Hacker News Reader

Comment by epistasis | Hacker News Reader

I hear about regulatory cruft, but have never heard of examples of what would be changed.

We have two recent build sites in the US, one failed entirely, and the other is hobbling along. I've read lots of analyses and postmortems, and the only regulation-related criticisms I've found are that the NRC doesn't regulate enough. By only looking for safety of the design, Westinghouse was able to submit designs that were safe, but not particularly buildable. If the regulators had checked the work of Westinghouse to include basic build ability in addition to safety, tens of billions of dollars might have been saved, and we might have been building more nuclear reactors.

But I would like to hear more specific complaints about how regulations could change, if it has the chance to improve nuclear.

View on HN

I have one that's specific to new technology.

A few years ago I got to sit in a meeting between reps from a bunch of GenIV reactor startups, and a former head of the NRC. The reactor people had one complaint: that the NRC required near-complete blueprints before they would even look at a design. It cost several hundred million dollars to get to that point, then the NRC would give a flat yes or no. If no then you were out of business, and if yes then you still just had a paper reactor.

That's a really difficult environment for investors. They said it would be a huge help just to have a multi-stage process. The NRC person was unsympathetic, said it wasn't the NRC's job to help develop nuclear technology, and brushed off climate change arguments.

Fortunately Congress has gotten involved since then and things seem to be improving a bit.

Tbh I can understand the need for near complete blueprints: Trying to judge a systems safety, based on plans of only half the system, does not sound like it would be a very useful judgement.

What's to prevent cutting massive corners after the original half plan was approved?

They weren't asking for final approval at an earlier stage. They were just asking for feedback along the way. That way they wouldn't have to guess whether their basic idea had a chance, and they could make changes to address concerns before spending hundreds of millions nailing down the details on something the NRC could have told them it considered fundamentally flawed. It would also help them with investors, when they could show favorable early feedback.

> But I would like to hear more specific complaints about how regulations could change, if it has the chance to improve nuclear.

https://rootsofprogress.org/devanney-on-the-nuclear-flop is a review of "Why Nuclear Power Has Been a Flop" by Jack Devanney.

Based on the review, the short version is that the regulators use the wrong threat model for radiation (LNT), and a regulatory model which effectively requires nuclear to be unprofitable (ALARA).

The review describes briefly what needs to be changed.

This is great, thank you!

Setting a firm limit on radiation release ahead of time, rather than one based on economics like ALARA just seems like it would be far safer. The other examples also seem like they are bad regulations that don't help safety or construction either. Would be great to see if they could result in more efficient construction.

Another way of setting a limit on radiation would be: 1/10 as much as an average coal plant per kWh. Current nuclear plants are already below this, because coal plants release a fair bit of radioactive material into the air, from trace elements like thorium in coal.

When regulations prevent deploying something 10x safer than the currently deployed alternatives, they're not making us safer.

Nuclear plants may be below this in normal operation, but what matters is the dose in accident scenarios. That's what drives things like containment buildings, which are large and expensive.

Acronyms used:

LNT stands for Linear No Threshold: cancer risk is directly proportional to dose, that doses are cumulative over time (rate doesn‘t matter), and that there is no threshold or safe dose. This contradicts studies that we have about people who received enough CUMULATIVE doses of radiation that they would be dead if it was at the same time.

ALARA: Radiation should be As Low As Reasonably Achievable. In practice this means that any cost reduction simply means freed money that you're now required to spend on safety.

Yeah, that's utterly stupid. We need to define a value per life saved and apply it across the board to safety regulations. Any safety regulation that comes in cheaper than that is required, no safety regulation that comes in more expensive than that is required. There should be no singling out of industries. All such regulations must include a reasonable description of how it can be done for that cost--which can be challenged in court. Likewise, outside groups can challenge proposals which have been rejected by showing there's a cheaper way to do it. (And this would include companies that stand to benefit. A company that comes up with a cheap enough way to implement a safety measure can get it mandated. While this could cause a monopoly situation they can't exploit it because if they make it too expensive the rule goes away.)

It looks like private enterprise is maybe incompatible with Nuclear Power then. I find this to be pleasing, and we should simply side step private enterprise and build public plants.

