow high thermodynamic efficiency to be achieved using off-the-shelf (600C) ultra-super-critical steam equipment.



High temperatures at the heat exchanger makes for more efficient heating of the working fluid. That increased efficiency allows a shift from the use of steam to the use of air to drive turbines. When they talked about a Brayton cycle in the post, all that was implied was a hot gas turbine. Tom commented that a higher temperature of the PRISM implied “much smaller and efficient systems”, that is, as a hot gas turbine. The simplest would be an open cycle hot air turbine. Ambient air is filtered, compressed, passed through the heat exchanger, through the turbine and dumped directly into the atmosphere.

There are distinct advantages for the expansion of energy delivery systems away from the main grid. For one thing, there is no condenser, cooling system, cooling tower or public eyesore to attract public resentment. Indeed, the nuke and its turbine are likely to be truckable. It is also significant that the power plant doesn’t need to be sited near water. That means that the town or industry it serves does not have to be on a watercourse, so industry can expand away from the existing conurbations along the coast and rivers. Deployed away from the main grid, the small air-turbined nuke would save on the need for transmission lines to connect these consumers to the main grid.

Further, being so transportable, the power plant can be rapidly deployed. As climate change advances, we must expect that the industrial and urban environments will change, possibly rapidly. We need an energy system that can be put in place as fast as we can move our homes. Similarly, we could dismantle such a unit rapidly. Depending on the design, many and perhaps all of its components might be moved from one site to the next with good fuel still in the core. (Would even an MSR be so mobile?)

Fans of giant nukes can point to their superior fuel efficiency, and in the current marketplace, their superior capital cost per power unit. However they are giant things to plan for, requiring gigantic funds to be found, suffer giant costs when delays or disruptions falter construction, and make century-long assumptions about a fixed climate, fixed public values and a fixed consumer base.

Mass production can make small, deployable nukes cheap enough to compete. Hot air turbines can make them rapidly deployable, anywhere. Climate change will require such adaptability.


Dr. Peterson has pointed out some thermophysical advantages of molten fluoride salt coolants: high volumetric heat capacity (5x sodium), high boiling point (>1400 Celcius), electromagnetic transparency (easy online inspection), low reactivity with air and water (some reactivity with water but nothing like sodium’s explosive behaviour).

I don’t think it is testiment of great in depth knowledge from Stanford, to state

The ONLY significant advantage a thorium cycle would have over fast reactors with metallic fuel (IFR/PRISM) is its lower requirement for start up fissile.

Any serious research will reveal, that thorium metal has a higher melting point and superiour mechanical strength moduli, and a higher non-phase transition temperature than uranium metal. It is very much a superiour fuel compared to uranium. It does not make plutonium that can be difficult to reprocess without losses, it makes uranium (U233) as fissile fuel that is much easier to do with fully automated reprocessing. Fast spectrum means U234 absorption product and Pa233 intermediate fast fission very well. There are many other advantages but I don’t have time for that right now, I just want to point out that there are many advantages to thorium fuel. People who are involved in one research area can have little knowledge or patience for other research areas.

I would like to make a suggestion for a combined research effort between LFTR and IFR and suggest a fluoride cooled IFR with thorium metal fuel. Possibly started with plutonium from spent nuclear fuel.


Any serious research also will note the awkwardly high melting point of molten fluoride salts. Reactors are not always operating — and you’ll have to have a very good heat source to keep your coolant in molten state.


@ Dan Meneley

Could you please explain the apparent contradictions in your various e-mails?

You forcefully disparage the use of thorium reactor technology in your first e-mail, ending by stating: “Thorium if absolutely necessary, but absolutely no thorium if not necessary”.

Later, you move on to champion the CANDU reactor and take issue with its having been lumped in with thermal LWRs. You state that the CANDU’s great advantage is that can operate with mainly thorium fuel.


Douglas: India and China are both rich in thorium resources, but very poor in uranium. These countries also are very poor in petroleum resources. For these reasons, thorium is a necessary part of their search for energy independence. I expect they will make use of it even though it does complicate operations.

A very well thermalized spectrum such as that existing in the CANDU design is the best for thorium utilization because of the resulting high internal conversion ratio.

I have never “disparaged” the use of thorium, but I do insist on presenting both the positive and negative aspects of its utilization.


@ Cyril R

“I would like to make a suggestion for a combined research effort between LFTR and IFR and suggest a fluoride cooled IFR with thorium metal fuel. Possibly started with plutonium from spent nuclear fuel.”

Could you elaborate? You are, I think, implying a fast spectrum. What fluoride salt combination would you recommend? What vessel lining material would you use? What delays would be incurred in qualifying materials? I like the principle, but not the possible lost time.


Great discussion. The NRC coolant comparison linked upthread is informative and shows that, for a number of reasons, sodium makes an excellent coolant.

It seems to me that the perceived safety threat of sodium’s reactivity is sometimes played up a little too much. It does react with water, which is why there is a separate sodium loop for generating steam. Sodium leaks to the air are apparently not a huge problem, and leak occurrence itself isn’t a big concern since the reactor doesn’t operate at high pressure. So these issues are not insurmountable, and seem to be about on par with sodium’s opaqueness as an engineering problem.

Lead is *really* heavy, the activation products have a long half-life, it’s difficult to keep it in a molten state, and there’s much less experience with it as a reactor coolant.

One thing that’s not clear to me about the LFTR: what kind of neutron source is used to get the reactor going?


Dan Menely: the molten salt reactor experiment worked fine, they used heat tracing on small piping and put the entire reactor in an oven. No lack of heat source after operating a few weeks – lots of fission products that make heat. Only startup that is tricky. But again the MSRE had no major issues… using 1960s technology. I’m currently modelling a design that uses a pool of buffer salt that keeps the entire primary loop immersed in hot salt, preventing freezing and allowing easy passive decay heat cooling.

Douglas Wise: the spectrum of the fluoride salt cooled IFR would be fast, but not as fast as the sodium cooled IFR due to inelastic moderation of fluorine. This is possible with thorium, which can breakeven on breeding in any spectrum. It is also likely possible to breakeven on breeding using only uranium and a fluoride coolant.

The coolant would be based on NaF with either BeF2 or RbF as secondary to lower the melting point. So you have NaF-BeF2 eutectic or NaF-RbF. NaF-BeF2 has a melting point of 340 degrees Celcius, the lowest so I’ve been looking at this one.

The vessel is fairly easy, use Hastelloy N like the MSRE used. Per Peterson’s AHTR will also use this material in contact with fluoride coolants.

The cladding is the most tricky part of the system. MIT is developing a triplex silicon carbide cladding for PWRs that fits the bill. If it works then this concept becomes a realistic option with many advantages.

The nice thing about thorium is that some of the negative aspects can be used as an advantage with the right design. For example, the presence of U232 complicates fuel processing, requiring it to be done fully automated. But this will be done for pyroprocessing anyways, and it is actually an oportunity to improve the proliferation (divergence) resistance. It is almost impossibly to steal fissile from a hot cell.

The biggest inherent disadvantage of thorium, in my opinion, is the lack of a fissile isotope. This requires a fissile starter material. The best material is plutonium. Low enriched uranium is also possible, but requires a high enrichment (because most of the fertile needs to be thorium it leaves little uranium which then needs a high fissile percentage to get a reasonable fissile loading).


Cyril: Plenty of high-activity fission products to keep the system hot? I guess that would depend on the rate of cleanup chosen for the core and blanket circuits, and the immediate past history of power production. Also, one would have to be careful to ensure that the FPs are evenly distributed in the salt.

The last sentence in your second-last paragraph indicates that you don’t think proliferation is an issue for either IFR or LFTR. I think some skeptics in the Administration will have to be convinced.

Short- and medium-term reactivity control in LFTR seems to depend on the rate of extraction of the protactinium. I would pray for a very highly reliable and steady extraction system, if that is the case.



“Liquid salt coolants allow high thermodynamic efficiency to be achieved using off-the-shelf (600C) ultra-super-critical steam equipment.”

But is thermal efficiency a primary consideration? I think not.

Readers may not understand the relevance of the critical point of water, which is 647 K (374 °C; 705 °F) and 22.064 MPa (3200 PSIA or 218 atm).

Above that temperature and pressure, water behaves differently than below it, because it does not exist as a liquid and it is thus possible to obtain greater operating efficiency due to lower losses, especially condenser losses.

Supercritical points for other fluids can be found on-line, eg at

Supercritical operation is a good thing from a thermodynamic point of view because it enables higher efficiencies – more electricity for the same input heat energy. This comes at the price of increased operating pressures and temperatures, hence possible higher capital cost.

One impact of Chris’s point is that water can be used as the fluid which carries heat from the boiler to the turbine. Candidate designs which use other fluids – lead, sodium, helium, etc – need to manage risks associated with those fluids, whereas steam, if it escapes, is water vapour, the risks of which are very well understood.

