¿ìè¶ÌÊÓƵ

French nuclear engineers are planning to simulate nuclear accidents
on a far greater scale than those that happened at Three Mile Island and
Chernobyl. In April, they intend to use the Phebus experimental reactor,
located in a sparsely populated area about 50 kilometres northeast of Marseilles,
to start a unique series of tests that will provide more information about
what goes on inside a nuclear reactor during a serious accident.

France generates three-quarters of its electricity from nuclear sources,
using pressurised water reactors (PWRs). Phebus, a 40 megawatt swimming
pool reactor, will simulate a faulty PWR, and six tests over six years will
assume that not only will the PWR core melt, but that safety procedures
(such as automatic shutdown and backup cooling systems) or anything to limit
the consequences of a meltdown (such as the containment buildings that house
reactors) also fail partially or totally. The experiments will be carried
out on fuel rods one-fifth the normal length, pre-irradiated to build up
the content of fission products in a Belgian reactor.

The simulation programme, called Phebus PF (for produits de fission,
or fission products), is being run by France’s Institute for Nuclear Safety
and Protection (IPSN) at the 1625-hectare Cadarache Nuclear Research Centre,
where the Phebus reactor is sited. Its partners in the project, which will
cost an estimated 900 million francs ( £100 million) over 10 years,
include the European Community’s Joint Research Centre at Ispra in northern
Italy, Japan’s Nuclear Power Engineering Corporation, the US Nuclear Regulatory
Commission, the Korea Atomic Energy Research Institute and Canada’s CANDU
Owner’s group.

The aim is to give researchers more reliable data for evaluating the
environmental impact of fission products released after serious accidents.
The amount, timing and type of radioactive materials that escape into the
atmosphere after an accident is called the ‘source term’. ¿ìè¶ÌÊÓƵs try
to calculate the source term for a variety of accident scenarios at different
nuclear installations so they can advise local authorities how to prepare
for a nuclear disaster. But Maurice Haessler, head of laboratory research
at Phebus PF, says it is impossible to determine just how accurate these
source terms are because it is difficult to validate the results of the
models used.

The American nuclear industry first started to research the source term
for a variety of accident scenarios, using mathematical models, in the late
1960s. The IPSN itself used a similar reactor to Phebus, called Scarabee,
at the Cadarache centre in the early 1970s to examine how nuclear fuel interacted
with its surrounding steel cladding during a meltdown in a sodium-cooled
fast-breeder reactor. About a decade ago, the accident at Three Mile Island
prompted more studies to calculate source terms for PWRs in the US, France,
Sweden and Germany.

These experiments, however, have all used separate models to deal with
cores, fission products, cooling circuits and containment buildings. Phebus
PF is unique because it is designed to simulate not only what is taking
place within the core but also the path by which fission products would
leak through the cooling circuits to the containment building before escaping
into the atmosphere.

There are two ways a nuclear reactor can go wrong. One is through an
increase in the number of neutrons bombarding the uranium fuel (caused,
say, by withdrawing the control rods in the pile), which releases a burst
of energy and causes overheating. A PWR core can also overheat if its coolant
stops flowing and all the automatic systems fail to restore the cooling
circuit. Even worse, both processes could occur together. In such situations,
the core temperature can shoot from 325 °C to 2850 °C in 10 minutes.

In 1979, a valve failed and cooling water stopped flowing at the Three
Mile Island PWR plant. However, the fission products remained safely inside
the core. At Chernobyl in 1986, a surge of energy at low power made the
cooling water in the core boil; the fuel heated up and finally melted.
The resulting explosion and fire blew the roof off the reactor and spewed
some 962 x 10 ¹⁶ bequerels of fission products into the environment:
the reactor had no containment building.

In the worst-case scenarios being simulated at Phebus PF, the coolant
flow to the core is to be interrupted. The heat produced by the radioactive
fuel boils the coolant left in the core. In the resulting steamy atmosphere,
the fuel rods heat up, burst and melt. Some 111 x 10 ¹³ bequerels
of radioactive fission products such as caesium-137, iodine-131 and strontium-90
are carried into the reactor’s primary cooling circuit, which leaks or fails,
releasing radioactivity into the containment building. That also leaks,
releasing radioactivity into the environment.

