Ok, So Maybe It is Gridcrash that Makes the Lights Go Out

Quite a few years ago I wrote a piece arguing that the single most likely scenario for most of us having to deal with long term electrical shortages doesn't involve gridcrash scenarios, but the growth of poverty and utility shut offs. I suggested that people should be prepared to deal with electrical outages in large part simply because of the economic consequences of our situation. It isn't that I didn't believe anything could shut down the electric grid, I simply felt that realistically, the probabilities of more than short-term outages in the near term were pretty small.

In news that falls more in the astrophysicist's department than my own (hat tip to him for pointing this out to me), however, it turns out that at least one expert in Space Weather places the probability of a Carrington-style solar storm that could knock out transformers on a large scale at much, much higher than I'd ever considered - a one in eight chance by 2020. Ok, I was wrong - maybe you should be worried about gridcrash.

Given that NASA and other previous analysts have suggested that such an event could knock out electricity in regions for four to ten YEARS, that's a number to think about. I keep fire and health and flood insurance against probabilities that are not that likely, and I bet many of you do too.

What's new about the current data, btw, isn't that this is a possibility - people have known about this for years and there have been a number of major reports. What's different is the new estimation of probabilities, which, if correct, is extraordinarily high.

The possible impact of a major solar storm of this type range from knocked out satellite service and really cool auroras to losing power for extended periods, and as the 2008 NRC report on the subject said:

"A longer-term outage would likely include, for example, disruption of the transportation, communication, banking, and finance systems, and government services; the breakdown of the distribution of potable water owing to pump failure; and the loss of perishable foods and medications because of lack of refrigeration," the NRC report said.

Now that's the worst case scenario and it is worth noting that the Space Weather research article does NOT claim a 1 in 8 chance of this happening. Still, I think it goes on the list of things that people should think about. The good thing about general preparedness is that it makes you ready for anything - but too many of us individually, as communities and as a nation remain unprepared for a lot of stuff.

That's being facilitated by at least one Electrical Industry Regulatory group, which is back on the typical response of anyone who might have to spend some money to prepare for a plausible danger - deny, deny, deny:

The North American Electric Reliability Corporation, or NERC, which is supposed to help the federal government regulate the electric power industry, has issued a report that minimizes the impact of even an intense solar storm which emits an electromagnetic pulse, or EMP.

My favorite strategy of denial (over the objections of many, including NASA, NAS, the US Department of Energy and more...) is this one:

Over the objections from GMDTF participants, the NERC report deferred to the opinion of industry transformer designers to assess the vulnerability of transformers to geomagnetic storms. The assessment did not include the opinion of electromagnetic pulse experts.

Why, after years of following peak energy does this sound so very familiar?

To my mind, all of this is just one more reason why we need to be very careful about how dependent we are on vulnerable fossil fueled infrastructure, and why the precautionary principle needs to be part of our thinking. But then, we knew this already.


More like this

Hmm ... so if this is an EMP, that would mean it'd take out home solar systems as well, right?

That would be nasty.

Is there a geographical probability distribution to these things? I am thinking that us northerners, who actually see the aurora borealis regularly, might be at more risk, but I don't know ... and I'm not married to an astrophysicist so could you maybe ask yours? :)

What?? You didn't know about this one already!? And here I thought you were a true connoisseur of Apocollapses.

I first heard about this one in 2002, at a global warming conference in Boston. Electrical transmission expert, knew all about it. If I recall correctly; his statement was that a major pulse could destroy a couple thousand grid transformers; and there were about 30 replacement transformers on hand in North America. And manufacturing capability for about 100 a year, in a pinch. (My recollection at this point is approximate; but that's the ballpark.)

Everybody in our little conversation group was appalled- "My god; somebody needs to tell Congress!" To which he responded "Oh, we've tried. They listen; say 'oh, my god, that's horrible!' - act really worried for 3 minutes; then go back to whatever they were doing. It's just too much for them to cope with- so they don't."

It's also a favorite fantasy of Newt Gingrich's now- that a "rogue state" will use a nuclear explosion in space to cause a highly destructive electromagnetic pulse. I'm afraid that's actually a much more real threat from Iran or North Korea than that they'd try to hit a city- it's way easier to just shoot a rocket up high, and set it off-

And at this point, the Kingston Trio's song "They're rioting in Africa!" starts to run through my brain....

People in the electric grid business should know better. They have the example of the March 1989 geomagnetic storm, which took down the electric grid for all of Quebec for several hours. Contributing factor: Hydro Quebec runs one or more hydroelectric stations along Great Whale River, which is 1500 km or so north of Montreal. That makes a nice long antenna for picking up geomagnetically induced currents in the ground, located at auroral latitudes where such currents are likely to be strongest.