Water is also cheap and plentiful.

If Chris is suggesting that water is better than some other candidate fluids and that supercritical operation is desirable, it may be on the basis of lower cost, safety and efficiency and because we have knowledge gained from the many thousands of water boilers in service worldwide, a percentage of which operate in the supercritical range. I agree, but only to the extent that these impact final cost expressed as LCOE and public/regulatory perceptions of risk.

A final note on efficiency was provided by G Stanford, above:

“It also seems likely that thermal efficiency per se will not be a major issue, in view of the very low cost of fuel for breeders.”

I agree. Chasing efficiency when the raw cost of fuel is low is not important, because it will have minor effect on LCOE.

Caution: I am not an expert in thermodynamics. Have I overlooked any significant other factors?


@Dan Meneley on proliferation … “reactivity control in LFTR seems to depend on the rate of extraction of the protactinium” – did you really mean “extraction”?

Sure, after an LFTR powers down, the transient Pa 233 (that was near equilibrium with the previous rate of creation and fission of U 233) will continue to decay and increase the reactivity. It would be an automatic process to remove or dilute the fertile salt that remained in the core to ensure that its reactivity could never reach criticality.

However moving some of the salt outside of the core would be enough. If an LFTR installation was found to be taking the extra step of extracting Pa 233, there would indeed be a proliferation concern as any protactinium isolated would decay to highly enriched U233.


Roger: It is my understanding that, in order to sustain a conversion ratio of 1.0 or above, the LFTR must remove protactinium from the circuit, more or less continuously. The decay product, U233, is returned to the circuit after an appropriate delay.

Am I wrong?


Yes Roger, he means extraction of Pa-233 and out-of-core decay to U-233, this is how the proposed LFTR fuel cycle works. The question I guess is the purity of the removed salt, as you say.


Could someone explain to me, as a lay-person, the advantages of both the LFTR and the IFR over advanced CANDU reactors (or point me in the direction of some good reads on the matter)? I’ve read numerous times that CANDUs can run off of natural uranium, thorium, and even “burn” weapons grade plutonium.

Are there any operational CANDUs which are using thorium as a primary fuel source now? Why is the LFTR considered so important if CANDUs can already use thorium as a fuel?

And what are the advantages of the IFR over the CANDU, if CANDUs can use natural uranium or plutonium as a fuel source? Does the IFR have a higher burn up rate/fuel use efficiency?

And how do costs (or projected costs) compare?

I assume that if either the LFTR or IFR has more advantages over existing CANDUS than the other, that would give that reactor a competitive edge.


Tom: Yes,the CANDU can operate on natural uranium, slightly enriched uranium, RepU, MOX. Further it can run as a “near breeder” with top-up of fissile material, either U235 or Pu. None of the 30-something operational CANDUs are now operating on thorium as a primary fuel source. Both India and China are well along on the development pathway to that goal. (In Canada and other countries the simplicity of the natural uranium cycle has, so far, won the day.)

However, the fraction of the original potential energy that can be extracted by CANDU is only a few percent — running on a natural uranium once-through cycle it extracts 0.75 percent – only about 25% more than a PWR. So long as one has a goodly supply of cheap natural uranium, all is well. In the long-term future however, especially as the demand for nuclear energy expands by 1-2 orders of magnitude, uranium will become expensive (thorium too). The then-existing world fleet — mainly water reactors — will require a steady flow of economic used fuel.

It is prudent, therefore, to develop advanced reactors that can assist CANDU and other thermal reactors to sustain operation as uranium prices rise — at least over their expected 100-year life span. It also is prudent to do so in order to reduce the hazards and volume of “used fuel” discharged from those older reactors, and that must be stored for a long time.

The prime candidate for this duty is the SFR, especially the design known as IFR. This new IFR fleet will be able to utilize all of the accumulated used fuel from thousands of thermal reactors and extract about 100 times more energy from each ton. At the same time the final used fuel from the SFR will be much easier to manage than is the thermal reactor used fuel.

Personally, I am not sufficiently knowledgable regarding the LFTR to say whether or not they will be able to produce a sufficient amount of excess fissile material to sustain the older thermal reactor fleet — I suspect not. In addition, I doubt that they will be able to “dispose” of the accumulated used fuel from the older thermal reactors, at least as efficiently as the SFR can. These two tasks are “must do” assignments for any advanced reactor, in order to justify its development costs.

Capital + operating costs of CANDU reactors today are comparable to those of LWRs, based on recent experience in China. Future costs of all “fleet capable” units will decrease slowly as technology and experience is accumulated. I have no firm data, of course, for the IFR or the LFTR.

Best regards



@Dan, Barry. A thorium MSR design can ensure maximum conversion ratio by “removing” (which Dan may actually have meant by saying “extracting”) occasional bucketloads of salt from the core, replacing it with bucketloads of salt which had been sitting out-of-core for a few months. (Where xenon is removed at the pump, fission product cleaning might be quite infrequent, if at all).

On the other hand, it would worry the proliferation people, if pure Pa 233 were chemical isolated, and then deliberately accumulated until it had decayed to highly enriched U 233. The extra process seems unnecessary, expensive and requiring levels of radiological control and security unwelcome in a power station. Am I missing something?


Roger: Certainly, that is how a “normal” operator would operate. However, the first assumption of the safeguard folks is that the plant owner (a national government) has a deliberate plan to produce illicit weapons-grade material. Cost is no barrier.

I understand that there would be some degree of protection afforded by the small amount of high-activity U232 that will be present in the protactinium stream.


If I understand this correctly this seems to be saying Australia may never need a uranium enrichment industry if it takes the CANDU–>IFR path, apart from imported start charges that is. Australia will soon produce as much thorium as uranium purely because of demand for zircon and rare earths, not as nuclear fuel.

This is a setback to those who were hoping the AP1000 could be built in Australia somewhat closer to the Korean price. It locks us in to one particular Gen 3 manufacturer so we don’t get to shop around.


Dr. Meneley,

Thanks very much for your response, it clarifies well my questions, and your framing of the two “must do” assignments to justify investment in advanced nuclear power is a compelling argument for IFR development.


Tom Keen, in the paper “Transition to Large Scale Nuclear Energy Supply“, Dan Meneley has gone over the CANDU/IFR scenario is some detail. It is really worth reading. Here is a snippet:

The next step of site development could be addition of more generation capacity; if this step is taken in the near future a good choice will be CANDU reactor units, either of the type now operating or the new ACR type. Later on, integral fast reactors might be added as a first move toward a system with a closed fuel cycle. These reactors could utilize the used CANDU fuel now stored on site, given the addition of a processing plant. (There is already sufficient used fuel on site to power an integrated generation complex of ~15 Gwe for several hundred years.) A U238 extraction plant could upgrade this fuel and supply the first charge to each fast reactor as well as recycling mixed-oxide fuel to onsite CANDU units. Depending on the rate of capacity buildup it may be necessary to supply a limited amount of enriched uranium or separated plutonium to the site from external sources.


Please bear in mind there are several types of LFTRS. The workhorse is the single fluid “Converter”, the breeders are usually dual fluid.

The single fluid version was the one actually built and tested for about 5 years.


Generally the mining companies that are mining zirconium, titanium, and the lanthanide elements such as Iluka and Lynas want to try and find mineral deposits that contain the minimum amounts of U and Th because they don’t want to get caught up in the political and regulatory headaches associated with “uranium mining”, “radioactive waste” etc., because they don’t consider those headaches to justify the value they might get from producing a small amount of processed saleable uranium oxide, and the demand for Th at the present time is very low.

I would be very interested in hearing what Martin Ferguson and Martin Hoffman have to say about nuclear energy at this upcoming thorium nuclear energy conference in Canberra.

@John Newlands: Personally I don’t see any value in the “energy amplifier” aka. accelerator-driven subcritical reactor.

You need a very large, very powerful particle accelerator (something roughly on the order of 10 mA of proton beam at 1 GeV, so about 10 MW of accelerator power) driving the reactor (the reactor is cooled by Pb-Bi eutectic, which acts as the neutron spallation target for the proton beam) and adding this very large particle accelerator to your fission reactor obviously adds a lot of complexity and cost.

But what’s the point of all that added complexity and cost? What justifies it? What does the accelerator give you that an IFR or LFTR or Pb-Bi cooled fast reactor does not already give you?

Basically nothing, as far as I can tell.


Dan Menely wrote:

Plenty of high-activity fission products to keep the system hot? I guess that would depend on the rate of cleanup chosen for the core and blanket circuits, and the immediate past history of power production.

Processing out the fission products doesn’t make the heat go away – that’s the fundamental challenge with the fission products. But we can use this to our advantage. Either process online and let the decay heat warm up the hot cell (oven) elsewhere, or don’t process and run as converter reactor (a DMSR) which leaves the decay heat in the fuel. Either way you can design for this heat to keep everything in liquid state even during extended shutdowns.

