Monday, February 16, 2009

The learning curve for serial reactor production

One assumption of the Aim High concept directly challenges an assumption of the conventional nuclear industry. That is the Aim High concept assumes that there are major cost advantages for serial reactor manufacture at factories rather than custom onsite reactor manufacture. This argument might be challenged by reference to reported Chinese cost projections for factory produced Pebble Bed Modular Reactors that were recently discussed by Brian Wang. The Chinese do not assume that early factory built PBMRs will be lower cost than on site manufactured reactors. The Chinese envision manufacturing hundreds of PBMRs in their reactor factory. Thus it would appear that the Chinese do not anticipate that PBMR production will lead to significant savings. Not even the serial production of 248 PBMRs lead to significant cost advantages for PBMR. The Chinese do anticipate a significant 30% to 40% per unit savings from experience based learning.

What then is responsible for the Chinese cost data? First, it should be observed that the PBMR has extremely low power density. Thus the PBMR pressure vessel is the same size as that of a Light Water Reactor which produces 5 times the power output. Thus the PBMR will require greater materials input for a given unit of power output than the LWR. This would suggest that PBMR components may be assembled in the factory but not the entire reactor. Even relatively small PBMRs would be too heavy and bulky for truck or rail transportation. Thus reactors would be assembled on site from factory manufactured kits. The Chinese probably anticipating on-site construction of LWRs from factory produced kits, so actual manufacturing conditions for PBMRs would not differ greatly from those of Chinese LWRs. It is also argued that PBMR simplicity would lead to lower reactor costs, but a glance at PBMR design suggest that it might not be all that simple compared to LFTR design.
Wang refers to production plans for about 250 reactors. While this is a very large number, it would not be large enough to justify the installation of large labor saving production machines. If reactos are manufactured in kit form at the factory, there would be no assembly lines. Thus many of the cost saving advantages of factory production would be lost.

The Chinese appear to anticipate that the primary savings from factory manufactured PBMRs would come from the learning curve. This would suggest low capital costs, since it would be assumed that the capital cost per unit would decrease over time as loans were paid off. Indeed financial costs, taxes, insurance and contingencies account for around 20% of Chinese reactor costs. This combined category is not significantly higher for Chinese built LWRs than for PBRs. We can assume then that manufacturing techniques for Chinese PBMRs do not differ significantly from those used to manufacture Chinese LWR, and that materials inputs will be, if anything more expensive per KW of electrical output than would be the case for Chinese LWRs. It would appear then that the cost advantages of serial production do not greatly outweigh the cost advantages of economies of scale.

Financial costs are a far less significant cost factor than they would be for reactors built in the other countries. Because financing costs are low to begin with the shorter PBMR construction time would not be translated into a significant cost savings. Finally any savings in PBMR labor cost would probably be balanced by the greater cost of materials.

It would not appear then, that Chinese PBMR costs would not provide us with a comparative insight into LFTR costs. Recent reports from China indicate that the Chinese are paying about $1.60 per watt for new reactors. This cost would appear to be less than half of the cost of reactors in Europe or North America. Recently Indian reports anticipate costs as low as $1.40 per watt for Indian manufactured LMFBRs. This would indicate a significant post-carbon electrical cost advantage for the emergent Asian economic super powers. This advantage does not stem from a potential cost savings in design or manufacturing techniques. Thus roads to greater nuclear cost competitiveness are still open to the North American, European, and North East Asian economies.


Anonymous said...

Just take a close look at your PBMR figure, you will see how much complication TRISO fuel handling, the high pressure helium system and its overall low power density characteristics contributes to the overall complexity of the PBMR approach. So many ugly system design tradeoffs have been made to support the perceived proliferation resistance of TRISO fuel.

The use of TRISO fuel is now pervasive throughout all governments reactor design efforts in one form or another. Major size, cost, and performance tradeoffs have been made and are continuing to be made by very smart men who should know better. New designs have TRISO fuel cooled by molten salt. So why not just use fuel dissolved in the salt? It would make for a very, small simple and effective design, easily built in a factory.

If you want to work in the government, you need to conform to the perceptions, goals, and fears of the bureaucrats. For a reactor designer it’s conformed or drive a taxi.

So why is it happening? Fear of proliferation in the government of course. Proliferation is very hard for the politicians to understand. Anything that is very hard and complicated, no matter how important, doesn’t do well in the government.


Charles Barton said...

It is curious then that the Chinese would fall into the same trap. Part of Making progress in solving energy issues is "the rectification of names" - calling things what they really are.

Anonymous said...

"Recently Indian reports anticiapte costs as low as $1.40 per watt for Indian manufactured LMFBRs"

Charles, have you got any reference for this statement?

Charles Barton said...

"With the Rs.35-billion PFBR project progressing at good pace at Kalpakkam, 80 km from here, the Indian government has sanctioned building of four more 500 MW fast reactors."
Rs.35 billion = $700 million for 500 MWs. that would be $1.40 per Watt.

