THE SAFETY RISKS
OF USING MIXED-OXIDE FUEL
IN VVER-1000
REACTORS:
AN OVERVIEW
Edwin S. Lyman, PhD
Scientific Director
May 20, 2000
Introduction
The United States and Russia are in the final stages
of negotiating an agreement that would establish a framework for the disposal
of approximately 34 tonnes each of plutonium withdrawn from nuclear weapons
programs. According to the U.S. Government,
the agreement stipulates that 25 tonnes of weapons-grade plutonium (WG-Pu) in
the U.S. and 33 tonnes of WG-Pu in Russia would be incorporated into mixed-oxide
(MOX) fuel and irradiated in existing nuclear reactors. The MOX fuel would be substituted for a fraction
of the low-enriched uranium (LEU) that normally fuels the reactors.
The remainder of the WG-Pu (9 tonnes in the U.S. and 1 tonne in Russia)
would be combined with high-level nuclear wastes after being stabilized in a
ceramic matrix, through a process known as immobilization.
The agreement calls for development
of an infrastructure in both the U.S. and Russia that would be initially capable
of processing 2 tonnes of WG-Pu per year. As a result of its reluctance to declassify the isotopic composition
of WG-Pu, Russia will add about 12% reactor-grade plutonium (RG-Pu) to the WG-Pu
before making it into MOX, for a total of 37 tonnes under the agreement.
Therefore, a capacity of about 2.3 tonnes of total plutonium per year
in Russia will be required.
The plutonium disposal infrastructure will consist
of three main facilities: a plant to
convert plutonium warhead components into oxide powder, a MOX fuel fabrication
plant and an immobilization plant. In
addition, modifications to the nuclear reactors designated to receive MOX fuel
may be required.
Recognizing that more WG-Pu may in the future be declared
surplus to the weapons program, the agreement will also require that a plan
for increasing the annual disposal capacity in both countries to 5 tonnes of
WG-Pu be developed within a year. According
to the director of the Office of Fissile Materials Disposition (MD) of the U.S.
Department of Energy (DOE), Laura Holgate, both governments hope that the capacity
expansion plan will be completed in sufficient time so that it can be implemented
immediately, and the 2 tonne per year stage can be bypassed.
Funding for implementation of the agreement in Russia, which according
to current estimates will cost nearly two billion dollars, has not yet been
secured, but ultimately will have to be provided by the U.S. and other Western
nations.
To meet the initial MOX disposition target, the U.S.
is planning to use four pressurized-water reactors (PWRs) in the U.S. --- two
at the McGuire nuclear station in the state of North Carolina and two at the
Catawba nuclear station in the state of South Carolina. Both nuclear stations are owned by Duke Power,
a private utility. In order to use MOX,
the operating licenses of these plants will have to be amended by the U.S. Nuclear
Regulatory Commission (NRC). A second
utility, Virginia Power, was originally part of the private consortium that
received the MOX contract, but withdrew in March of this year, stating that
its participation no longer made economic sense for the company.
Russia plans to use all seven of its operating VVER-1000
reactors (four at Balakovo, two at Kalinin and one at NovoVoronezh), which are
similar in concept to Western PWRs, as well as the BN-600 fast neutron reactor
at Beloyarsk. Russia will require more
reactors to process a similar amount of MOX fuel than the U.S. will because
it is planning to use a lower MOX core fraction. In order to increase the rate of MOX disposition
of Russian plutonium per the follow-on agreement, either the core fraction in
the Russian reactors would have to be increased or Russian MOX fuel would have
to be exported to other countries such as Ukraine, Western Europe and Canada.
The preferred alternative of the Ministry of Atomic Energy (Minatom)
is to build a generation of new fast breeder reactors, known as BN-800s, but
the U.S. continues to oppose this alternative, which poses additional proliferation
risks.
Increasing the amount of WG-Pu to be immobilized is
another alternative for increasing the disposal rate that would be safer than
the MOX options. However, Minatom vehemently
opposes this approach because of a misguided belief that WG-Pu should not be
"thrown away." As this report
will show, Minatom's insistence in using MOX fuel could well have disastrous
consequences for Russia and Eastern Europe.