One idea I have is to just make the hot cell thermally leaky so that it always loses some heat, based on thermal emissivity, the hotter it gets the more heat it loses. That’s a guaranteed decay heat cooling system. If it gets to cold during a shutdown the thermal emissivity drops exponentially. That prevents freezing during extended shutdowns. For really long shutdowns you’d eventually need supplementary electric heating or let the stuff freeze in a dump tank.

I’ve suggested to put the entire primary loop in a bath of buffer salt which shields the radiation and is a heat buffer for transients.

Also, one would have to be careful to ensure that the FPs are evenly distributed in the salt.

There wouldn’t be any fission products of significant quantity in the coolant for this system… its a fluoride cooled IFR. For a LFTR with fuel in the salt, the stuff that wants to stay in the salt stays there and the noble stuff – gasses and metals – come out easily in simple sparging and particle spunge filters. Oak Ridge ran the molten salt reactor experiment very successfully for several years and was able to solve all the operational issues. They were getting ready for a large scale demonstration when the funding was cut for political and budget reasons.

Short- and medium-term reactivity control in LFTR seems to depend on the rate of extraction of the protactinium. I would pray for a very highly reliable and steady extraction system, if that is the case

It’s not the case. The equilibrium Pa233 concentration in the system is very small, and one can easily make do without any Pa seperation while still breaking even on breeding. This is done simply by lowering the power density – the slow decay constant of Pa233 being sensitive to power (flux) density.

It goes without saying that the processing plant should be able to shut down without the reactor powering down, for reliability/economic reasons. This is not very hard since any practical processing will be slow (months to years).


Dan Menely wrote:

Personally, I am not sufficiently knowledgable regarding the LFTR to say whether or not they will be able to produce a sufficient amount of excess fissile material to sustain the older thermal reactor fleet — I suspect not.

I strongly disagree with this line of thinking, for several reasons.

By the time LFTRs are developed the exisiting thermal burner solid fuelled fleets will be older than cranky.

Moreover, in the grand scheme of things, the existing nuclear plants are almost completely unimportant. We need at least 10000 GWe, and probably more like 20000 GWe, of nuclear new build to solve the CO2 emissions problem adequately in the future (growing energy demand). What’s 500 GWe of existing capacity in that scenario? It’s nothing.

Besides, there’s plenty of mineable uranium available at high energy return, the stuff being log normally distributed in the earth’s crust.

Finally I’d note that using bred fissile to keep the older burner reactors operating is a massively inefficient way of transitioning. You’d be much better off using the bred excess fissile to startup more Gen IV reactors which can also breed themselves. Feeding the stuff in old inefficient burners will retard exponential growth of Gen IV, whether LFTR or IFR or both. That’s a short sighted way to go about with your fissile resources.

I doubt that they will be able to “dispose” of the accumulated used fuel from the older thermal reactors, at least as efficiently as the SFR can.

Surprisingly, they can. The use of thorium means very little transuranics are produced, and the lower fissile loading (softer spectrum) means they can use existing TRU supplies more effectively. Also keep in mind that uneven numbered TRUs tend to burn out better in a thermal spectrum.

If you consider TRU as a waste, then the SFR makes more sense since it can gobble up more per reactor. But that is not how it works in the real world. In the real world, TRU costs an arm and a leg to process out of spent fuel; any reactor that can start up on less of it has an economic advantage.

A Pu started LFTR or IFR running on thorium is a very efficient waste eater. I like the IFR but I don’t like sodium at all.


Cyril: You make good points. I think that our reasoning differs in only one important respect; that is, the number of water reactors that will be operating at the end of the current century. Your thought is that there will be about 500 of them. My guess is that there will be much closer to 10,000 of them in operation. I think the underlying driver will be the sky-high price of petroleum — this is an urgent and critical issue, according to the OECD-IEA (see their last few World Energy Outlook reports).

Some years ago the IAEA began to look at this issue. Here is one report that summarizes the situation as it was seen in 1997: N. Oi, L. Wedekind, International Atomic Energy Agency (IAEA), “Key Issue Paper #3, Proceedings of the International Symposium on Nuclear Fuel Cycle and Reactor Strategies: Adjusting to New Realities” (June 1997).

That report was a small part of a very general conclusion, which was that economics, the abundance of uranium, the urgency of installing this new energy source, and the ingrained habit of very cautious change within the electrical utilities — who are the ones who eventually make the decision as to whether or not ANY power reactor will be installed — will combine to favour the choice of water reactors for installation over the next several decades.

Of course that conclusion could be wrong, but it will take a very large tsunami (or equivalent) to alter this trend in the short term (about 50 years), I believe.


John Newlands, whether Australia goes CANDU->IFR or steam engine -> water wheel probably doesn’t bear on setting up an enrichment industry, given the small size of the domestic market. We should at least do it for commercial exports. Not that we’ve ever shown an inclination to add value to our quarries, but hope springs eternal.

I doubt we’d skip LWRs. A mix of CANDUs and AP1000s looks very attractive. The CANDU reactors can be small, 6-700 MW, compared to 1150 MW for the AP1000, so they’d fit better into our grid.

We could take the relatively easy first step of fabricating CANDU fuel from natural uranium, without enrichment, as a low barrier entry into fuel services and a value add to the raw material. The door would appear to be open to selling oz-fabricated CANDU bundles to India.

And ultimately, for the fastest large scale rollout, the different reactor designs have different rate limiting components in their supply lines. The world could probably build more CANDUs at once than AP1000s, and certainly more at once of both than either.


Dan Meneley wrote:

I think that our reasoning differs in only one important respect; that is, the number of water reactors that will be operating at the end of the current century. Your thought is that there will be about 500 of them. My guess is that there will be much closer to 10,000 of them in operation. I think the underlying driver will be the sky-high price of petroleum — this is an urgent and critical issue, according to the OECD-IEA (see their last few World Energy Outlook reports).

OK, I can see this is a more realistic scenario which I foolishly ignored. In that case though, we’ll just use mined uranium for the water reactors and use the resulting byproduct transuraniums to startup newer Gen IV reactors. I just don’t see the point in using bred fissile to keep the water reactors going – they need a lot of fissile, especially light water reactors because they are burners at low conversion ratios. Keeping feeding them bred fissile from Gen IV while they use up only a portion of it and then send the rest to reprocessing again, I think is not a likely scenario. This plutonium is much better used in IFRs or LFTRs. If CANDUs are used in stead of LWRs it gets a little better with faster burnout of Pu and lower fissile start charge. If CANDUs can use ThO2 fuel with transuranics from spent fuel as fissile, to make U233 for starting up “pure” Th-U233 cycles (whether in IFRs or in LFTRs) then it gets better still. In fact, on the I’ve suggested to use CANDUs to do just this in order to make the startup charges for LFTRs. If IFRs are first to the scene then these could probably be used to make U233 for slower spectrum reactors as well.

If we look at the different deployment scenarios, the fastest deployment comes from using two generations of reactors, first faster spectrum reactors to burn out transuranics and later on using thermal spectrum reactors that leverage the fissile a factor of 5-10x. So if IFRs are first to the scene that is fine by me. I remain worried that the public won’t accept sodium coolant as being walk away safe. Barry Brook has convinced me more or less, I hope he can convince enough people of sodium’s safety.


“If you’re using thoria fuel, how are you going to extract the U233 economically?”

The “Thorium 2011 Remix” DVD covers this. According to Sorenson:

1) Blanket is thorium fluoride salt. Neutrons convert some to UF4.

2) Pump some into another chamber and hit it with fluorine gas. UF4 will convert to UF6, which is a gas. Thorium only has tetrafluoride so it stays put.

3) Now you want to convert that UF6 gas back to UF4 for the reactor core. You do that in another chamber, hitting it with hydrogen, which strips off the two extra fluorine atoms to make HF.

4) Put the HF through an off-the-shelf electrolyzer, feed the F back to chamber (2) and the H back to chamber (3).


Isn’t HF a gas that only becomes acidic if exposed to water? It would seem to be a dangerous material to be exposed to but not in and of itself a problem if properly isolated. My understanding is that the LFTR requires that the salt be extremely pure regarding the exclusion of water in order to avoid such problems. I would think that this process would be somewhat assisted by the high operating temperature which would tend to drive off moisture.


Dr. Dan Meneley wrote:

Roger: It is my understanding that, in order to sustain a conversion ratio of 1.0 or above, the LFTR must remove protactinium from the circuit, more or less continuously. The decay product, U233, is returned to the circuit after an appropriate delay.

Am I wrong?

