Sunday, April 12, 2009

LFTR Cost may run as low as 1 Billion per GWe of generating capacity

Yesterday I argued that a klawed 1970s computer model for Light Water Reactor costs had mislead the ORNL staff members who fail to see the competitive cost advantage of MSR technology over LWR technology. I argued that the technological advantage of the MSR still held. ORNL-TM-7207 assumed that the cost relationship between MSRs and LWRs assumed by ORNL-4541 (1971) was still valid, but the estimated cost of LWRs offered by ORNL-TM-7207 is far lower than what was actually the case in the index year of 1978. ORNL-TM-7207 stated that its cost estimates were based on computer modeling of reactor costs, rather than actual construction costs. However, it retrospective it is obvious that computer models of 1970's reactor costs used by ORNL was seriously flawed, and failed to capture reactor price increases between 1970 and 1980. The MSR design indexed for the DMSR was based on the the 1970 MSBR, and unlike LWRs had not undergone significant evolution. Such was not the case for the LWRs of the 1970's. And indeed in 1980 when ORNL-TM-7207 was published, the design of LWRs was continuing to evolve as the result of the Three Mile Island accident.

Thus the comparable price of LWRs and MSBRs ca. 1970 could not and should not be assumed for 1978 and even less so for the post Three Mile Island environment. The reason for this lack of evolution in MSR design can nor be fully explained by a lack of MSR design research. At least one major MSR design study emerged after the publication of ORNL-4541. But it should be noted that the MABR design had been relatively fixed at ORNL during the 1970's. There were obvious reasons why this was the case. Compared to the LWRs of 1970, 1978 or for that matter 2009, the MSBR arguably represented a significantly better form of nuclear technology. The evolution of the LWR since 1970 represents an unsatisfactory attempt to bring the LWR up with the potential of the MSBR. The MSBR was under no pressure to evolve, because it set a mark which the LWR could not equal at any price.

I believe that the case I make about the post 1978 relationship between MSBR and LWR costs requires more proof. As I noted the cost relationship between the two, set out in ORNL-TM-7207, is based on a data set found in ORNL-4541. An alternative approach to verifying the cost relationship I suggested yesterday, would be to reprice the ORNL-4541 cost estimates by determining 2009 prices. This would be a serious project, and would require much more that my once over lightly approach, but it would yield a far more definitive estimate of 2009 LFTR costs.

I believe that it is this cost advantage that lies at the heart of the case for a LFTR based energy future, I believe that this argument can and should be subjected to further tests. In particular fairly detailed design studies of two and one fluid MSR designs published by ORNL in 1979 and 71 contained fairly detailed cost information.

Until 1967 ORNL had envisioned a two fluid commercial MSR design. However, that year ORNL Reactor chemist discovered the Bismuth Protactinium recovery process which allowed thorium breeding in a single fluid core reactor. The single core reactor approach was considered advantageous By ORNL because it eased core graphite problems. The improvement in graphite performance was considered desirable enough to justify a switch from a one fluid 1 GWe commercial design.

ORNL MSBR had considered the possibility of both 1 GWe and 250 MWe modular two fluid designs. Both contain data that is relevant to LFTR costs. ORNL-4528 is especially interesting because it describes a facility powered by 4 250 MWe MSRs. In addition we have the 1962 Sargent & Lundy Report CAPITAL COST EVALUATION, 1000 MWe MOLTEN SALT CONVERTER REACTOR POWER PLANTS. This study was prepared under contract from ORNL. Unfortunately the Sargent & Lundy report was the most detailed, and it was perpared when thinking about large MSR power generation projects was thew least evolved. I am pointing these studies out because they serve as good starting points for determining LFTR costs.

Sargent & Lundy estimated the construction cost of the MSR to have been $65,481,000 1962 dollars, say 460 million 2009. No one who knows about power generation capitol cost would take that figure seriously. ORNL-3996 suggested 103 million in direct construct costs and another 30 million in indirect costs, and $110 million in interest, all in 1966 dollars. ORNL-3996 appears to have attempted to find a more realistic picture of commercial MSR costs., and we get a total cost of $243 Million 1966 dollars, or 1.6 billion 2009 dollars.

ORNL-4528 estimated facility costs to run to $146 million, about the equivalent estimate for a 1 GWe LWR facility was estimated to cost in 1967. ORNL-4541 estimated capitol costs for a 1 GWe MSR was 202 Million 1970 dollars, including 30 million accumulated interest during construction. Assuming another 200 million in interest we get a total price tag of 400 million to tax payers. That translates to 2.64 billion 2009 dollars. I take note of the fact that my Thursday calculation of MSR cost were off by a billion dollars or so, no doubt because the DMSR study did not include the cost of interest in their calculations. So we have a little reason for confidence in our ORNL-4541 cost estimate. There is also egg on my face as a consequence of my little calculation error of Thursday. I don't suppose you would believe me if i said that it was the first time I had ever made a mistake? The second time? Oh well.

