I will get to my particular reasons for working on a small modular IFR at the end, but first let me review some underlying considerations.
1. Technically small reactors are very sound. If we can build 1000 Mwe plants, then 10 Mwe or 100 Mwe plants are that much easier. So I find a small reactor design, such as 4S (Super-Safe, Small & Simple)is technically sound and well designed. There is no magic about a long-life core. If you start with a derated core, say 6 times more fuel compared to convential 5 year core, then you have 30 year life with the same burnup level. Neutron damage on cladding will be higher, but since the netron flux level is lower at a derated core, the neutron dose increase is less than a factor of two, which is doable. However, this requires a good neutron economy, which metal fueled fast reactor provides. [By the way, the TWR is based on the same principle, except the moving part, which complicates the cooling arrangements.
2. However, small reactors do not fare as well when it
comes to economics. My own experience will illuistrate this
point amply. About 20 years ago, I had a chance to work
together with a potential reactor vendor in designing a 10
Mwe terrestrial reactor for a client (I believe it was Air
Force) with potentially large number of orders. A detailed
design was developed with a preliminary cost estimate. The
proposal was well received by the client, who then issued a
RFP for a fixed price contract. When the vendor worked with
component suppliers to come up with a fixed price, it
turned out many times the original estimate. Ovbiously the
client's interest quickly evaporated. Small reactors cannot
compete with the economies of scale.
3. Can factory manufacturing of multiple units overcome the
economies of scale? All nuclear plant components are
factory manufactured -- small or large. Even large pressure
vessels are now forged in few pieces. Furthermore, modular
construction approch allows more and more factory
assembling -- small or large. If manufacturing 100 small
components repeatedly can lower the unit cost, so would
manufacturing 10 large components repeatedly. So I do not
buy a faster learning curve arguments.
4. Large reactor projects are too complex causing delays?
This was the perception in the late 80s, prompting
Westinghouse to develop AP-600. AP-600 was the first and
earliest design certified by NRC, but no takers. Hence
AP-1000 was developed, but the first customer, China
demanded uprate to 1400 Mwe. All 26 reactors that utilities
have applied for NRC licenses are large units in the
1400-1800 Mwe. Large units have been built without delays
in Japan and Korea, and now in China.
5. Small grid system demands small reactors or incremental
expansion? In the U.S. essentially all grids are tied to
regional systems and interconnected. That's why all present
reactor oders as indicated above are large units, which
represents a small fraction of the national grid of more
than 500 Gwe. Small utilities can buy into portions of a
large unit. For example, a small utility serving San
Antonio bought 20% equity position (and electricity) of the
NRG's South Texas project. In overseas market, small grid
countries like Vietnam and Indonesia indicated interests in
nuclear but they are looking at 1000 Mwe class for economic
6. How about remote sites or dedicated power source for
industrial process heat application? There could be a niche
market, but still the bottom line will be the generation
cost. Small reactors cannot compete with gas turbines or
other fossil plants that can be sized to the demand.
In spite of my prejudice, I believe there might be a case
for small modular reactors of the PRISM size, namely
300-600 Mwe. The economic penalty is only nominal but
offers other intangible advantages. Below that level, if we
go down to 10-50 Mwe range, the economic penalty will be
too high to be practical.
Why then did I work on a 50 MWe IFR design? The motive was
simply to initiate a project to maintain the technology
base and train next generation enginners required for the
future. For this purpose, almost any reactor size will do
but a small reactor project cost will be more affordable.
"To get the kind of breeding gains that enable a 7-year doubling time,
you need a mighty high breeding ratio, and to get that, you need a
mighty fast spectrum and super-rapid reprocessing.
Since one of the tenets of IFR/PRISM is no separated plutonium, please
help me understand how you're going to accomplish this? Please also
explain how you're going to keep the reactor controllable in this hard
spectrum, since resonance absorption (Doppler effect) is really your
only self-control mechanism, and you're above all the resonances in
energy in this hard spectrum.
Softening the spectrum to make the reactor controllable has been what
every LMFBR around the world has had to do to make the reactor even
mildly controllable, and this kills off the breeding ratio really fast."