Since then, our electrical systems have become more interconnected. We still, last I heard, have three separate regional grids (east, west, and Texas), but somewhere in there is at least one de facto antenna with an effective length of more than 2000 km. (It's about 4300 km by air from Boston to San Francisco, and about 2000 km from Boston to Miami.) We're probably far enough from the geomagnetic pole to not worry about storms like March 1989, but a Carrington event would be another story.

By Eric Lund (not verified) on 06 Mar 2012 #permalink

So my question with a large power outage over an extended period of time is what about the nuclear power plants? Isn't what did Fukashima in not the quake and tsunami, but the loss of power which prevent the cooling of the reactor cores? So......if one is living near a nuclear plant and they can't keep the core cool or shut it down (which I heard you can't ever really shut those down but I'm not a scientist), then that population is really in trouble. Our own preparedness really isn't an issue then because the radiation leaks would be horrendous.

By Denys Allen (not verified) on 06 Mar 2012 #permalink

Some doomsday scenarios, I think we will be totally helpless no matter how prepared one is. There is one thing few of us who presumably read this blog can or want to be doing, that is, create a powerful defense system for our farms.

I suspect that in a doomsday scenario, no matter how well prepared some families are to be self-sufficient on their acreage -- there will be looting. Are we ready to decent, compound-style .. not me. I'm afraid my animals or my crops would simply be looted, we'd wind up like in a Western movie, where we have to defend the homesteads.

Oh no, this isn't totally new to me at all, Greenpa - if I weren't generally the sort of person who pays attention to this stuff, Eric, whose work specifically focuses a bunch on satellites and other space objects would have that one covered - it is the dramatic change in probabilities that is new. The highest credible probabilities before this were in the 1 in 100 or more range - this is very different.


A scenario that is all the more gruesome to contemplate as nuclear power plants would need those four to eight years for cooling of reactor cores and spent fuel pools -- presumably by running diesel generators the whole time. It would not really take very much social disruption to bring such efforts to a complete (and catastrophic) halt.

Well, I finally get to play electrical engineer, 28 years after graduating.

The answer to the question depends largely on how strong the solar flare is, or how strong the nuclear-induced EMP is. Larger, interconnected systems would feel it first, due to the antenna effect. It would, however, take a much larger pulse, probably from a more local EMP burst, to take out stand-alone, house systems, though modern inverters, if they are at all grid-connected, could be destroyed. Simpler house solar systems, with mainly direct current to load designs, would probably fare best.

Given that we saw Japan suffer greatly when power could not easily be restored to the nuclear plants, I too would think we'd see massive nuclear plant damage and release in an extended regional and/or national grid collapse, depending especially on whether the electromagnetic pulse also destroyed the control electronics for the plants' diesel back up generators.

Something that bad would simply be impossible for the rest of us to prepare for, as I would think the social destruction would be truly something that only the most serious Preppers, holed up in their remote, Idaho cabin, would contemplate.

By Stephen B. (not verified) on 06 Mar 2012 #permalink

It wouldn't be a good thing, Denys. Applejack Creek, I've asked - he'll look it up and post his answer. His first instinct was that it probably wasn't more likely to affect northern latitudes, but he's not entirely sure. It would likely be geographically distributed depending on how the planet is turned at the time, but the question of whether you can calculate probabilities is another one.


"The highest credible probabilities before this were in the 1 in 100 or more range - this is very different" - ah. We were hearing different numbers there- the guy I was talking with in 2002 was putting it way higher than 1/100; but we were also talking long range; it was more like "30% chance in next 10 years; 70% in next 20.." and of course, if you run it out to 40 years; it's 100%. The guys who want to ignore it will always quote you the "it's only 1 in 100" ... deleting the information that that's the calculated probability for the next year; only.

They all get to 100%; in the long run. :-)

Hmm ... so if this is an EMP, that would mean it'd take out home solar systems as well, right?

No. Home solar systems have wiring loops measured in tens of square meters. Electrical grid wiring loops are measured in thousands of square kilometers.

A really spectacular EMP can, of course, take out both. But you're several sigma out to even get to where the ginormous grid loops take damage; it'll take something much, much less likely to take out home solar setups.

By D. C. Sessions (not verified) on 06 Mar 2012 #permalink

We were hearing different numbers there- the guy I was talking with in 2002 was putting it way higher than 1/100; but we were also talking long range; it was more like "30% chance in next 10 years; 70% in next 20.." and of course, if you run it out to 40 years; it's 100%.

And yet ...

