Radiation at Fukushima No. 1 plant has surpassed legal limit, TEPCON, briefly reaching 1204 micro sievert via Kyodo

An excellent explanation of what is happening at Fukushima is available here

How do nuclear power plants work?

In fission, atoms are split in half, which generates heat; some materials that are created during this process will decay for some period of time, and they also generate heat. This heat is used to generate steam. A pressurized water reactor (water does not boil) uses a heat exchanger to generate steam in a separate chamber (a two-stage process). A boiling water reactor is a one-stage process; when it is operating, the core is kept covered in boiling water and the steam is collected above that level.

What happens when you shut down a nuclear power plant?

The self-sustaining (critical) atomic reaction that runs a reactor can be shut off in a matter of seconds; the goal is to keep the reactor sub-critical. The core material, which is radioactive, will generate heat, at a decreasing rate, for some time. Therefore, nuclear engineers need a way to keep this heat from building up in order to protect the radioactive fuel (rods) and the reactor.

The General Electric Mark I reactor design used in the Fukushima plant was determined to have a possible containment problem, should a meltdown occur, in 1986.

The unit in question is a General Electric Mark I reactor design, also called a “pressure suppression” system. It’s a pretty common model; of the 104 reactors in the US, 23 are this type. But there are concerns that design doesn’t necessarily provide the best containment in the event of a meltdown.

If they are unable to cool the core and it does melt through the reactor vessel, this model doesn’t have most robust containment system, says Bergeron. If it did meltdown and the reactor core slumped to the floor, it likely “would result in containment failure in less than a day,” he says.

Critics of nuclear power say that this has been a known issue for decades and these reactors aren’t designed to sustain a meltdown. Greenpeace’s nuclear policy analyst Jim Riccio writes:

In 1986 Harold Denton, former director of NRC’s Office of Nuclear Reactor Regulation, again acknowledged this vulnerability while speaking to utilities executives at Brookhaven National Laboratory. Denton noted that, according to NRC studies the GE Mark I reactors had “something like a 90% probability of that containment failing.”

Live Broadcasting by Ustream