Response by Dr. Yoon Chang:
The metal fuel used in the IFR, due to its high density,
results in a most hardened spectrum and the best neutron
economy (more excess neutrons that can be used for
breeding). Some of these excess neutrons leak out of the
active reactor core but captured in the external blankets
to convert depleted uranium into plutonium. The harder the
neutron spectrum, the higher the breeding ratio.
Parenthetically, the neutron economy (excess neutrons) is
dictated by fissile isotope and spectrum, and Th/U-233
cycle has a worse neutron economy than U/Pu cycle in fast
spectrum, whereas the opposite is true in thermal reactors.
Even then, achieving a breeding ratio of unity in thermal
spectrum is a great engineering challenge.
Super-rapid reprocessing is not necessary to achieve the
7-year doubling time. A two-year ex-core inventory is
already accounted for in the doubling time calculation. We
have two years to reprocess and refabricate.
In the IFR pyroprocessing, all actinides including Pu, Np,
Am, Cm, etc. as well as some U and rare earth fission
products (trace amounts) are recovered in a single product
stream and electrorefining is incapable of separating out
Pu from the rest of actinides. The blanket actinides are
rich in Pu and less minor actinides. However, actinides
from the blanket will be mixed with those from the driver
fuel in the electrorefiner or in the injection casting
fabrication furnace. Hence, separated plutonium cannot be
produced and the entire reprocessing and refabrication are
carried out in the same hot cell, with no accessibility.
The excellent neutron economy also implies that the excess
reactivity requirement to overcome the reactivity deficit
by fuel burnup is minimal. Hence the reactivity control by
control rods (with neutron poisons) is also minimal and the
accidental reactivity insertion events can be dealt with
simple design features. Doppler reactivity feedback is
smaller by about 20-30% compared to oxide fueled fast
reactors. But that is still more than adequate to deal with
prompt reactivity requirements. What is most important is
the overall temperature and power reactivity coefficient.
When the coolant temperature rises or the power increses
for whatever causes, the IFR responds with a negative
reactivity feedback due to coolant density or structure
expansion, which tends to reduce the power and hence the
Even in worst case accident events (loss-of-flow and/or
loss-of-heat-sink without scram like TMI-2 or Chernobyl
initiators), the initial coolant temperature rise will
cause thermal expansion of fuel assemblies which increases
neutron leakages, and hence the power is brought down all
by itself without operator actions or safety systems.
Ironically in these events, as the inherent feedbacks try
to bring down the power, the Doppler feedback actually
contributes positive reactivity. (Recall that Doppler was
necessary to protect against inceasing power. When power is
coming down, it tries to raise the power.) This feature is
unique only with the IFR. The metal fuel operates at low
temperature because of a high thermal conductivity (a
factor of 10 higher than oxide), so the stored reactivity,
(Doppler coefficient) x (temperature difference), is too
small to override the negative feedback due to coolant
temperature rise. In other words, it's the temperature
difference rather than Doppler coeffient itself that
enables this unique inherent safety. Therefore, in IFR the
Doppler feedback is adequate to deal with overpower
transients, and at the same time it enables inherent safety
features in the other extreme accident conditions.
Current Fast Reactor Construction Projects
Russia resumed the construction of BN-800, primarily
driven by the weapons Pu disposition application.
India is constructing a 500 MWe Prototype Fast
Breeder Reactor (PFBR), to be on-line in 2010.
Subsequently four more units of the same size are
planned in two sites by 2020.
China is constructing 65 MWth/20 MWe China
Experimental Fast Reactor, to be on-line in 2009.
Follow-on 800 MWe prototype FBR planned ~2020.
Both China and India envision rapidly growing demand
for nuclear and consider fast breeder reactors to be
essential part of their future energy mix.
There is a growing international consensus that these
five criteria are what the next-generation advanced nuclear
system must meet to be broadly acceptable for the 21st
century and beyond, namely:
* Reduce the volume and toxicity of nuclear waste.
* Keep nuclear materials unsuitable for direct use in weapons.
* Be passively safe based on characteristics inherent in the reactor design and materials.
* Provide a long-term energy source not limited by resources.
* Be economically competitive with other electricity sources.