On Answering Questions

I got into an argument last night on Twitter about hazards around the Zaporizhzhia Nuclear Power Plant. The other person argued that I should have emphasized worst cases and had a couple poorly thought out. What I try to do in my posts on such things is to answer the question “What is likely?” I think that’s what most people want to know.

“What is likely?” is different from “What is the worst thing that can happen?” It also takes more work to answer. Probabilities, quantitative information, and subjective judgments all enter in. Unfortunately, a large number of nuclear experts, particularly those opposed to civilian nuclear power or nuclear weapons, have used the worst-case approach. It’s time for them to change.

There’s a lot of quantitative or semi-quantitative information available on ZNPP. Much of it can be treated as “more than” or “less than” something else. For example, there is much more containment at ZNPP than there was at Chernobyl. Probabilities can be arranged in a rough order.

The worst thing that can happen requires multiple failures in the system. The Chernobyl reactor, under normal operating conditions, contained design flaws that were failures waiting to happen. A poorly thought out experiment at a time of night when few other operators were available set off the cascade. The earthquake and tsunami at Fukushima, along with poor engineering against them, set off multiple failures there.

Multipoint failures are less likely than single-point failures. That puts the worst thing that could happen at the low end of the probability scale.

Could the worst thing happen? Yes. But analyses that rely too heavily on that word must be questioned. One of the articles my interlocutor served up was modeling of what would happen if all the radionuclides at ZNPP were aerosolized and released. The editors had the good sense to include, up front, the disclaimer “Editor’s note: The simulations described in this article do not reflect a current or predicted situation in Ukraine.” The article did not say how all the reactor cores, all the spent fuel would be vaporized and all the containment opened.

I usually leave out worst cases, because they are unlikely and because I know that others will bring them up. In a formal hazard assessment, I would present them and several other scenarios and their relative probability. But that’s not what people are asking about ZNPP. They are asking how a nuclear incident would affect them.

I’ve been gently chided for leaving out worst cases, and in last night’s bad-faith argument, told that I believe the worst case to be impossible. No, just unlikely, and I focus on what is more likely.

Worst-case fearmongering has skewed people’s perception of what is possible and has fanned fears of radiation, whose hazards can be hard to understand. People who are fearful do not make good decisions. Fearmongering has been successful in helping to generate a broad opposition to nuclear power plants. It has been less successful in eliminating nuclear weapons.

With global warming, we must consider nuclear power as an option for supplying electricity. I have stayed out of that argument because it is so bitter and so divided. I think I am seeing signs that the worst of that may be subsiding. There are real issues for and against nuclear power, as for other sources of electricity.

Reactor designers learn from the accidents, so comparisons to earlier situations will always be at least slightly out of date. Control panels were improved after Three-Mile Island. Nobody manufactures reactors of the Chernobyl RBMK design, and existing RBMKs have been retrofitted. Inherently safe reactors that shut themselves down are being designed in the wake of Fukushima.

And, of course, having a war around a nuclear plant could contribute to failure modes. Best not to have one at all. I’m looking at you, Russia.

Cross-posted to Nuclear Diner