We've been monitoring space weather (to at least the crude level required to notice huge EMP-type events) for much more than 40 years. It's not like human activities have materially changed solar weather (as distinct from terrestrial climate.)

I'm not trying to sugar-coat the risks of grid crashes. To the contrary, I pointed out a grid-crash scenario twenty years ago on the RISKS Forum. As far as I can tell, it's still valid. But assigning a near-unity probability of an event over 40 years that hasn't happened in much longer than that? I suspect some slight error in the calculations.

By D. C. Sessions (not verified) on 06 Mar 2012 #permalink

Does it have to do with the uptick in solar activity recently? I follow space weather.com intermittently and the last few days have been fun. It even prompted me to recheck our food stores (not that we can store much in this tiny ass apartment, but something is better than nothing).

Ok, but help me out: even a huge flare takes a few days to arrive at Earth, right? And major utility providers and whatnot DO watch the solar flare info so they know when a CME is incoming, right? Wouldn't they have time to shut down major pieces of electrical infrastructure before the CME arrived? I mean, we might lose power for a few days, in that case, but not months.


The guys who want to ignore it will always quote you the "it's only 1 in 100" ... deleting the information that that's the calculated probability for the next year; only.

If that's the case, then the probability has increased somewhat, but not that much: less than a factor of 2.

Assuming the probability P that the event happens this year doesn't change from one year to the next (not quite true, but good enough for illustration), then the probability of at least one such event in the next N years is 1 - (1 - P)^N. If P is 0.015 (which is 1 in 67), that translates to about a 12% probability over nine years (the paper was written last year, and has only just been published). The probability never quite reaches 100%, but it does get noticeably high rather quickly--and of course there might be more than one event.

By Eric Lund (not verified) on 06 Mar 2012 #permalink

Responding to jerah @12: Yes, the power companies do monitor these things. Yes, they have contingency plans, but whether these plans would be enough to deal with a Carrington event is open to question (you can't just shut the entire grid down, and that may be what would be required; even if you do, power lines still make nice antennas). Yes, they usually have about three days of lead time for a CME, but not always: the CME from the Carrington event arrived in less than 24 hours (Carrington observed the flare during the day, and London, normally well south of the auroral zone, had spectacular auroral displays that evening).

So a stonewalling response by the power companies, rather than a statement to the effect of "we are aware of the risk and have ways of mitigating its effects," is not a good sign.

By Eric Lund (not verified) on 06 Mar 2012 #permalink


I would like to caution you against taking the argument from this paper with too much credulity. The model used has some tremendous uncertainties associated with it which make the "12%-in-the-next-decade" figure quoted an extraordinarily low confidence prediction.

For some details, you may want to check out this recent publication: http://arxiv.org/abs/1202.5995

In particular:

These show that Poisson probabilities perform comparably to some more complex prediction systems, but the overall inaccuracy highlights the problem with using average values to represent flaring rate distributions.

I'm surprised no one has referred to this article:


@D.C., you're correct, solar storms of this magnitude seem to happen about once every 70 to 100 years. The last one was in 1921, and the Carrington event was in 1859.

@Risa Bear, it takes about 1 year of running the diesel generators to cool down the core enough to prevent a melt down. It still would likely be dangerously hot for a while, but at least a catastrophic failure would be avoided. Most nuclear plants have at best a few weeks worth of diesel fuel to keep the cooling system going.

Hey! This is my husband's area of research--he does predictive simulations of space weather events. I'll try to get him to comment later tonight. We've been talking about this issue for years, and it's one of the main things he talks about when people say "why should we spend money on that kind of research anyway?". Plus the bit where communications satellites get knocked out of commission by these same events.

So what does a X5.4 flare, like the one that just happened, mean? Some satellite disruption? Possible ground current disruptions at higher latitudes?

@ D.C. Sessions
Bear in mind that 40 yrs ago, the electrical grid was much smaller & not yet interconnected to the degree it is today, so the antenna effect was significantly reduced. Also, the grid is aging, and many of the transformers & cables have withstood many surges already. Another big one could overwhelm them.

Well, if solar storms of this sort of magnitude happen about once every 100 years, that's 1 in every 9 solar cycles...

Without wanting to launch a long discussion on Statistics- to DC and Eric - sure, your calculations are fine, and validated by many- but I don't agree with them. I adhere to an obscure branch of statistics, seldom mentioned but nonetheless considered academically respectable, which holds that real world probabilities change, depending on past history. So if you toss a fair coin heads 8 times in a row- yeah, the probability for fair toss #9 is not 50-50. The discussions on "why" get really hairy.