Dan, yes this is wrong. Even in the 1970s version MSBR protactinium removal was not necessary to break even. If not done (and it probably never should be for many reasons, the simplest economic) then the Breeding Ratio only would drop from about 1.065 down 1.01 or so (see fig 5 where S=1.5 for the MSBR from this 1970 Nuc Tech paper that had a whole issue on MSBR)

Click to access NAT_MSBRfuelcycle.pdf

And that was for a single fluid design. Two Fluid designs have the thorium in a separate blanket salt which sees far lower neutron fluence and thus sees even lower losses to Pa233 absorption. Pa removal was only really called for since the mandate of Oak Ridge in those years was to get as high a breeding ratio as possible back in the days they thought there was very little uranium (actually doubling time was the real metric and the MSBR with 20 years was pretty darn good).

David LeBlanc


HF is a weak acid but is capable of etching glass and silicates whereas mineral acids cannot. Its not dangerous because of its acidity, its much scarier than that. If I spilt concentrated HCl on myself, I’d go, “Oh bugger” and wash it off. If I spilt HF on myself that would be a full blown five alarm medical emergency.

HF only causes deceptively mild irritation on exposure. But the molecule easily diffuses through tissue with consequences days later causing bulk, deep tissue necrosis. It diffuses to the bone where it converts the surface to CaF2, apparently a very painful condition requiring abrasion of the bone surface to correct. And it interferes with calcium metabolism, causing cardiac arrest and death with a skin exposure as small as 160 cm2, according to wikipedia.

Its not quite on my list of things I won’t work with, but it scares me. We keep calcium gluconate gel in labs where its used for emergency application, right next to the cyanide antidote. Use of either would be a major emergency. When The New York Times was my local paper in the 90s, it carried a report of some garbage collector being doused from an improperly discarded bottle in someone garbage. Poor bastard died. Horribly.


I would hope in the future we can bring the debate around to evaluating Molten Salt Reactors or if one wishes, Liquid Fluoride Reactors not just LFTR (which I’d also caution is different things to different people). It is the reactor type itself which is the true wonder, not thorium. As Per Peterson has rightfully pointed out, in the foreseeable future the price or availability of uranium is not a true obstacle to even LWRs, let alone CANDUs. MSRs run as converters with a mix of low enriched uranium and thorium or even just uranium alone are an extremely attractive option needing very little annual enriched uranium and I encourage the reader to learn more. Oak Ridge’s last major effort on MSRs around 1980 was on this concept and proved very compelling.

Click to access ORNL-TM-7207.pdf

With a fuel cycle cost for such a converter being about 0.1cents/kwh* it is hard to ignore this option which has far less technological uncertainty than any breeder option which must first at least prove salt processing methods and in many cases much more. The breeder is perhaps the logical second step but it is highly likely a converter comes first so I’d wish we’d discuss it more (not nearly as sexy I guess so I’m not likely to get my wish).

(Caution: Shameless self plug). For any interested I’ve tried to give a review of the basic principles along with some thoughts on improvements in this Nuclear Engineering and Design paper.

Click to access MSR-Molten-salt-reactor.pdf

Just as many believe it should be IFR AND LFTR not IFR OR LFTR, please don’t forget many of us also support both breeder AND converter Molten Salt Reactors.

* 1980s DMSR converter averaged 40 t U/GWe-year (4M$ at 100$/kg) and 47 kSWU at 0.2% tails (at 150$/SWU is 7M$) is 0.125 cents/kwh (no fabrication needed and conversion only UF6 to UF4)


David (wrt #comment 143325): Good comment. I do, however disagree with your comment emphasizing the supposed uncertainties of “any breeder option”.

India, China, Russia and possibly Japan will very soon eliminate the small residual uncertainty of the ifr-type of breeder option. Thirty years’ operation of EBR-II with lifetime capacity factor around 75%, and the BN-600 with a similar long-life capacity factor, have essentially eliminated the operational questions. Proof of the viability of used-LWR-fuel reduction is still an essential step, but that is no big deal.

Oh, and we also have a pretty darn good converter in the good old CANDU 6 — running on a mainly-thoria fuel, once through cycle. And what will that produce? Lots and lots of used thoria, containing U233. Good startup feed for LFTR and its cousins.

Comment to Per Peterson: The “foreseeable future”, to me, includes a time when petroleum will be a small-scale industry. Fission will be carrying the big load, in 10,000 to 20,000 units of 1 GWe each or equivalent. On that scale, we need to worry a lot about sustainable supplies of fissile material.


Thanks Dan for the <a href="question “>clarification .

From the discussion, it seems depressingly easy for a proliferator to extract uranium from a thorium fluoride melt. The IAEA (TECDOC 1450) seems to head off the possibility by recommending the addition of U238 to ensure that any extraction is of low enrichment. However, the presence of U238 absorbs scarce neutrons, and creates higher actinides after all.

However Dan’s question in the post was not about mixed fluorides, but about thoria, ie solid thorium dioxide fuel. It was derisive, Implying that it would be particularly difficult to extract traces of uranium from something so refractory and chemically resistant.

Being so tough that is unlikely to be recycled by a commercial operator, the proliferation people would not be much worried by its use. It would seem that thorium dioxide is an ideal non-proliferating fuel for long residence, high burnup and eventual burial intact. Would ThO2 fuel would be used in the above references to a CANDU?


Roger: Yes, ThO2 fuel will definitely be used in any CANDU application. This fuel is already well known as an excellent performer with regard to dimensional stability and fission gas retention.

We hope to achieve about 5% burnup of thoria in a once-through cycle, in a checkerboard fuelling system including U-Pu or enriched U driver fuel. Special reprocessing will be essential to extract U233 — but, of course, the U233 will always be accompanied by high-activity U232.


Replying to David LeBlanc, at (#comment- 143313): Thank you David. This clears up a long-held misconception. This leaves me, with regard to Pa233 management, only one remaining question — could deliberate diversion of Pa233 be achieved, or not?


Dan Meneley,asks

Replying to David LeBlanc, at (#comment- 143313): Thank you David. This clears up a long-held misconception. This leaves me, with regard to Pa233 management, only one remaining question — could deliberate diversion of Pa233 be achieved, or not?

For a breeder MSR who will be processing the salt I suppose if you take the “what if” question deep enough one could imagine some way. However if one is not trying to pull out Pa on purpose, the chemical processing needed will in no way produce separated Pa so a very major and obvious refit of the plant would be required. I’d add that processing on its simplest level (for Two Fluid designs) is simply fluorination to remove uranium (a mix of isotopes including U232) and vacuum distillation afterwards to recover the carrier salt (fission products collect in a “still bottom”).

For a converter the case is even better as one needs no salt processing on site whatsoever. We’d hope operators at least recover and recycle all actinides and the carrier salt with one batch job at the end of the salts life (ORNL calculated for 30 years, I prefer shorter batches of 10 or less). This operation could be at a central facility and there is no rush (or build what you need onsite after many years of revenue generation). A “base model” though could entail a true Once Through cycle if desired and just eventually send the actinides along with fission products to geological storage (after likely some sort of vitrification). Even this somewhat wasteful mode would still need far less uranium than LWRs, CANDUs (it is what ORNL assumed in their study).

Finally of course the converter design might even forgo Thorium altogether. This brings a few advantages including no months long reactivity changes post shutdown due to Pa turning to U233. The downside is a bit more annual uranium but still only a fraction of current converters.

Thanks again for your interest,

David L.


Dan said,

Good comment. I do, however disagree with your comment emphasizing the supposed uncertainties of “any breeder option”

Sorry, I wasn’t clear in my comment. I was meaning the technological uncertainty of going straight to a MSR “breeder”, not breeders in general. Although I can’t resist commenting that while sodium fast breeders have a far more extensive operational history I think many would argue the track record hasn’t been stellar (certainly good enough to go further though).

If you then are thinking why another converter when we have LWRs and CANDUs now. I won’t get into a long list of “opinion” but many of us feel there is a very strong case for just about any angle you look at it from uranium usage to costs to safety to the long lived waste profile.

David L.


Yes, the modern environment is different — the Monju sodium leak was transformed from reality (a fix requiring 1-2 months to complete) into a massive failure needing 20 years of study, by chicken-little people who were convinced the sky was falling. And, the Fermi-1 subassembly melt — caused by the demands of an overanxious regulatory body — seemed to be the cause of the project’s failure when it was actually brought on by an ignorant Washington bureaucrat. Also, the long string of successes of the French fast reactor program were converted into disaster at Creys-Malville by the meddling of an anti-nuclear government. The British program failed primarily because the British welders didn’t have the benefit of learning the good techniques of their American counterparts. EBR-II suffered only one small sodium fire in its 30 years of operation — that caused by a fool cutting into a sodium-filled pipe with a hacksaw. That leak was fixed immediately. The S/Gs of EBR-II never did leak.

Uranium usage, etcetera? I argue that the CANDU-IFR combination already has all of those issues well in hand. Ah yes, David. But CANDU is a “bird in the hand” and LFTR is still “in the bush”. There will be a few surprises along the way, as the concept is developed from lab scale to a commercial enterprise in which each unit must produce about 500 to 1000 MWe on a 24/7/52 basis. Then, will the ultimate decision-maker — the electrical utility — actually buy them?