Anonymous said...

It is curious then that the Chinese would fall into the same trap.

In the late 80s or early 90s, Andrew C. Kadak Professor of the Practice, Nuclear Engineering at MIT wanted to restart nuclear power around the world. He and his students looked at all existing nuclear reactor designs and decided on the pebble bed design as the most politically acceptable reactor technology; see

The molten salt reactor was rejected because of the political problem of fuel salt reprocessing.

The Chinese academic community became part of the PBMR design effort network. Kadak worked with them to develop there prototype which became the foundation of their PBMR reactor effort.

As the twig is bent so the tree shall grow.


Anonymous said...

Just curious, given the fact the power density of the reactor, if I understand correctly, is a restricting feature in terms of factory manufacture, modularity and definitely costs, I ask you two questions :

1) what' s the typical power density of a moderated (graphite, for example) thermal spectrum LFTR
2) what does it change in case of non moderated, epithermal-fast spectrum LFTRs, as those proposed from a French group (the so called TMSR, thorium molten salt reactor with no moderator operated in a fastish spectrum)
If I understand correctly, in the first case life of the graphite concerns (which indeed becomes high level waste at the end of its life) are going to reduce the power density, while I guess that a non-moderated fast spectrum design avoids this limit, correct if I'm wrong (even if a little of graphite is always needed, if I understand correctly)

Pratically, besides the graphite HLW intrinsically avoided production, I tended to prefer the non moderated option for the (slightly) better breeding and transuranics transmutation features, but I'd like to know your opinions about this point, I think it's a very interesting and important point

Charles Barton said...

There is a clear advantage to the use of graphite. The amount of fissionable material required to sustain a chine reaction declines significantly. There are ways to deminish the graphite problem, for example using LFTRs in load following or peak reserve capacity increases graphite core life. I believe that annealing radiated cores at 2000 C reverses the neutron damage to graphite. Thus it would seem possible to recycle graphite cores. I also believe that it is possible to seal graphite from penetration by some fission products. This would not be the case for tritium, but tritium has a relatively short half life.

I would favor graphite cores if large scale LFTRs deployment is considered because it existing supplies of U-235 and plutonium would be start a large number of graphite core LFTRs,thile far fewer unmoderated LFTRs could be started without large scale uranium enrichment.

Anonymous said...

But, at last, what's the typical power density of a LFTR core? And does it change anything from a non-moderated fastish spectrum to a graphite moderated LFTR version?

As far I know, initial fissile start-up is not a great concern for countries which have many operating nuclear plants like Us,Japan and many European countries, because they are still plenty of TRU wastes
If there are no other issues, for example material corrosions or salt solubility in faster spectra, I' d tend to support rather the faster version of LFTR

Charles Barton said...

Alex I am not the go to person for this question. I would suggest that you ask Kirk Sorensen of David LeBlanc, for a more detailed account. As to the amount of fissile material needed to start up a LFTR, I believe that it amounts to a ton in a moderated 1 GWe LFTR. In a unmoderated LFTR it would take several times that amount. I am too lazy to look up the exact number. The problem is that most of the electrical generating capacity in the United States is devoted to peak generation. This means that we need to find fissile material to operate several times the average annual electrical output of the United States.

Do the math yourself. Find the weight of a start up charge needed for an unmoderated 1 GWe LFTR. Count up how many 1 GWe LFTRs would be needer to provide American Peak demand capacity. Multiply the two, and then start looking for start up charges.

Anonymous said...

Thanks, Charles

One number I read is a few hundreds of kg per GWe for the moderated Lftr and about 1,5 tonn per GWe for the non moderated version; based on this, there is plenty of transuranics waste to feed the fissile start-up for the first worldwide hundreds or thousands GWe of Lftr (TRU production in LWR is about 250 kg per GWyear).
Of course, it' s not clear to me if there are other safety and reliability issues (corrosion, salt solubility, etc..) with faster spectra LFTR versions

Charles Barton said...

Alex, My father who was involved in MSR research at ORNL for 20 years, rorked on both MSR prototypes, including a patent on the fuel formula for the first prototype, and who did a lot of research on protactinium extraction, favored the a moderated 2 fluid MSR design. He was undoubtedly familiar with all of the compromises and trade offs involved in MSR design.

Robert Hargraves said...

Regarding the nonproliferation aspects of TRISO fuel, I note that page 71 of this IAEA document ... 50_web.pdf
illustrates how to reprocess TRISO fuel.


Blog Archive

Some neat videos

Nuclear Advocacy Webring
Ring Owner: Nuclear is Our Future Site: Nuclear is Our Future
Free Site Ring from Bravenet Free Site Ring from Bravenet Free Site Ring from Bravenet Free Site Ring from Bravenet Free Site Ring from Bravenet
Get Your Free Web Ring
Dr. Joe Bonometti speaking on thorium/LFTR technology at Georgia Tech David LeBlanc on LFTR/MSR technology Robert Hargraves on AIM High