The Risks of MOX in Western PWRs
The substitution of MOX for LEU in
LWRs raises serious safety risks that have not been adequately assessed in either
the U.S. or Russia. These risks, which
apply broadly to both WG-MOX and RG-MOX, can be grouped into two categories.
First, the probabilities of certain severe accidents
may increase when MOX is used. Fundamentally,
the introduction of MOX fuel into LWRs reduces the effectiveness of the materials
used to absorb neutrons in the core, such as the control rods and the boron
dissolved in the coolant. This makes it more difficult to control the nuclear reactions in
the core and reduces the margin available to safely shut down the reactor if
problems arise. At the same time, the
"delayed neutron fraction," a parameter which determines the speed
at which the power level of the reactor responds to changes in conditions, is
smaller when MOX fuel is used. This
means that the operator not only has less control over transients but has less
time to respond to them as well.
Measures can be taken to restore some of the lost control
worth, such as increasing the number of control rods or the concentration of
boron in the coolant. However, these
are only partial fixes and may introduce other problems as well.
Second, the consequences of a severe accident (as measured
in latent cancer fatalities and early fatalities from acute radiation exposure)
involving containment failure or containment bypass (i.e. a steam generator
tube rupture) will be greater if MOX fuel is in the core. This is because MOX cores have higher concentrations
of actinides, including isotopes of plutonium, americium and curium.
Most of these are alpha-particle emitters with large radiotoxicities
if inhaled or ingested.
A recent report by the Nuclear Control
Institute (NCI) analyzes a number of these issues with regard to WG-MOX use
in U.S. PWRs.
[1]
This report calculates that in the event of
a severe, Chernobyl-type accident with containment failure or bypass, at a PWR
with a 40% WG-MOX core, the number of latent cancer fatalities would be about
25% greater than for a PWR with an all-LEU core. Depending on the population density in the vicinity of the accident,
this could correspond to hundreds to thousands of additional cancer deaths.
The report also finds that the use
of MOX could have serious negative effects on other aspects of PWR operation.
Three examples are:
Overcooling transients and
pressurized thermal shock. Certain types of accident initiators, such as a break in a main steam
line, can result in a rapid temperature drop in the primary coolant system.
In PWRs, when the temperature of the coolant drops, the reactivity increases
(the "moderator temperature coefficient" is negative).
Therefore, even if the reactor is immediately "scrammed" (that
is, the nuclear reaction is stopped by inserting control rods), the reactivity
and core power will increase and automatic safety systems must be activated.
One potential outcome of an overcooling transient is
a severe accident known as pressurized thermal shock (PTS), which is of particular
concern with respect to VVER-1000s (see below). PTS can occur when a steel reactor pressure
vessel has been embrittled by exposure to neutron radiation. If the reactor vessel is cooled down below
a certain temperature (which depends on the vessel material, the radiation exposure
time and many other factors) while it is still pressurized, small cracks can
rapidly expand and cause the vessel to split open. Once this happens, cooling of the fuel cannot
be assured and a meltdown is likely. In
addition, as the pressure vessel breaks apart it can also breach the containment,
resulting in a large radiological release. After a main steam line break, active intervention
by operators is required to prevent the reactor from entering a state at which
PTS can occur.
There are a number of ways in which MOX fuel can increase
the risk that an overcooling transient will progress to a severe accident.
First, because the moderator temperature coefficient is more negative
in MOX cores and the delayed neutron fraction is smaller, the rate at which
the power increases will be greater, reducing the time margin for activation
of safety systems.
Second, the risk of PTS will be higher if MOX fuel
is in the core, since the rate at which the temperature decreases will be greater.
This is because the decay heat of MOX fuel is smaller than that of LEU
fuel immediately after the reactor is scrammed.
An analysis of the main steam line break accident by Westinghouse shows
that the temperature drops into the range at which PTS can occur after six minutes
for a partial MOX core, whereas for LEU fuel the dangerous temperature range
is avoided entirely.
[2]
MOX will also lead to an
increase in PTS risk by accelerating the neutron embrittlement of the reactor
pressure vessel, since plutonium-239 fission produces a greater number of
"fast" neutrons that have a high probability of escaping from the
core than uranium-235 fission. While
reactor operators may try to minimize this problem by placing MOX fuel elements
away from the reactor vessel (i.e. near the center of the core), this creates
additional power-peaking problems, since the neutron flux is higher near the
center of the core. This has caused
fuel management problems in at least one nuclear plant in Germany. It is unclear whether this difficulty can be
adequately resolved.