There’s many a slip ‘twixt the cup and the lip.

Best regards


There was a sodium fire at INL recently and a worker suffered some burns. Not too much damage but this kind of thing doesn’t stimulate confidence in the supposed benign operating nature of sodium coolant.

There was also a sodium fire at a Dow chemical plant here in the Netherlands recently.

Some worker cutting a sodium pipe open seemingly unaware of what he’s doing is also not instating much confidence in the safety culture.

I’m not the one that needs convincing, it’s the public that has gone to college and has had elementary chemistry lessons (we all remember the sodium+water reaction). It is a tough mental image to fix.

As for CANDU, they are awesome, but alas, they may be a bit of a dying bird in the hand with recent business developments…


Looking into how it’s running, things are A-OK for now – SNC-Lavalin has plenty of refurb work to keep them going. It would be nice, of course, if the economies of the world would recover.

Canada still holds the intellectual property rights.


There’s a long list of properties that make sodium the natural choice for LMFBR coolant, and one manageable downside, which is sometimes played up to an extent out of proportion to its seriousness. I don’t think this FUD will detract from the appeal of the IFR, but it does reduce the credibility of those who spread it.


I would just add to David’s excellent points that the simplest possible uranium MSR will have the shortest development and licensing period and the lowest development and licensing cost of any MSR, due to very simple system design and similarity to the demonstration reactor that ran very well. It will also be the easiest design to mass produce in large numbers at the lowest cost.

We must keep our eye on the goal; replace fossil energy sources as fast as possible by developing cleaner, safer, less expensive sources of energy as fast as possible. The goal is not to produce the most elegant reactor possible.


Well said Bill Hannahan (#comment 143401 and -02).

There will be scale-up difficulties with ANY new alternative, and those will take time to resolve.

Fortunately we already have the PWR and the PHWR; both machines have passed all of these hurdles over the years, and can be mass-produced whenever we gather the courage to do so. The IFR also is very close to that final stage. All that is needed is a scale-up of the pyro-processing facility for conversion of PWR used oxide fuel to metal fresh fuel.


On second thought, Elegant Simplicity is a hallmark of the best engineering, which is what we want.

Rube Goldberg engineering solutions are easy to come up with, Elegant Simplicity is harder. Steve Jobs was a master at that.


@ Bill Hannahan:

My limited understanding leads me to believe that the IFR would have the shortest development and licensing period (from now) of any Gen IV design. I think, therefore, that a commercial demonstration reactor/pyroprocessor are needed as soon as possible. Perhaps you were thinking specifically of designs involving salt coolants or salt fuels when making your comments?

From an economic perspective, molten salts appear to offer eventual advantages. The most elaborate of designs based on these would probably be descibed as LFTRs and they would appear still to require considerable development time and money. There are two evolutionary designs which could probably be more quickly licensed than a “final” LFTR type and it seems possible that their successful deployment might be even obviate the need for evolution at all. You make mention of the uranium only, single fluid MSR of which David LeBlanc is a major proponent and you clearly favour its promotion. The alternative seems to be the PB-AHTR espoused by Per Peterson. This seems to have attracted some financial backing from the US Government.

I would be very grateful for an assessment by you and other posters of the relative merits of the two “evolutionary” approaches and guesstimates of time to deployment of commercial demonstrators for each, given no financial or unreasonable regulatory constraints.


Taking a USA perspective. Materials qualification is the biggest technical licencing issue facing new reactor designs.

The AP1000 is fully qualified and completing its license in the US with Chinese builds already going on.

If I understand correctly, the IFR has a more or less qualified and generally well proven fuel, and non-qualified integral processing equipment. It is not licenced in any meaninful way, I understand.

The PB-AHTR uses a qualified fuel (TRISO) so that reduces the development and commercialization timeline. It also uses qualified ASME materials for construction, such as Hastelloy N. This alloy is not qualified for nuclear pressure boundary but the PB-AHTR cleverly solves this by using the TRISO fuel as the nuclear pressure boundary (ie the fuel has both cladding and pressure vessel function). So the Hastelloy is non-nuclear pressure boundary. So all the materials are qualified but the system as a whole is innovative and not qualified.

The single fluid MSR has no fuel to fabricate so that is an advantage in the development cycle. It is also a disadvantage in that it is so deviant from existing regulations. Which are very bureaucratic and any change will be slow and painful. For the US there is a problem with the regulations that requires certified materials for nuclear pressure boundaries. The Hastelloy N is certified but not for the nuclear pressure boundary, and unlike the PB-AHTR there is fuel and fission products in the coolant. We know the corrosion is low with Hastelloy N but the problem is with licensing and certification of this alloy. One way around this that I and many others are working on, is to use a containment as ‘nuclear pressure boundary’ (there is actually no pressure but that is a licensing term). The entire reactor vessel and piping, then, is just a ‘flow guide’ (again that is the terminology required for licensing) and we can use non nuclear qualified materials like the PB-AHTR can. Only the containment needs be a qualified material but we can use already certified materials for this, basically the same stuff LWR vessels are made out of. Think about an AP1000 containment type with passive wall cooling, except it would be double walled with the first wall being the ‘vessel’ and the second wall the actual containment.


Regarding the IFR licensing, one pragmatic way to get started as early as possible is to make a non-processing IFR. Just a converter like a water reactor. That way we can get started without the processing plant. This can be developed simultaneously, if necessary it can be retrofitted to the non-processing IFR that is built already.

I feel the same way about speeding up molten salt reactor development: get started with simple single fluid converter reactors, add processing equipment later when it is ready.


Dan Menelely, how about getting started with a low enriched converter IFR without processing? This way we can focus on the reactor itself while the processing plant gets developed.


Cyril: Would love to, but I don’t have the money.

Seriously, the processing plant development is already complete — the EBR-II guys did that. Minor scale-up is required.
The major development requirement for IFR is oxide reduction of LWR used fuel and demonstration of resultant metal stream, then a burning demo in a medium-size plant — PRISM size.


Dan, what I meant was, startup the IFR with low enriched uranium so that you wouldn’t need the oxide reduction processor. If the online fuel processor is not a development issue then you add that to the plant. Low enriched uranium is also easier to get than SNF TRU. We currently (with aqeous processing) only use SNF Pu, not Am and Cm so that doesn’t do all that much for long lived wastes reduction. Also if you rely only on SNF TRU for startup the processing capacity of that SNF TRU becomes a bottleneck for Gen IV buildout.

With a fast reactor you have the possibility to startup on low enriched uranium while breeding, so I think we should take that option.


From a Global Warming perspective, it makes absolutely no sense to develop the IFR further. It is far too cool to replace most fossil fuel applications. The MSR is barely hot enough.

The Russians have an amply-sized, fully developed IFR that only the Chinese, who will buy two of anything for learning purposes, have bought. The MSR cannot be bought so the Chinese have initiated an MSR development program and have repeatedly stated they will assert international intellectual rights on what they come up with. The MSR race is on, there is no IFR race.

The Converter MSR (a pot, a pump, and a pipe) approaches a fossil boiler in its physical simplicity.

The author’s MSR applications call for large amounts of FLiBe. FLiBe’s cost and availability could be an Achilles heel.


To Jim,

Not sure I follow the logic of IFR not helping since it runs cooler. Yes, having possible industrial heat options are nice but electricity production is by far the biggest need. In the IFR and/or MSR debate I do guess it adds a bullet point to MSRs though.

Not sure who you refer to as “author’s MSR application” but I should add that while Flibe availability is a valid concern most feel Li enrichment is not going to be a huge challenge, just it hasn’t been needed in quantity before (except by the military and cost or the environment was no concern). I’d add though Molten Salt “fueled” reactors do not need flibe, especially for converters but surprisingly enough for many breeder designs as well. Molten Salt “cooled” do though need flibe for negative void reasons (except for fast spectrum designs).

To all,

In the very beginning of the exchange, G. Stanford refers to a link to “Carlsen’s enthusiasm for thorium”. I don’t know who this is but the link is indeed without substance, just lets do it to save the planet. I’d add that I was at the Thorium conference at Google (thanks again Google) and I certainly didn’t see Bill Gates or Al Gore there (would Al Gore even consider Nuclear?). Every “promoter” adds some hype (including IFR supporters) but like the media does with news, sometimes the LFTR hype gets way too recycled and exaggerated. Please don’t paint us all with the same brush…

David LeBlanc


Douglas Weis wrote; “the IFR would have the shortest development and licensing period (from now) of any Gen IV design.”

Probably true Doug, in a Business As Usual environment. But, given the high level of complexity, it is not clear to me that the plants could be built quickly, at a low cost, so as to produce reliable abundant kWh’s cheaper than fossil fuel.