How did you get that from the linked article? Or are you just shooting from the hip based on bias?

> is that the regulators use the wrong threat model for radiation (LNT)

Nuclear fans like to complain about the LNT, but I don't think they're really thinking this through.

In a nuclear accident, most of the population exposure will be a minor increase spread across a vast population. At those doses, we basically cannot check whether LNT is true or not -- the small cancer incidence it predicts is statistically invisible against all the other causes of cancer.

So foes of LNT want to say "we can't show LNT is correct", which is fine, but then they say "so we must assume the actual effect is smaller, perhaps zero", which is not fine. The evidence doesn't support that second step, and this is not a court of criminal law where radiation must be presumed innocent unless found guilty beyond a reasonable doubt. One might very well argue that one should take a precautionary approach, which is to assume that low level radiation has the worst effect it could have that is not ruled out by evidence. This would imply even larger threat than under LNT.

We know the mechanism by which radiation causes damage: Particles knock into DNA, changing its structure. This causes mutations, and once too many of those accumulate, you get cancer, or your cells just stop working and you die.

This would suggest that the risk should be linear with radiation dose. But cells have DNA repair mechanisms. If a person receives a very large dose in a short amount of time, it makes sense that the repair systems might be too overwhelmed to fix all the damage that has occurred. Of course, the repair system isn't perfect, and some small fraction of damage will be permanent. This suggests that LNT should be true for small doses, but the harmfulness of radiation per particle should increase at larger doses.

A few consequences if that's true:

Depending on how the constant coefficient is determined, a fully linear health risk model will tend to overestimate risk at low dose rates, but underestimate at high dose rates.

This also suggests that radiation concentrated in a particular spot on the body is particularly dangerous. The cells in that location will be bearing the brunt of the radiation dose, so their repair systems are more likely to fail. So inhaling a bit of plutonium dust means you're in for a worse time than absorbing an equivalent dose spread out over your body.

Of course, whatever health risk model we choose, it should also be applied to the regulation of coal plants, since they put radioactive isotopes into the atmosphere as a part of their regular operation.

It's also possible that repair mechanisms can be induced, with the body making more repair enzymes under higher damage load. If so, this would imply the danger from a single particle would be higher at low doses where less induction has occurred. This is radiation hormesis, but of a kind that means LNT could be underestimating the danger of very low doses of radiation.

You could be right, maybe there's a hormetic valley somewhere in the risk curve. I think it still has to be linear close to the background radiation level, and grow faster than linear as one approaches the lethal level, though.

Isn't it true that the proportionality coefficient for LNT was calculated using cancer rates for atomic bomb survivors? The dose for them would be very rapid, and there would be little time for enzymes to be produced. This would explain why such a valley didn't show up in that particular set of data.

LNT has been checked as far down as the statistics lets it be checked (including with animal experiments). Those doses are much higher than the doses I was talking about, the doses that actually matter for nuclear accidents. It's conceivable that the slope of the effect/dose curve at very low doses could be very steep. The results at those higher doses wouldn't be able to tell.

> By only looking for safety of the design, Westinghouse was able to submit designs that were safe, but not particularly buildable.

Because the commission is biased towards safety above all else. They need to be more realistic and not cave in to fearmongering. The design needs to be safe, but against a realistic threat model.

The designs also need to be assembly line and not so highly customized. French style reactor designs are good for this reason. Part that holds the reactor? Fine, evaluate for weather and calamity resistance and build per location. The reactor? Hope you like black.

What is an example of this fearmongering, and what expense did it cause?

From the outside, you have large-scale accidents that caused a lot of fear that killed the drive for us. Notably Three Mile Island in the US and the effects of seeing what happened in Chernobyl that was drummed up by the US government to show failures by the Soviets (even though the Soviet reactor design was far more dangerous). Additionally, I live near a plant and they hold extreme safety training on iodine distribution and such to school children that makes kids fear the plant.

No denying that there's been fearmongering in general. But I'm still asking for an example of what regulation on the construction of nuclear plants should be changed.