Reactivity insertion
accidents. In a hypothetical accident in which a
control rod is ejected from a PWR core, a sudden increase in reactivity will
result within adjacent fuel assemblies.
This can cause rapid increases in temperature and power in fuel rods
that can cause the cladding to burst open or, at a higher level, the fuel
itself to fragment. At a minimum, this
could cause a release of fission products into the primary coolant. At worst, the expulsion of fuel particles
could block the flow of coolant, leading to higher temperatures, more fuel
damage, and a potential meltdown.
It has now been documented that MOX fuel of the type
produced by the French company Cogema, after being irradiated for more than
three annual cycles, is more vulnerable to failure during a reactivity
insertion accident than LEU fuel at a similar burnup. (Cogema is a key participant in the U.S. MOX disposal consortium
and will design and operate the U.S. MOX fabrication plant. It is anxious to play a similar role in
Russia as well.) This conclusion
results from the Cabri test reactor experiment series in France, in which a MOX
fuel pin that had been irradiated for four annual cycles underwent a severe
failure, whereas an LEU pin with similar characteristics did not fail.[3]
The failure is understood to be a consequence of the
inhomogeneous nature of MOX fuel, which contains plutonium-rich clumps. These clumps experience extremely high local
burnups and consequently accumulate large quantities of fission gas in small
volumes. When the fuel pin is subject
to a reactivity insertion, the gas can be rapidly released, resulting in
fragmentation of the fuel and rupture of the cladding. Two other experiments with MOX at lower
burnups were carried out as well.
Although these rods did not fail, the fission gas released into the gap
between the fuel and cladding was considerably higher than that seen from LEU
fuel rods at similar burnup. Production
of inhomogeneous fuel is a fundamental flaw of the French MOX fuel fabrication
process.
Loss-of-coolant accidents
and station blackouts. Probabilistic risk assessments
carried out for Western PWRs find that loss-of-coolant accidents (in which a
rupture in the primary coolant system reduces the amount of water available to
cool the core), and station blackouts (in which off-site power is lost) are
among the largest contributors to core melt and early containment failure. The likelihood that these events will lead
to such severe consequences is related to the extent to which damage occurs to
the fuel cladding before the emergency core cooling systems are activated. This in turn strongly depends on the
centerline temperature of the fuel rods.
The centerline temperature of MOX fuel rods is about
50C higher than that of LEU fuel under similar
operating conditions. As a result, the
rate of cladding corrosion during the initial stages of a loss-of-coolant
accident will be more rapid for MOX fuel, and faster-acting safety injection
systems may be necessary to prevent fuel damage and core melt.
The Risks of
Using MOX in VVER-1000s
According to the International
Atomic Energy Agency, compared with Western-style PWRs, the VVER-1000 has a
number of design features which make it more vulnerable to experiencing severe
accidents that could lead to radiological releases. These include design of the plant layout, which does not ensure
sufficient physical separation of diverse safety systems to prevent common-mode
failures. These design flaws are compounded
by a number of other issues that affect safety, including the type and quality
of materials used in construction, the reliability of instrumentation and
control systems, the inadequacy of the maintenance during the current difficult
economic situation and the lack of thorough and independently reviewed safety
documentation (accident analyses, emergency operating plans and severe accident
management guidelines).
Although Western nations have joined in efforts with
states in Eastern Europe and the former Soviet Union to correct some of the
deficiencies in reactors of Soviet design, most of the emphasis has been on
reactors of the RBMK or VVER-440 type, which are considered to be of higher
risk than the VVER-1000 series. In no
instance, however, has the West committed to financing upgrades of
Soviet-designed reactors to conform to Western safety standards, an endeavor
which has been estimated to cost hundreds of millions of dollars (billions of
rubles) per reactor.
To the extent that safety margins
are smaller in VVER-1000s than in Western PWRs, the further reduction in margin
associated with use of MOX will be of even greater concern for VVER-1000s.