They contain huge quantities of pu239 arranged in a geometry that is very poor in terms of nuclear criticality, in order to have sufficient heat transfer. Being fast reactors, they have the potential to ramp up power at blindingly fast rates that could release a very large amount of energy, sufficient to disperse the core over a large area, if they ever get into a configuration that is well beyond prompt critical. How do you write a computer simulation that is guaranteed to rule out every possible mechanism that could produce a sudden change in configuration resulting in such an accident?

Thermal reactors are physically limited in how fast they can ramp power (that is why there are no thermal spectrum nuclear weapons), and of all thermal reactors the MSR has the lowest amount of excess reactivity. It is at or near its most reactive configuration during normal operation. Volatile fission products are removed and stabilized as they are produced, and the remaining fission products are in forms that are stable at very high temperature. I have tried without success to think of how an MSR could be blown up with enough energy to produce a large release of fission products beyond the plant boundary, any suggestions?

I do not think that the regulatory requirements for MSR’s have to be as strict as for IFR’s, just as the requirements for handling fire crackers is not the same as for dynamite.

Had there been no WWII it would probably have taken several decades to do what was done in 3.5 years, the development of nuclear weapons, at a time when knowledge of nuclear cross sections and fission products was very limited. With that level of effort we could be mass producing simple MSR’s in a similar time frame.

But my recommendation is to spend $100 billion per year to push every technology as fast as possible, and use what ever works best.


Thank you, Bill Hannahan, for this excellent comment (#comment- 143471). And that you also for the reference to the discussion in The Oil Drum.

I take issue with you on your insistence that the safety of the IFR is profoundly different from that of the MSR — a comparison of fire crackers with dynamite. Having spent several years in studying exactly the accident question you pose – the energetic disassembly question – for fast reactors, I must insist that the metal fuel used in IFR is entirely separate from oxide fuel, in that its behaviour under severe accident conditions is profoundly different.

Indeed, the metal fuel IFR behaves in a manner much closer to the MSR than to the oxide-fuelled sodium fast reactor. That fact, in addition to the fact that the liquidus temperature of metal fuel is actually lower than that of the fuel cladding, leads to a design that is extremely unlikely to reach the “prompt critical” configuration essential to generation of an energetic disassembly. (Apologies to all those to whom these words are unclear, but it would take many days and many pages to write this statement out in plain English).

It is grossly unfair and demonstrably incorrect to make your “firecracker to dynamite” comparison between MSR and IFR. Public protection is vital for all power plants, and comparison of one design against another deserves a full exposition. A one-liner cannot do the job.


David LeBlanc.

I like David’s enthusiasm, but the “just do it” folk gave us solar panels and wind turbines, with which to save the planet.

The planet is not being saved.

To deal with the evil twins of climate challenge and global energy needs, we need to be much more persuasive and much better founded in science, engineering and common sense than either wind or PV. “Just do it” cannot work at present.

Perhaps: “Just trial it” has a chance, but is it good enough to get IFR or LFTR, for that matter, across the line?

As David pointed out, only China is interested and capable at present. China is not only a huge nation, it has several advantages not currently present in Western nations:
… massive trade surplus
… centrally controlled economy
… long planning horizons – well beyond the 5 year plans
… preponderance of engineers in upper political circles.

The Western comparison is:
… cash strapped
… banker controlled economies and bean-counter decision makers
… No effective planning environment
… Political classes consisting of lawers and other rule-makers with little strategic vision
… anti-science approach is common
… personal & corporate short-term perspectives win every time against national or global realities.
… USA election cycle of 24 months, Australia 3 years – a planning horizon of 1 year, perhaps much less. Why commit to projects which take a decade or two, in certainty that the next mob will likely abandon them?

David’s war cry might as well be “Just leave it to China!”

Let’s get back to problem solving and knowledge building and hope for a few strong, visionary leaders in nations such as Germany, USA, Australia, Canada, France (can she do it again?), India, Brazil and so forth.


Dan Meneley, here’s an issue I’m thinking about.

The high breeding ratio IFR, especially if a small modular reactor, needs a blanket of U238 rods. U238 fast fissions if the neutrons are really fast. If there is a void in the sodium in some beyond design basis accident (BDBA, traditionally not considered in the core damage frequency calculations) the spectrum hardens due to reduction in inelastic scattering and the blanket produces more power. This could lead to a positive feedback. I know IFR proponents like to believe that below grade pool type design makes this scenario almost impossible (and I tend to agree with this notion) but sodium is reactive with air and water and there is a leak path through the RVACS space for voiding. The boiling point of sodium is also not high enough to preclude boiling induced voiding in some BDBA.

I would like to know how to avoid positive void coefficients with IFRs that have blankets without resorting to inefficient arrangements such as pancake or other high leakage cores.


Dan, thanks for the response. You may well be right in claiming that it is physically impossible for an IFR to have a high energy criticality accident, but I have yet to see a convincing analysis, can you point me to one or more?

I think it could be shown that a core can be designed such that the initial meltdown will take place without a high energy event; I am more concerned with what happens after the core is melted. For example, could a collapse or low energy criticality in one region force a large mass of plutonium in another region rapidly into a more favorable geometry?

How do you poison a melted core to guarantee there will be no criticalities after a meltdown, and is this part of existing designs?

People can differentiate between a plane crash and a train crash, but they do not differentiate between a PWR, IFR or MSR accident. A large IFR accident would put the industry back another decade or more.

The potential for a big accident is my biggest knock against fast reactors, especially if there are other options that can produce abundant cheap kWh’s without that risk. If big accidents can be shown to be impossible I would be more supportive.


Bill: During my youthful career as a fast reactor major accident analyst, I came to the firm conclusion that the only way to be absolutely certain that an SFR would remain shut down after the rods were inserted was IF THE FUEL DID NOT MOVE.

This is one of the good features of metal fuel — it swells like crazy if it is suddenly over-powered – from expansion of fission gas inside the fuel, which is soft like bubble gum at high temperature. The reactivity effect is negative — and not recoverable. The melting temperature of the fuel is lower than that of the fuel clad, so the geometry remains fairly constant. The decay heat is taken away via natural convection in the sodium.

If I were a regulatory staffer I would say “show me the proof”. I’m trying to get that now — give me a few days.

As for big accidents, I think we’ve already seen one of the biggest possible. The steam-fuel vapour explosion/hydrogen-oxygen explosion of the Chernobyl-4 reactor was mighty impressive, even for the most skeptical regulator. Death toll? around 30-40, similar to the death toll from the crash of a small commercial aircraft in Central America, that same week.


Dan Meneley wrote:

This is one of the good features of metal fuel — it swells like crazy if it is suddenly over-powered – from expansion of fission gas inside the fuel, which is soft like bubble gum at high temperature. The reactivity effect is negative — and not recoverable.

Dan, this is true if you have only a reactivity initiated accident and it stays with that. If you have a structural failure in the passive decay heat removal system the decay heat will rapidly heat up the fuel – a high power density core with a high decay heat power density. After the sodium pool has heated up a bunch the fuel will start to melt from decay heat. If the fuel rods are not vented they will swell from fission gas expansion and block sodium coolant flow, causing further overheating and melting. Fuel ballooning is a serious issue for high power density fuel rods.

Imagine a scenario where an earthquake or terrorist aircraft attack damages or blocks final heat rejection from the passive decay heat removal path. Could be a damaged or debris plugged chimney or a sheared guard vessel, (partially) blocking air flow. Imagine this, there is no power and you’ve got a blocked passive heat removal system. What happens?

I’ve a personal interest in analysing BDBA for different reactor designs, in order to understand what makes a reactor robust and ideally walk away safe. If we have a design that we can definately sell as walk away safe, that will help with public acceptance. The LFTR and AHTR so far are the most resistant designs I’ve come across, due to the lack of chemical reactivity/stored energy in the entire system until extreme temperatures and the high boiling points and heat capacities of the coolants.


Now, we’re getting down to specifics. For instance, I could imagine a MSR of some design that has developed a blockage in its piping, due to insufficient local heating of the salt. At some other point in the circuit, fission products cause overheating of the salt to the point of failure of the piping. After all, the fission products that are the direct result of power production have to be SOMEWHERE — can they always be presumed to be adequately cooled?

It is not my normal practice to conduct a formal safety analysis via e-mail. What set me off in this case was a rather flippant comparison of LFTR with IFR as being similar to comparing a firecracker with dynamite. That statement is so patently false that it raised my hackles.

Risk of death from nuclear plants of any sort depends much more on the people who build and operate them than on the characteristics of the machines themselves. Machines are much too stupid to make mistakes — but human beings have developed that particular art to the n-th degree.


Dan, yes, flow blockages can be imagined in MSRs as well. In that case though you have less volatile coolant, boiling point >1400 Celcius, less chemically reactive coolant, no significant reactivity reserve, and no positive void coefficients.