What killed the drive for nuclear in the 1970s in the US was more that the costs came in very high, that electric power demand growth suddenly moderated, and that the grid was opened to external non-utility competition by PURPA in 1978 (four months before TMI).

How about starting with mountainous environmental impact reports that no one reads? Theres lots of room for maintaining effective oversight while making the overall processes more efficient. The problem is the current processes were designed to be onerous due to those lobbying against nuclear power.

Also with new concepts like micro reactors or reviving long abandoned technology like liquid thorium reactors that would burn what we stupidly label “waste” and if the active systems are interrupted coast to a stop on their own instead of running away like our fast breeder water based designs a lot of the existing regulations and requirements are suddenly moot.

Most importantly Nuclear is the only “clean” technology that is predictable and controllable. Until you have a way to reliably meet base load requirements, fossil fuel generation is going to continue.

So if you really do think that climate change represents impending doom, resisting nuclear power is pretty dumb. It’s not perfect - but there isn’t any technology that is perfect or without some risk. Pretending nuclear is the only energy technology with serious issues is also dumb. 50 years of people painting nuclear as the boogie man hasn’t helped either. If you strongest arguments are emotionally based those aren’t very good arguments at all.

This isn't fearmongering:

https://news.ycombinator.com/item?id=26348520

I don't see any evidence humans are smart enough to operate nuclear power safely. There are too many examples of humans not fully understanding nuclear, or just being stupid with nuclear. For example, building a nuclear power plant near an earthquake fault line in CA, or where tsunamis occur in Japan. We can't get the basics of safety right here.

Even with all the disasters included, nuclear power is safer than almost all other kinds (the exception being very large hydro plants), per unit energy.

Interesting. I know nuclear power is far safer than it's general reputation.

Is nuclear really safer than solar?

This[1] has some data and estimations for death rates measured based on deaths from accidents and air pollution per terawatt-hour (TWh), which suggests nuclear has 0.07 deaths per TWh, which is marginally higher than wind (0.04), hydro (0.02), and solar (0.02).

So, it's very close!

1. https://ourworldindata.org/safest-sources-of-energy

That's bad data. Nuke is getting blamed for the Fukushima deaths that were due to the evacuation--neglecting the fact that the safest option was to stay put. If you replace the evacuation deaths (IIRC ~500) with the stay-put deaths (most likely zero) you about halve the nuke death rate.

The larger deployment of utility-scale solar does seem to have reduced it's death rate. (Many of the solar deaths are from falling off the roof during installation or maintenance. Utility-scale solar is normally on the ground and with better safety measures.)

> Nuke is getting blamed for the Fukushima deaths that were due to the evacuation

I think this is fair. /All/ deaths from nuclear and renewable power are due to accidents and bad decisions. Accidents and bad decisions aren't going to go away. It takes a monumentally boneheaded decision to make a nuclear power plant dangerous, but apparently the rate of monumentally boneheaded decisions is one per thirty years at our current level of nuclear power usage.

> but apparently the rate of monumentally boneheaded decisions is one per thirty years at our current level of nuclear power usage.

That rate is very likely to increase as time goes on and reactors become older and thus more prone to failure/some freak low probability incident happening.

At the # of deaths produced by nuclear in normal operation, or by wind or solar, the "deaths" are dominated by the statistical lives due to the cost of energy itself.

The NRC uses a value of $9M for the value of a statistical life. That is, it is worth spending $9M if that will save one expected life.

Nuclear, solar and wind have deaths/energy somewhere in the ballpark of 1 life per 10^10 kWh. So, at $9M/life this cost is roughly $0.001/kWh. This is very small, which says that even minor differences in the cost of energy from various sources will be more important than the direct number of lives lost.

(This would not be true of fossil fuels, though.)

TLDR: it's more important to reduce the cost of energy from these non-fossil sources, and to choose the sources with lowest cost, than it is to make them safer. For nuclear, inherent safety could be useful if it would enable cost to be reduced, but not because nuclear needs to be safer.

I have a hard time coming to the same conclusion.

All of the problems with nuclear reactors have happened to plants which were designed and constructed in the 1950->1970s. As it turns out, we've learned a ton about safely operating nuclear plants. The problem is upgrading these old plants rarely happens and getting newer plants to replace them is equally daunting.