ORNL-4528 looks like it calculated contingencies too low, way too low probably and underestimated interest accumulated during construction. That gets us a figure that is about 10% less that the ORNL-4541 cost estimate. There are other ample opportunities for fudging on the 1967 cost estimate, thus ORNL-4528 and ORNL-4541 are assuming the same universe of capital costs. This is not terrribly shocking since Roy C. Robertsons name went on the front cover of ORNL-3996, ORNL-4528, and ORNL-4541, and Ed Bettis's name went on the front cover of the first two.

I realize that I am pushing my analysis to a level of obtuseness that has long sent me readers packing, but there is a point here, and one that would make a nice master's thesis for a nuclear engineering student. That point is that in the 1962 to 1970 ORNL cost estimates for MSR power plants were not just off the top of the head guesses, and thus would be serious starting points for for attempts to better determine LFTR costs.

Lets go back to the 2.64 billion calculation, the inflation adjusted costs of the 1970 ORNL-4541 cost estimate. Let us assume that factor production decreases overall cost by 20% and that the rapid manufacture decreases accumulated interest by 2/3rds. Further let us assume that by recycling old power plant sites we are able to save 5% on capital costs. Finally assume a 30% cost savings due to the learning curve enjoyed by serial manufacturing of hundreds of LFTRs. That brings our capitol costs down to the neighborhood of 1.2 Billion dollars. This is lmost too good to be true, and needs more work,

I do see my goal, however. Lets see what we can reasonably say. First that ORNL made a realistic study of commercial MSR and noted that MSR costs at that time would have been similar to LWR costs. That during the 1970;s LWRs underwent design changes that made them more expensive, but there was no similar design change for the MSR. That given inflation in 2009 the cost of MSRs would probably run about half what LWRs cost if the LFTRs were built in the same way LWRs are built. However factory built LFTRs would cost significantly less than custom built MSRs because of the labor cost advantage of factory manufacture, the decrease of accumulated interest because ofrapid manufacture, the cost savings of recycling old power plants and the finally the learning related cost savings related to serial manufacture. Taken all together, we have a potential power generating product that appears may cost less than 2 Billion dollars per GWe of generation capacity, and it is not impossible that factory built LFTRs might cost as little as one billion dollars per GW of generating capacity.

There is a well documented data set derived from research don at ORNL in the 1960's. Further research using the ORNL data set and recent materials and manufacturing cost data, might refine the potential LFTR cost picture.

I intend to clean up my analysis, and make it a little easier to read, and present the repackaged text very soon.


Anonymous said...

You make a strong case the LFTR based on the ORNL history. I keep looking for a down side. One is that to take full advantage of the higher temperature, a yet to be perfected gas turbine is needed. A second is that lithium fluoride appears to be a promising salt, but tritium maybe a problem. Possibly technology exists to sequester tritium? A third problem is the graphite moderator which is affected by the neutron flux. I have seen proposals for graphite spheres which could be replaced in a manner similar to pebble bed fuel. Also, I have read that the French are looking at operation without a moderator. There is beauty in simplicity. Some place I saw a proposal for the use of synthetic diamond. Chemistry of the space age has produced several novel carbon structures. A pilot plant is needed to resolve the moderator issue. Since degraded graphite could result in positive nucleation, it is best to solve this problem first as the public demands higher safety standards for nuclear power than other power sources.

Charles you have propose replacing the energy source of our existing coal plants with multiple factory built LFTRs. I assume that you would settle for a secondary steam loop and a reduced efficiency. You pointed out the such a conversion would cost much less than a new build. If we could get some form of carbon tax, it is likely that nuclear conversion would attract interest. Thanks for your continued promotion of LFTR. I appreciate your effort and I rarely miss any of your blogs.

John Tjostem

Anonymous said...

The cost of reactor production and deployment is a function of the system assumptions that are made.

For example as discussed often in the past, the most cost effective system is one in which a sealed U232/U233 driven nuclear battery is deploy to retrofit existing coal fired plants. This reactor type is unattended, with no moving parts or maintenance requirements during its service life and deployed underground, cemented into a steel reinforced pit at or near existing coal power plant steam lines. The production cost of such a process heat only reactor will certainly be under $1000/KW when manufactured on a automated mass production line.

Either one or a few very large U233 production/breeder reactor(s) will provide the fuel for such a fleet of small reactors numbering in the thousands and whose cost is prorated over their total number that is deployed. The large reactor will also burn waste and recycle blanket products as necessary from the nuclear battery fleet. Taking advantage of the economics of scale, the large multi-core reactor with a pool type fuel production blanket may produce up to 20 gigawatts of electrical power.



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