Also MSR primary flow blockage isn’t caused by overheating, rather by overcooling, so you have negative feedback. If a plug forms it will overheat from fission product self heating. With the IFR there is no self heating of the coolant so if a plug forms it won’t self unplug. To be fair such a plug is also much less likely with sodium’s low melting point (though still well above room temperature). In pool type design, plugs in the primary loop are rather impossible so both IFR and LFTR can design this risk out, IMHO. The AHTR for instance uses a buffer salt pool at a temperature above the primary salt that is submersed in a closed loop in this buffer salt pool. That pretty much deals with this issue. Essentially the pool is your oven.

The problem I was thinking about was not primary flow plugging from overcooling, rather it is slow overheating of the incapacitated passive cooling system in a semi permanent station blackout scenario. Similar to Fukushima except there is a passive cooling system that is only working at small capacity or not at all.

In that scenario, the MSR performs much better, with a higher boiling point, lower chemical reactivity in case of primary loop failing, very low reactivity reserve, combined with coolant=fuel gives no chance of recriticality after loss of core geometry or LOCA. If the primary salt spills into the oven or buffer pool salt, or the buffer salt goes into the primary, that’s the end of any critical configuration (the buffer salt is a neutron poison).

The BDBA allowable temperature is much higher for MSR and AHTR, than for the IFR. The salts can soak up more heat per volume as well.

Even if everything goes wrong, the LFTR has the inherent advantage of lacking any mechanism for dispersal of radionuclides. With the IFR you have the reactivity of sodium to disperse radionuclides, and fission product cesium will be mostly in elementary form (the environment in the IFR being chemically reducing). The LFTR has cesium in nonvolatile chemically stable salt form, CsF. Its strontium is in the form of nonvolatile chemically stable SrF which is also non-soluble in water, unlike elementary strontium.

So it is my opinion that the LFTR can do better in a BDBA.


Cyril: “Walk-away safety” is mythology, pure and simple. You cannot sell it to anyone, anywhere, anytime.

Do you NOT agree that the humans are the most complex component of a power plant? Or that the theory of complex systems tells us that they will, in some unpredictable manner, fail? This is, after all, the reason designers incorporate defence in depth into all nuclear plant designs.

A first-rate crew can operate even a third-rate machine very safely. But a third-rate crew can soon create an unsafe state in even a first-rate machine. So, for example, we have two fully capable, testable, diverse, independent, and redundant shutdown systems in each CANDU reactor. These are poised to act automatically at all times during operation.

I agree that we should not be building reactors with strong reactivity coefficients EITHER POSITIVE OR NEGATIVE. And that does not apply only to voiding, but equally to ALL such coefficients. At the same time, the failure matrix considered must be related somehow to potential events in the real world — and not to nonsense such as a reactivity addition “sliding down a moonbeam” into a reactor.

Lets get real.


Risk of death from nuclear plants of any sort depends much more on the people who build and operate them than on the characteristics of the machines themselves. Machines are much too stupid to make mistakes — but human beings have developed that particular art to the n-th degree.

For future Gen IV reactors, I don’t agree with this line of thinking, at all. We should not be building reactors with strong positive void coefficients, no matter how good the operators are. We should not rely on electricity completely for decay heat cooling no matter how good your operators are at maintaining their diesel generators.

We can’t sell this to the public as walk away safe. If your operators walk away (or just happen to be dead) and you rely on operators for critical safety functions, that is not walk away safe.

People can be convinced by a simple physical mechanism that makes a reactor critically safe. People can be convinced by a simple passive cooling system that uses gravity and other physics for decay heat cooling. People won’t be convinced by assertions of a ‘safety culture’.

First you start out with a safe technology, that will work even if badly maintained, then you make it safer by making sure professionals maintain it.


@ Dan Meneley, Cyril R: The possibility of a MSR core blockage due to a broken pump impeller blade crossed my mind some time ago. Upon taking a look at a practical MSR design, the 1 GWe EBASCO reactor-confinement cell assembly (ORNL Subcontract 3560), I came to the conclusion that such a scenario was unlikely since there were actually four pumps and heat exchangers, the pumps were located on the discharge of the graphite core and discharged into four heat exchangers, the intake manifold of the graphite core could easily have an intake screen similar to an automobile’s oil pump sump screen. The discharge of the core is also manifolded.

1,300F is dull red hot. If you look at how the pump motors are located, one would suspect the temperature of the interior of the confinement cell was calculated to always be well above the freezing point of FLiBe heat transfer salt once the reactor was up and running. The “clear salt” leaving the confinement cell is non-radioactive and whatever happens to it is of no consequence.


This is good, Jim. A designer can do a whole lot to minimize the inherent weaknesses of any concept. For instance, passive decay heat removal in the IFR is provided by at least two separate systems. But the designer will never be able to beat the imagination of the reviewer.

We’ve recently been discussing “way-out” scenarios beyond the reach of any designer. One can ALWAYS imagine another hypothetical situation in which fission products would be released to the environment — unless, of course, you have none of these on site.

The Chernobyl 4 event would have been worse had it been raining at the time of the vapour explosion — in that eventuality a high concentration of fission products would have fallen directly on the town of Pripyat. But it didn’t rain that day.



Dan Meneley wrote:

Do you NOT agree that the humans are the most complex component of a power plant? Or that the theory of complex systems tells us that they will, in some unpredictable manner, fail? This is, after all, the reason designers incorporate defence in depth into all nuclear plant designs.

Yes, I agree that humans are a complex component of the operations. That’s why my point is, if you can design the human failure out of the safety equasion, you’ve got a robust system. We can never design humans out but we can design them out of the safety equasion.

So we need to minimize the degree that a nuclear powerplant relies on human actions for safety. If the systems failure mode is fail safe – that is, failure doesn’t impede safety functions – then human failure is of little consequence. In fact, the only area where humans can fail is in design of the systems itself. Once you have a design that is walk away safe and has benign failure modes, you’ve achieved the ultimately safe nuclear powerplant.

Relying on critical maintenance of diesel generators for safety, with dozens of moving parts, and requiring diesel refuelling after some time, is an example of how not to get a benign failure mode. Building a reactor critically safe by the virtue of its reactivity coefficients, and designing the containment to be thermally leaky so that the decay heat gets dumped without any action, is an example of benign failure modes. Having a low volatile coolant is also benign. Even if a pipe breaks, by poor maintenance, the heat still gets removed from the containment and the salt spill mixes with the buffer salt, definately shutting down the reactor and never challenging the containment by either pressure or temperature rise. No one would notice outside the plant. Even if the containment itself is poorly maintained and has holes in it, or is warped and sheared by an earthquake, the heat is still removed the low volatity of the salts ensure extremely low releases.

“Walk-away safety” is mythology, pure and simple. You cannot sell it to anyone, anywhere, anytime.

I have been able to sell this concept effectively to many people, including many nuclear engineers, nuclear chemists, and farmers.


Dan Meneley wrote:

the failure matrix considered must be related somehow to potential events in the real world — and not to nonsense such as a reactivity addition “sliding down a moonbeam” into a reactor.

Yes, again I agree with you. It needs to be scenario based and physically easily conceivable. As we’ve recently seen, earthquakes and tsunamis are clearly realistic scenarios challenging nuclear power plants in certain areas around the world. We’ve also seen that aircraft suicide terrorist attacks are real, even two in brief succession. Such attacks on a nuclear plant will and are incorporated in new reactor designs. For a reactor with fully passively decay heat removal systems, the challenge will be to show how such an attack does not block passive air flow for the passive cooling system. If I understand correctly, the AP1000 has convincingly proven this to the US NRC regulators.

My point is bigger than this, though. If we can claim superiour safety than the AP1000 then convincing people will be much easier. We can all understand gravity, hot fluids moving up, a hot room losing heat to the outside air, etc. The AP1000 still needs safety critical valves and a high pressure containment integrity (which does require testing and maintenance to ensure high pressure leak tightness over the years).

Walk away safe is a great hallmark. Trust me, it sells.


Dan Meneley wrote:

Now, we’re getting down to specifics. For instance, I could imagine a MSR of some design that has developed a blockage in its piping, due to insufficient local heating of the salt. At some other point in the circuit, fission products cause overheating of the salt to the point of failure of the piping. After all, the fission products that are the direct result of power production have to be SOMEWHERE — can they always be presumed to be adequately cooled?

Cyril answered the case of “overcooling” causing frozen blockage and the fact that decay heat actually aids MSRs in that if you stop overcooling, the plug melts. However I assume Dan you are also talking about other sorts of physical blockage. Yes we have to think about that but I disagree with the jump of logic to “to the point of failure of the piping”. You must remember that with the design already operating at very high temperatures radiant heat from pipes provides a cooling that can not be removed until the entire surrounding hot cell increases in temperature. The salts have boiling points around the melting point of the nickel alloys used (about 1400 C). While of course the nickel could lose most structural strength well before this, a temperature of 1000 C has long been thought fine for modest periods of time. With a surrounding “heat sink” of the hot cell at no more than 700 C, a pipe getting up to 1000 C will radiate upwards of 100 kw per m2 which goes up with the forth power of temperature (close to 400 kw/m2 near the boiling point). I’m asuming good emissivity and I’ll admit this is ad hock but even if pipes do fail, everything is within the hot cell which slopes its floor to drain to cooled storage tanks. Some might argue that there are more ways to imagine “expensive” accidents or higher priced maintenance with MSRs but it is just really hard to imagine dangerous accidents due to the lack of internal potential driving forces.

