When evaluating the empirical statistics and potential for hazard of safety failure of a nuclear reactor (such as an uncontrolled criticality excursion or loss of coolant/loss of containment), significant consideration needs to be given to not only the immediate impact of such an accident, such as explosive damage or direct irradiation effects, but also the persistence of such effects in the environment and the range of effects. Pointing to Chernobyl as only having directly killed approximately fifty people, for instance, glosses over the persistent radiation hazards, accelerated mortality from long-term low level exposure, the cost of evacuating and relocating the population of the nearby city of Pripyat, total remediation costs, and enduring economic impact to the Soviet Union and the Ukraine in the form of contaminated areas that are no longer arable, health impacts and property damage, et cetera. The damage to the Soviet Union is almost universally regarded among historians and economists as a significant contributor to the ultimate economic and political collapse of that nation, to the extend that it largely dominated Gorbachev’s later tenure.
Consider, too, what the impact could have been had the Chernobyl facility been located not in a remote area of the Ukraine but adjacent to or upwind of a major metropolitan center. The location around Chernobyl has been essentially sealed off with only a modest impact upon available property, but had such an event occurred near a major population center the economic impact could have been enormous, dwarfing the estimated US$20B for loss and clean up of the collapsed World Trade Center buildings. The estimates of the economic impact and projected remediation costs for Tokyo are already in the tens of billions of dollars, and this is without any evaluation of persistent radiation or other long-term costs. Certainly, petroleum refineries can catch fire, coal mines can explode, and oil wells can leak or burst, but the persistence of all of these, as massive as they can be, pales in comparison to what the long-term effects can be from massive radiation release or leakage.
Nuclear power generation facility safety design and risk mitigation steps are based upon what what are called design basis accidents, i.e. the at-present best engineering judgment as to the likely worst-on-worst conditions the reactor and containment systems are likely to experience in the operational lifetime. However, if that threshold is exceeded–as in the case of a massive seismic event that is beyond facility code experience to withstand–all bets are off; the provisions that would normally prevent or limit catastrophic failure may no longer be adequate, as with Chernobyl #4 fire or the ongoing events at the Fukushima facility. Because they are beyond anticipation doesn’t mean that they are inately unlikely to occur or that the impact shouldn’t be considered, any more than our prior ignorance of the likelihood of catastrophic meteorite impact has protected civilization from impact.
A brief perusal of other threads on this topic reveals a great deal of speculation and no small amount of fundamental ignorance on the practical operation and function of nuclear fission power generation masquerading as self-proclaimed expertise. Large scale power generation facilities are not like small research or mobile reactors, where SCRAMing reduces function to negligible levels and the decay heat can be readily carried away. Large fission cores can be both highly sensitive to modest tweaks and yet resistant to attempts to reduce activity once hey have exceeded a criticality threshold, which is why the first researchers who worked on fission weapons and later energy production referred to control of the fission reaction as “twisting the dragon’s tail”.
Diagrams in textbooks or glossy pop-sci magazines give the impression that this is all a tightly controlled process of shooting a neutron into a nucleus and receiving a specified result like an expert bowler performing a strike; the reality is that the entire process is stochastic, and there are a lot of phenomena that can influence the criticality threshold that have been mostly learned by trial and (occasionally lethal) error. While modern Generation III reactor designs have implemented the lessons learned from previous failures, it would be very ostrich-like to baldly assert that such reactors are immune from any future failures, particularly those that result from a condition that is beyond the design basis accident envelope.
Despite the current knee-jerk response by Germany and other nations to abandon or roll back nuclear fission power production, the reality is that current capabilities in renewable sources will not fill the gap created by dwindling supplies of hydrocarbon fuels. Nuclear fission power will be part of any realistic energy planning for the foreseeable future. But that doesn’t mean that there aren’t some very significant risks, costs, and drawbacks to the entire nuclear energy production cycle, from fuel refining through waste disposal. To justify not considering those impacts on the basis of “it’s just a political problem” or “won’t happen because it is too unlikely” glosses over the fact that our understanding of those probabilities is non-rigorous and based upon a lot of assumptions.