There are 3 examples of major nuclear plant problems. That doesn't seem like too many.

In contrast, there are hundreds of operating plants. The newer ones are particularly safe because they require positive input to keep the nuclear reaction going. Any sort of earthquake, tsunami, mudslide, etc that causes the plant systems to fail will cause the nuclear reaction to be halted.

Chernobyl, 3 mile island, and fukushima are all impossible in plants built in the last 25 years. (Gen III or newer)

The US army tried and failed. The US Air Force tried and failed. The marines keep trying. Japan has had several fatal accidents in their civilian program just handling fuel. Even the navy limits them to specialist roles, and their success and safety record might all hinge on the legacy of one gifted man (Rickover).

I support research and trials of the SMRs, but you might want to consider the possibility that it really is hard at the full-system level. The human mind does not readily understand invisible, exponential process like radiation.

Every power source has accidents. Deaths per TWh for nuclear are comparable to wind and solar. Every form of fossil is much worse. Hydro beats everything for major disasters; Banqaio Dam killed 26,000 people immediately and many more in the aftermath.

> Deaths per TWh for nuclear are comparable to wind and solar.

A statistic that only works because epidemiological studies into the long term effects of radiation exposure are extremely difficult, complex and time consuming.

Something made even more difficult by the fact that we blasted uranium fallout in the atmosphere that's hanging around to this day, so getting a non-affected control group has become pretty much impossible.

Ain't helping that any research attempting to investigate the problem will very quickly be labeled as highly controversial by pro-nuclear lobbies [0]

[0] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2696975/

What makes it even more difficult is natural background radiation. The global average is 2.4 millisieverts/year, with the US averaging 3.1 and Japan averaging 1.5. Medical scans add 0.6 mSv/year. Airline crews get an extra 2 mSv/year.

By comparison, atmospheric nuclear tests added 0.11 mSv at their peak in 1963, declining to 0.005 mSv/year today. Chernobyl added 0.04 mSv in 1986, declining to 0.002 today. The nuclear fuel cycle adds 0.0002 to the global average, and is required to be less than 1 mSv for all members of the public.

The highest natural background radiation is in Ramsar, Iran, with 6.0 mSv/year. Studies are ongoing but the evidence so far shows no negative health effects.

Note that Sieverts are normalized to the health effects on the human body. Any concerns about different types of radioactivity are already accounted for in this measurement.

Chernobyl and Fukushima of course caused larger exposures to nearby inhabitants, and these exposures are accounted for in the statistics I mentioned.

https://en.wikipedia.org/wiki/Background_radiation

https://en.wikipedia.org/wiki/Sievert

I'm sure the Army, Air Force, or Marines would if they practically could. The promise of power density with no logistics tail is magic. They choose not to because of practical, operational reasons.

Look at the US Navy operational record for nuclear power. That might convince you that humans can do it.

> or where tsunamis occur in Japan

This one is particularly interesting considering Fukushima wasn't the first time something like that happened. On the other side of Japan is the Kahiwazaki-Kariwa plant [0], the largest of its kind on the planet.

In 2007 that plant was already hit by an earthquake, shaking the plant beyond design basis, it was shut down for 21 months after that.

And even tho it wasn't affected by the 2011 earthquake that blew Fukushima up, it still was shut down to implement safety improvements, it remains shut down to this day with no date for resuming operations.

[0] https://en.wikipedia.org/wiki/Kashiwazaki-Kariwa_Nuclear_Pow...

In the documentary 'inside bill's brain', you hear that regulatory problems led his team of physicists to look to China for a manufacturing partner and a location to build a new reactor (which the trade war between US and China also squashed). If they could have built it on US soil, they would have.

A lot of it is chicken-and-egg problem. New nuclear plants are so rare that they each end up with being slightly different from one another, which adds to regulatory requirements.

If we started mass producing plants, there would be stronger push for uniformity in design and that would translate into significant cost savings. But no one is going to start mass producing plants because it's so difficult to get just one plant online.

Their job is to enforce safety, not good business decisions. :)

Reply on news.ycombinator.com