David LeBlanc


So David, is it true you assume that the fission product density distribution in the salt moving around the circuit is uniform? Also, that the “surrounding hot cell” is never drained inadvertently?

We are, with the best of intentions, getting into the early stages of what amounts to a regulatory review. From long past experience I’ve found that the questioning goes on and on and on — the reviewers’ job, after all, is to find ALL of the chinks in the armour of a supposedly impregnable fortress. That takes time, and (hopefully) most of the probing returns reassuring proofs. But that is not assured, and it probably will be necessary to run some experimental safety verification tests.

The LFTR is in the early stages of this process: the IFR is some years further along but is not yet at the end. This is what takes up so much time in development of a new reactor concept. We used to count on one prototype built and operated, followed by one demo plant (200-500MWe) built and operated, and finally one or two commercial size units built and operated before the machine was truly on the market. Finally, the “fleet build” stage occurred, depending on whether or not the customers wanted to buy the product. Now, the times will be longer unless some of the accumulated delay factors ar swept away.

The total elapsed time used to be about 10-15 years to the startup of the first commercial-size unit. Before the fleet began to have some effect on the consumer market, about a further 20 years went by.

The world cannot wait for these new machines. The IFR is going ahead in one way or the other in Russia, India, and China. Perhaps Japan will get back on her feet as well, but IFR cannot have much impact of the generation fleet until well after 2050. National governments, with their myopic view, will have to concentrate on supplying the energy needs of the day — they are unlikely to be deeply interested in developing yet more reactor types.

Long before either the LFTR and/or the IFR becomes important to the future of world energy supply there must be thousands of rather ordinary water reactors in service around the world — otherwise, the main interest will be in preventing mass starvation.

Does it not follow that the two “must do” tasks for either of the new machines are (a) provide the means for managing a huge mountain of used water reactor fuel, and (b) provide the “topup” fissile materials to keep the water reactor fleet operating, at least until their duties can be taken over by the new machines?

Sorry for going off on a tangent. Sometimes I think we’re not concentrating on the correct problem.


Dan, probably your are right with the estimation of the time required to bring to market these promising new designs in ordinary conditions, but I’m fascinated by two projects that US (I’m italian) did accomplish in the past:
– Make the bomb
– Going to the moon

During Manhattan project a dream-team of scientists (and I’m proud to think that our Fermi was the best among them) could make the atomic bomb and in the process invent and BUILD all the infrastructure needed: reactors, BIG factories, many enrichment plants, in JUST 3,5 years!

And with Apollo project in just 7 years they invented everything needed to be invented to go to outer space: giant multistage rockets, moon capsules, space suits, even moon cars!! And in the process invent new materials, processes, and so on.

I really can’t believe now we need 15 years to bring to market 2 PROVEN technologies like MSR and IFR, of which prototypes where already build in the past. Processes and materials needed already invented, technical difficulties mostly solved.

They just need some serious engineering work to bring all together, and above all someone who can firmly decide what to put inside and what NOT put inside these things (no more tinkering, someone have to decide the final design and stick with it) and put a FIRM DATE to the end of the project.

I love to read all these discussions about different ways to make a MSR. Nice. But… someone have to take a reasonable and pragmatic decision of how to make it, even if is a bit suboptimal. I’d love to see all you nuclear geniuses to (mostly!) agree on a single design and build the thing.

I’m the last one to think that “good enough” is “good enough”, but if you want to really finish something sometimes you should compromise to 85-95% perfect, instead 100% perfect. You can refine to almost 100% perfect in the following iterations of the thing, if you have been successful.


Certainly, Alberto, there is no doubt that the US could solve this energy supply problem overnight — if they had the will.

It has been said that the French people have only two reactor types, but 100 kinds of cheese. Sadly, the American people have 100 reactor types, and only two kinds of cheese.

The central idea of those US citizens who actually want to solve their energy supply problem is to build a few hundred of the reactor types that they already know how to build – PWR and BWR, Generation 2 or 3. At the same time invest a relatively small amount of effort for development of one or two advanced reactors on behalf of the prosperity of their grandchildren.

And, most important, don’t mix up these two distinct tasks.


Jim: I guess I’m no longer easily stunned.

I watched, many years ago, as a set of PWR license application documents arrived at the USNRC office. There were several copies, of course — but the sum total filled a full size semi-trailer to a depth of about six feet.

Now, that was stunning.


In this very long, interesting exchange, I especially appreciate Per’s clarity and Chris’, Andy’s & Alberto’s remarks.(Comment re moderator deleted.)
In any case, we indeed need to be working on multiple paths with dedication to getting the job done well & quickly — the IEA just told us
we’ve 5 years at most to get just some control of climate extremes. I believe we’ve an even bigger issue with ocean acidification, already attacking the base of oceanic food chain in Nordic waters — Google the topic in AAAS Proceedings, etc. That 70% of all human food protein comes from the sea and that we’re ~01.pH away from killing plankton in some regions should be a wake up call, even louder than sea rise or weather extremes.

In any case, the arguments for proceeding with IFR, MSR & AHTR are clear. The regulatory hurdles must be addressed quickly, in the sense that Congress allocates funding for the necessary expansion of NRC & DoE research & rules development staff. Without that, we’re toast. I specifically discussed the issue with Jaczko some months ago when he was here at Stanford — NRC does nothing they’re not told they can do and given $ to do. You likely know this all too well. It appears folks here know there needs to be a concerted effort to present a few best choices in a unified, substantive way, to all relevant members of the agencies & bodies who together must indeed be convinced to act effectively, now. There’s been no time left for a decade or so, which the IPCC Fall 2009 Copenhagen Diagnosis document made clear. And, for those unaware, JFK was told how to solve our nuclear energy (and emissions-reduction) needs in 1962 via breeders — we were to have 700GWe by 2000, no coal, no gas, no LWR generation….

We screwed up. The urgency for responsible action has increased every day from 2000.

As to LFTR, we all know the ORNL MSRE was very successful & being readied for a Thorium-fluoride breeding blanket when funding was cut. A 17,000-hour full-power test of any generating system speaks volumes. Most all the docs are at

So, reading those reports, as an engineer, I .see MSR/LFTR as being about as far along to demonstration prototype as most any other advanced machine. This is because the essential features of good engineering were already investigated & implemented 40 years ago. The 1969 teardown report (ORNL-TM-2009-181) & Grimes’ wonderful reports on reactor chemistry are excellent examples of what was nearing a prototypical demonstration product.

The Chinese have discovered this too, and as you may know, the chinese Academy of Sciences allocated about $1B early this year to take ORNL’s MSR work and develop it to prototype by 2020.

Good. Maybe their progress will scare us into getting our proverbials moving. I agree with the assessments that MSR/LFTR is about as ready as any advanced reactor, because its features were designed for safety, economy, efficiency, simplicity & as competent a group of scientists & engineers as we now have working on any other reactor designs. With the additional choice of external Thorium breeding blanket, or the 1-fluid denatured state mentioned by David, there’s no reason our country’s science & engineering communities can’t come together to make it an equal candidate with IFR, etc. for immediate implementation — we can use the Chinese competitive threat to help us define “immediate” & awaken legislators about the implication of failure for both the US’ economy & its technical standing in the world.

My last comment is on “safety”. As a Sierra Club member, UCS, NRDC, NAPF… supporter, I know firsthand how hard it is to urge support for nuclear power, of any kind, despite its unmatched safety record around the world — Chernobyl or no Chernobyl. That record must be exceeded to gain meaningful support. Thus, indeed molten lead, sodium, even salt, are flags. Pressurized systems encompassing radioactive materials are flags. Wastes & incoming fissiles are enormous public flags. Many of us promote MSR/LFTR because those flags are tinier for it, some even absent. A reactor that in Fukushima would have already dumped its core fluid by powerless gravity to safe underground storage would have made as much of a non-story of the Tohoku quake/tsunami as would TEPCO actually listening to engineers (in my acquaintance) saying backup power must be above tsunami levels — reactor 6 actually had one such generator and was fine. We know TEPCO was documented often as corrupt & foolish, but that’s not what average folks see as the disaster’s cause — they see “nuclear”.

To be accepted for safety means using Ma Nature, not simply engineer’s minds to invent work-arounds. Average folks & exploitive media folks have very different standards for nuclear in relation to coal, gas, etc. — they/we are accepting of many thousands of deaths/year due to simple sources of plain old fire.

You obviously haven’t visited BNC in pre-moderated times or any other unmoderated/loosely moderated blog.


Leave a Reply (Markdown is enabled)

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s