But Phebus will not irradiate the Marseilles region. The experiments
will study 20 fuel rods at a time, which is 0.02 per cent of the inventory
of a 900 megawatt PWR. Even so, Haessler says that the site follows the
same three-safety-barrier philosophy as most working reactors. The first
layer is a series of zirconium oxide and zircalloy shells. This assembly
is isolated from the core of the reactor by an in-pile cell with zircalloy
and nickel alloy barriers. And finally, the building housing Phebus has
half-metre thick reinforced steel and concrete walls to contain radioactivity
released during the experiments. These walls are designed to withstand the
impact of a 5.7 tonne Lear business jet.

The original Phebus reactor, built 14 years ago, simulated only part
of the core of a 900-megawatt reactor. The Cadarache team has now almost
completed modifications, which included building structures to model cooling
circuits and the containment building. A separately connected metal alloy
containment vessel simulates the containment building of a full-size reactor,
so that Phebus can simulate the entire sequence of events in a faulty power-generating
reactor.

During the tests, the fuel rods will be heated to between 2000 and 2500
°C (or higher) at a rate typical of a severe accident. At such temperatures,
the fuel is damaged and the fission products escape. Such conditions will
produce aerosols of radioactive particles, which will be swept out of the
core’s protective shroud by steam and travel into the primary cooling circuit
containing a steam generator and pressuriser where the walls will reach
temperatures as high as 1000 °C. In some experiments they will then
travel into the 10 cubic metre containment vessel. Finally, any gases left
over will be collected and retained in a larger tank.

Each of the six tests lasts only a few hours, but the analysis far longer.
To find out what (and how much) is going where, Phebus is equipped with
some 400 sensors. In the core, they will monitor temperature, pressure and
flow rates. The primary cooling circuit has sensors at the intake of the
steam generator and pressuriser, and just outside the containment vessel.
Gamma detectors and aerosol samplers will measure the amount and type of
radioactivity released, while other sensors measure deposits on the pipe
walls and the composition of liquids. The containment vessel is equipped
with a similar set of sensors.

The meticulous analysis of the fission products escaping from the core
during the Phebus PF experiments should reduce much of the uncertainty in
the calculation of source terms for accident scenarios within reactors.
And although Phebus models a PWR, the data can be applied to other types
of nuclear reactors.

There are, however, still unanswered questions about the safety of Phebus
PF. And the final decision on whether the programme can start in April has
still to be taken.

Following meetings in June and October 1989 with the permanent group
of experts on nuclear safety attached to the French Nuclear Installations
Safety Directorate (DSIN), the IPSN submitted a preliminary report outlining
the Phebus PF experiments, proposed modifications to the site and safety
precautions that would be taken. The DSIN group, which includes experts
on nuclear safety from industry, universities and the government, responded
favourably to the plans, and in November 1991 the French environment and
industry ministries approved the project, in principle.

Last August, the Cadarache team submitted a second report on the modifications
that have been carried out and which must be approved before the experiments
can begin. The IPSN is expecting the final decision from the government
by next month.

Besides major modification to Phebus for the experiments, the IPSN
has strengthened the underground sections with steel-reinforced concrete
to conform to new norms for major earthquakes. The Cadarache site is not
far from a fault line and Phebus PF must be able to withstand the effects
of an earthquake measuring 9 on the Medvedev Sponheuer Karnik scale (equivalent
to the surface moving at Cadarache by 20 to 40 centimetres per second).
The region has an estimated annual risk below 1 in 10 000 of such an earthquake.

But one of the advisers to the DSIN group of experts says there are
other considerations. For instance, as the zircalloy cladding around the
fuel melts during the experiments it will react with the water coolant to
produce hydrogen gas. This will then escape, along with the radioactive
fission products, into the containment vessel. If any oxygen is present,
there is a risk of explosion. There is also a risk of a vapour explosion
in the core as the water coolant and molten fuel come into contact.

He says what is still lacking is knowledge about exactly how the meltdown
during the experiment will be controlled. The experimental risks have still
not been precisely defined; nor has anyone calculated a source term for
the experimental reactor. Once this is done, safety authorities will decide
if the risks are too great. He favours approving the six tests separately
so each can be evaluated before the next is carried out.

Haessler says that a lot of experience has been gathered from previous
nuclear accident research, and regards Phebus PF as a ‘safe’ experiment
– although not perfectly safe. ‘In France we borrow the English expression
‘Never Say Never’ when talking about nuclear accidents,’ he says. ‘Otherwise
there would be no need for this type of research. It is logical to take
precautions and then add an extra one – in case of human error.’