Not having read the book itself, I'd like to clarify a point for the record. The general public has a definition of "risk" in mind that's only the likelihood of an untoward or catastrophic...
Not having read the book itself, I'd like to clarify a point for the record.
The general public has a definition of "risk" in mind that's only the likelihood of an untoward or catastrophic occurrence. The U.S. NRC's effort in Probabilistic Risk Assessment, as illustrated by the Rasmussen graph, rather deliberately left this common understanding undisturbed. We know that a nuclear power plant accident could be a terrifying event, but sleep soundly enough when told it's vanishingly unlikely.
The engineering definition of risk can be summarized as:
risk = probability of occurrence x consequences arising from occurrence
The book review gestures at this difference when it discusses the potential $10 trillion cost of evacuating Tokyo, had the Fukushima incident caused its maximum potential damage. [I'll submit that even this may be an underestimate, given the potential for longer term injuries and economic damage due to the contamination of food sources, land, and water; industrial losses; health costs; and extraterritorial radiation spread.]
Normal Accidents is an incredibly useful (though somewhat dated) book which discusses risk in terms of complex systems, with the then-current Three Mile Island incident in mind. It talks about coupled events - i.e. the tendency for one problem to greatly increase the likelihood of other problems around it. The prior Tildes topic about the Husky Superior Refinery explosion is an example. It wasn't only the cracking tower hydrogen explosion - the high-velocity debris release caused a puncture in a distant tank. The punctured tank flooded the whole refinery grounds with burning tar and put still more tanks at risk. Nuclear power plants are full of closely coupled systems where a single failure can let loose a herd of subsequent ones, which then trigger more - the "cascade of rice grains" mentioned in the Safe Enough? review.
At the end of the day, there are safer nuclear power generators currently in use and under development - smaller scale, modularity and standardization to avoid unique construction/design risks, passive safety systems that can function without external power, fail-safe designs, etc. Nuclear power will never be zero-risk, but we're in a time of terrible risk trade-offs due to carbon dioxide pollution.
This is a review by an anonymous author in Astral Codex Ten’s book review context. From the review:
This is a review by an anonymous author in Astral Codex Ten’s book review context. From the review:
Wellock sets out to tell this history, how the US public went from nuclear-lovers in the 1960s to suspicious in the early 1970s, hostile in the 80s, and ambivalent today.. Wellock does not try to hoodwink us with happy talk - he makes clear what the stakes are in nuclear energy, that in the case of Davis-Besse there was not simply a power plant at risk, but the potential to release radiation across America's industrial heartland. Wellock tracks regulatory victories for the nuclear industry, and expensive defeats at the hands of activists, and, always, political posturing over its future.
Yet "Safe Enough?" is less of a history of events than a biography of an idea, the birth of "Probabilistic Risk Assessment" as the guiding principle for understanding and mitigating risks in complex systems. The heroes of Wellock's book are not nuclear plant night shift assistant supervisors, or the Nuclear Regulatory Commission training and assessment specialists, though they each make important cameos. The city of Toledo, Ohio is not safeguarded by watchful superheroes. It is protected by a methodology.
Not having read the book itself, I'd like to clarify a point for the record.
The general public has a definition of "risk" in mind that's only the likelihood of an untoward or catastrophic occurrence. The U.S. NRC's effort in Probabilistic Risk Assessment, as illustrated by the Rasmussen graph, rather deliberately left this common understanding undisturbed. We know that a nuclear power plant accident could be a terrifying event, but sleep soundly enough when told it's vanishingly unlikely.
The engineering definition of risk can be summarized as:
risk = probability of occurrence x consequences arising from occurrence
The book review gestures at this difference when it discusses the potential $10 trillion cost of evacuating Tokyo, had the Fukushima incident caused its maximum potential damage. [I'll submit that even this may be an underestimate, given the potential for longer term injuries and economic damage due to the contamination of food sources, land, and water; industrial losses; health costs; and extraterritorial radiation spread.]
Normal Accidents is an incredibly useful (though somewhat dated) book which discusses risk in terms of complex systems, with the then-current Three Mile Island incident in mind. It talks about coupled events - i.e. the tendency for one problem to greatly increase the likelihood of other problems around it. The prior Tildes topic about the Husky Superior Refinery explosion is an example. It wasn't only the cracking tower hydrogen explosion - the high-velocity debris release caused a puncture in a distant tank. The punctured tank flooded the whole refinery grounds with burning tar and put still more tanks at risk. Nuclear power plants are full of closely coupled systems where a single failure can let loose a herd of subsequent ones, which then trigger more - the "cascade of rice grains" mentioned in the Safe Enough? review.
At the end of the day, there are safer nuclear power generators currently in use and under development - smaller scale, modularity and standardization to avoid unique construction/design risks, passive safety systems that can function without external power, fail-safe designs, etc. Nuclear power will never be zero-risk, but we're in a time of terrible risk trade-offs due to carbon dioxide pollution.
This is a review by an anonymous author in Astral Codex Ten’s book review context. From the review: