At the height of the nuclear war scare in the late 1950s and throughout the 1960s, American and Soviet military planners engaged in a spate of "civil defense" planning. This hole-digging mentality, parodied most famously in Dr. Strangelove (and possibly the best, most incisive episode of The Twilight Zone), meant digging nuclear fallout shelters and creating structures capable of surviving the blast of a nuclear warhead. Civil defense planners found that it is possible to design a structure at a reasonable cost to survive a nuclear explosion on the order of one megaton. To withstand a larger explosion, the design rapidly became prohibitively expensive. NORAD's famous Cheyenne Mountain complex, for example, can withstand a five to seven megaton blast within a 2 mile radius. And that thing is dug into a goddamn mountain. The moral of the story, of course, is that no matter how much "defense" the Americans or Soviets built, all the other side had to do was build a bigger bomb. Five to seven megatons? Great; hey NORAD, here's our 10 megaton warhead.
So safety can be engineered up to a certain point, beyond which additional safety is theoretically possible but economically impractical. The result is that almost any system (excepting those that are fail-safe or passively safe) is vulnerable to an adverse event that is either unforeseeable or highly unlikely. The World Trade Center was designed to withstand a hit from a 707, but not fully fueled 767s. New Orleans was engineered to survive a glancing hit from a Category 3 hurricane, but not the accompanying storm surge. The Banqiao Dam was designed to withstand a 1-in-1000 years flood, not the 1-in-2000 years flood that breached it in 1975. Your car's frame, restraints, and airbags are designed to protect you in a normal highway accident, but not if you drive head-on into a concrete pillar at 100 mph. History tells us that you can plan for a lot but it is neither possible or economically feasible to plan for everything even with the best intentions.
Fast forward to today and note what is happening in Fukushima at the nuclear facility. The reactors were designed to withstand an earthquake, but not an 8.9 earthquake followed by a tsunami. It would be easy enough to write this off as an unforeseeable event for which no reasonable enterprise could prepare. A disaster of this scale certainly supports that logic. Unfortunately, if we look more closely we see that a chain of human and engineering mistakes undermine the attempts to make nuclear power safe.
Two caveats. First, I am a fan of nuclear power, at least in comparison to coal- and gas-burning generation. Most liberals are reflexively against it (more on that in a moment). Second, I'm not a certified expert on this topic, but merely someone who has done a lot of reading. If at any point I misrepresent something feel free to correct me.
Boiling Water Reactors (BWR) like Fukushima are an obsolete technology from the late 1960s. BWRs, like any other nuclear reactor that relies on a supply of pumped coolants to prevent overheating, are inherently dangerous. Recent events illustrate this. Fukushima has backup generators to provide power to the cooling supply in the event of a grid failure, but…what happens if the backup generators are also damaged? That is the question no one asks during planning. It is the equivalent of "What happens if the Russkies build a bigger bomb?" The answer is always "Well, then I guess we're fucked." Someone translate that into Japanese, please.
Fukushima is better designed than Chernobyl (which was the RMBK-type reactor that was so dangerous Brezhnev couldn't even give them away to Third World countries) in that its defense-in-depth is stronger. What they share in common is two alarming design flaws. First, control rods have to be inserted mechanically from the bottom, which is not possible when power fails. Were they lowered in from above, gravity could do the work even in the absence of power. Second, the reactor continues to produce an incredible amount of heat after it has shut down. The reactors in Fukushima were long ago shut down, yet they continue to require extensive cooling to keep them subcritical. Modern nuclear technology does not replicate these flaws and thus, in my opinion, is a viable alternative to fossil fuel power. However, 99% of the operational reactors were built in the 1970s and feature the inherent limitations of that outdated technology.
The last step in any catastrophe is almost always human error. In Fukushima – again, this is my moderately informed opinion rather than fact – the people in charge attempted to save the economic value of the reactors rather than immediately recognizing the magnitude of the crisis and initiating their last-ditch safety measure: flooding the reactors with boron carbide and seawater, which would cool but also destroy them for good. They attempted less extreme measures – running the normal cooling systems on battery power, etc. – so that the reactors could be used to produce power again in the future. Accordingly, by the time they initiated the last resort plan involving seawater the reactors were already too hot, partially melted (as evidenced by airborne cesium), and beyond the point at which they could be cooled without adverse consequences if at all. It was, in a word, shocking to hear that 24 hours after the quake the Japanese authorities had yet to flood the reactors; the consequences are now apparent and will be increasingly so in the coming days.
Aside from the immediate tragedy – workers and residents exposed to radiation, thousands of gallons of radioactive liquid waste produced, etc. – the saddest thing about this is that it takes nuclear power off the table for a few decades much as Three Mile Island and Chernobyl did in the 1980s. It can be safe, but not with archaic 1960s technology that is fail-deadly and full of design flaws. Reactors operate here in the U.S. and around the world with very small margins of error. They are dangerous, and their "safe" operation depends on the assumption that nothing will happen to the reactor beyond what it is engineered to withstand. The more problems emerge from old Generation II designs, the lower the odds that advanced, passively safe, low-waste Gen IV reactors will ever go on line.