With regard to the additional public
health consequences associated with MOX use, the results for VVER-1000s should
be very similar to those cited above for U.S. PWRs --- about a 25% increase in
latent cancer fatalities following a severe accident with containment failure
or bypass. The relatively high concentration
of actinides in MOX cores compared to LEU cores is a generic feature of
light-water reactors. A recent study by
the Kurchatov Institute calculated that at a burnup of 60,000 megawatt-days per
tonne, a WG-MOX VVER-1000 fuel assembly would have around twice as much
plutonium and around three times as much americium and curium as an LEU
assembly.[4]
These results are nearly identical to those we found for U.S. PWRs.
In fact, the consequences may be
even greater than those calculated for U.S. PWRs. This is because Minatom intends to mix WG-Pu with about 12% RG-Pu
to obscure the isotopic composition of Russian WG-Pu. This will increase the quantity of higher plutonium isotopes in
the unirradiated MOX fuel to be used in Russia. As a result, the concentrations of americium and curium isotopes
in the irradiated fuel will be higher than if WG-Pu was used, and the
associated radiation exposures following an accident would be higher.
Because the most severe consequences
of an accident at a MOX-fueled VVER-1000 would result from a failure or bypass
of the containment, unresolved issues associated with VVER-1000 containment and
steam generator integrity will become even more urgent if MOX fuel is
used. The Nuclear Energy Agency (NEA)
has raised questions about the long-term integrity of the pre-stressed concrete
containments at VVER-1000s, and has cited the need for research "to
develop a model which could predict the loss of performance as a function of
loss of pre-stress...".[5]
The risk of containment bypass is a
significant concern, especially with regard to the integrity of the steam
generator collectors. A leak in a steam
generator establishes a pathway from the primary to the secondary coolant
systems and hence to the environment.
VVER-1000 steam generator collectors experienced numerous failures in
the mid-1990s, and despite temporary fixes the NEA states that "assurance
of the VVER-1000 steam generator collector integrity remains one of the most
important safety issues."
A number of the deficiencies of
VVER-1000s should be of particular concern with regard to the use of MOX
fuel:
Pressurized thermal shock. Foremost among them is the embrittlement of the reactor pressure
vessels (RPV) and the risk of pressurized thermal shock (PTS).
According to the NEA, one of the most urgent open
issues for VVER-1000 safety is an evaluation of the integrity of the reactor
pressure vessel "which takes into account the actual condition of vessel
material under operating conditions."[6] The radiation resistance of RPV materials is
highly sensitive to slight variations in the composition of the steel used in
the vessel and welds, as well as reactor-specific details of the operating
history. For instance, some VVER-1000
vessel weld joints contain more than 1.5% nickel and have significantly reduced
radiation resistance. The NEA cites in
particular the lack of a detailed understanding of the conditions to which the
weld areas are exposed during reactor operation, the lack of non-destructive
assay methods that could locate micro-cracks in the RPV and assess the extent
of its embrittlement, and the lack of verified PTS assessment methods.
To cope with this problem, Russian
engineers have developed a process known as "annealing," in which
heat treatment is applied to the RPV and welds to restore their crack
resistance. However, the benefits of
annealing have not been decisively verified and it is not known how long they
last. In addition, annealing itself can
have a detrimental effect on the RPV because it can cause "sensitization"
of stainless steel, a process that significantly reduces resistance to
localized corrosion.
Compounding the uncertainties in the
integrity of VVER-1000 RPVs is the lack of adequate analyses of overcooling
transients that could initiate PTS, such as the main steam line break.
The IAEA reaffirmed its concerns
regarding VVER-1000 RPV integrity and PTS risk in the report of a 1999
conference.[7] The author has also been privately informed
by an official from the IAEA's Department of Nuclear Safety that all VVER-1000
RPVs are embrittled. Moreover, the
official has little confidence in the effectiveness of the annealing procedure
to mitigate this problem.
Given these concerns, the potential
increase in the risk of pressurized thermal shock as a result of using MOX fuel
must be fully evaluated. Since this
risk is associated with plant aging, it will continue to become more acute if
VVER-1000 lifetimes must be extended to accommodate the plutonium disposal
program.
Reactivity insertion
accidents. There are differences between VVER-1000 fuel
rods and PWR rods which could exacerbate the problems seen in high-burnup PWR
MOX rods during the Cabri reactivity insertion accident tests. For example, unlike Western PWR fuel
pellets, VVER-1000 pellets have a hole in the center. The hole was intended to reduce temperatures in the center of the
fuel pellet and to provide extra volume for fission gas to accumulate, thus
reducing the internal rod pressure.
However, in practice this does not seem to work. Because the center of the fuel pellet is the
hottest region, the gas accumulating there reaches a higher temperature than is
typical in PWRs, so that "the internal fuel rod pressure is considerably
higher than in a PWR for a given filling pressure and plenum gas volume."[8] According to the IAEA, "the merits of
the hollow UO2 pellet design used in WWER [VVER] fuel rods are
debatable." Fission gas release
fractions in high-burnup VVER-1000 fuel rods of up to 70% have been observed,
compared to typical values of 5% or less for PWR fuel.[9]
The higher fission gas release in
LEU VVER-1000 fuel rods compared to PWR fuel rods raises questions about
whether MOX can be used in VVER-1000s without significant changes to the fuel
rod design, since one would expect fission gas release in VVER-1000 MOX to be
even higher. This would result in an
even greater vulnerability to clad rupture during reactivity insertion
accidents and LOCAs than was observed in PWR fuel in the Cabri test series.
Russian engineers claim that the cladding
used in VVER-1000 fuel rods, which is a zirconium-niobium alloy, is more
resistant to failure during reactivity insertion accidents, even for high
burnup fuel, because it does not corrode and become embrittled during
irradiation as easily as conventional PWR cladding, which is pure
zirconium. However, if true, this
property would not prevent the type of MOX fuel failure seen in the Cabri
tests, since the MOX rod that failed was not heavily oxidized.
Conclusions
The risk to public health and safety associated with
using MOX fuel in both U.S. PWRs and Russian (and Ukrainian) VVER-1000s is
considerable and must be fully assessed before plans to go forward are
undertaken.
In our view, these additional health and safety
risks are unnecessary to achieve the goal of warhead plutonium
disposition. The immobilization
alternative is capable of safely increasing the inaccessibility of plutonium
without incurring the risks associated with reactor irradiation. However, short-sighted technocrats and bureaucrats
in both the U.S. and Russia, encouraged and lobbied by nuclear industry
representatives from Western Europe and Japan, appear determined to load MOX
fuel into reactors no matter how large the risk. The only hope of stopping this dangerous program lies in the
principled resistance of the residents of communities which would be most
directly affected by a Chernobyl-type accident in their backyards.
------------------------------
Edwin Lyman, Scientific
Director of the Nuclear Control Institute (NCI) since 1995, received a PhD in
physics from Cornell University in 1992.
Prior to joining NCI, he was a researcher at Princeton University's
Center for Energy and Environmental Studies.
[1] Edwin S. Lyman, Public Health Consequences of Substituting Mixed-Oxide for Uranium Fuel in Light-Water Reactors (Washington, DC: Nuclear Control Institute, 1999), accepted for publication in the Princeton University-based journal Science and Global Security.
[2] Westinghouse Electric Corporation, Implementation of Weapons Grade MOX Fuel in Existing Pressurized Water Reactors, DOE/SF/19683--7, report prepared for the U.S. Department of Energy (Pittsburgh: Westinghouse Electric Corporation, 1996), 3-30.
[3] J. Papin, F. Schmitz and B. Cazalis, "Further Results and Analysis of MOX Fuel Behavior Under Reactivity Accident Conditions in CABRI," in the Proceedings of the 27th Water Reactor Safety Information Meeting (Washington, DC: U.S. Nuclear Regulatory Commission, 1999), 355.
[4] Margaret B. Emmet, Calculational Benchmark Problems for VVER-1000 Mixed Oxide Fuel Cycle (Oak Ridge, TN: Oak Ridge National Laboratory, 2000), 85-86.
[5] Nuclear Energy Agency, Safety Research Needs for Russian-Designed Reactors (Paris: Organization for Economic Cooperation and Development, 1998), 30.
[6] Ibid, 25-28.
[7] International Atomic Energy Agency, Report of the International Conference on Strengthening Nuclear Safety in Eastern Europe, 14-18 June 1999, Vienna.
[8] International Atomic Energy Agency, Design and Performance of WWER Fuel, Technical Reports Series No. 379 (Vienna: IAEA, 1996), 76.
[9] Ibid, 42-43.