We also were starting with quite different base assumptions. And; you need to keep in mind we are talking NOT about "random events" - but about "natural PROCESSES". Just as with The Big One for California Earthquakes, the question has never been IF; but only WHEN; and yes, the immediate chances increase every day, because what we're talking about is the increasing tension between rock formations as they continue to move against each other. They move- every day- the tensions do increase, every day. Eventually- yep.

Solar processes are generally believed to be similar- though even less understood than earthquakes. There are "processes" - happening regularly- which result in mass ejections, etc, regularly, and cyclically.

I would point out here, also, that the recently created academic "disciplines" of "risk assessment", and more humorously, "risk management" - were created by gifts to universities to set up such departments, from large business concerns; and their actual intention (not publicly stated) is not to understand "risk:, but to justify to law makers undertaking projects known to have potentially disastrous outcomes. For purposes of making quick profits, and getting out before the disasters actually occur. No, I don't have references for that. My own observations.

Nuclear power being an excellent case in point; and I will also point out here that over two years ago I am on record (NYT Green Blogs) as stating that "Nuclear power is not going to be revived. Apart from the truly huge startup costs, and time scales, for new plants to be built, when the next Chernobyl happens, public reaction will prohibit it. And yes, there will be another Chernobyl; it is only a matter of time. My own guess would be in France, or Japan- they have the most reactors, and the most arrogance regarding their operation." That quote is approximate, from memory.

That was more than a year before the multiple melt-throughs at Fukushima. And- statistically- when you use your conceptual model to make a prediction; and the prediction comes true- that is hard evidence in favor of the validity of the model.

Ethan, you are right - but looking back over the past articles, it doesn't seem like those that came up with much lower probabilities had vastly lower uncertainties - ultimately, predicting future events is a pretty uncertain exercise, the only real question is whether the margin of significance is enough to merit the precautionary principle, no? I'd say yes there.


I'm still trying to take in this nuclear situation, what having read the link in John @17's post.

I kind of knew that subject matter before, namely that nuclear plants need LOTS of active management, that is, cooling, even after they are "shut down", but the Japan situation really drove home to me the seriousness of this situation.

Face it, when we look at how our social, economic, and political situation has destabilized over the past few years around the world, and when we realize how we have hundreds of nuclear plants, and especially spent fuel rod pools, around said world, and we realize that it would be Fukushima crossed with Chernobyl many, many times over, I now realize that it really is a question of not if, but when, that we'll see major, major radiation releases around the world. It might happen due to a major, sustained grid crash due to solar flare or an EMP burst, but if not that, it's only a matter of time that it happens due to a quick economic, social, or civilization crash as well.

I wonder how many more times the people of the world will accept the nuclear industry's and the egghead physicists' assurances and pronouncements that all the other world nuclear plants are basically safe, that another Fukushima can't happen? How many times must world civilization see one of these nuclear disasters before people, the world over, "get it" ??

Anybody who says that humanity will safeguard all those plants in the face of a major, *human* melt down, just doesn't understand the history of war, civilizations, and the like, nor do they understand what happened in the US in late 2008, or in any country such as Libya, Syria, etc., in the more recent past.

I think that people love their electricity, and this will especially be true in the developed world as we embrace electric power and transportation even more, in the face of oil decline. Sadly, I think it will take several more major scale radiation explosion/releases before everybody understands this, worldwide, if at all.

I suppose there are nuclear plant designs that cool themselves passively in the case of prolonged cooling system shut down, but those designs are not what we have in the hundreds of existing world power plants. Then too, I suppose there might be ways of more quickly reprocessing and/or breaking up spent fuel rod collections so they don't need several years of active cooling water immersion, but again, the latter is unfortunately where the world has tens of thousands of fuel rod assemblies stored right now.


By Stephen_B. (not verified) on 07 Mar 2012 #permalink

Greenpa: Everything you wrote sounds almost reasonable, except for this:

I adhere to an obscure branch of statistics, seldom mentioned but nonetheless considered academically respectable, which holds that real world probabilities change, depending on past history. So if you toss a fair coin heads 8 times in a row- yeah, the probability for fair toss #9 is not 50-50.

I'd be really curious what the name of this "academically respectable" branch is. And exactly what the probability for toss #9 is. And most importantly, why actual tests with coins and other independent randomizers don't seem to confirm this, but rather refute it.

Lenoxus - "almost", eh? :-)

My statistics professor did not tell us the name; perhaps there is not one; just that there were several respectable academic statisticians who were working in this area. That was a couple decades ago, to be sure. The topic came up in class because on some level the idea that toss #9 should be 50/50 does not agree with "common sense"; nor with the way any sane gambler would bet.

Hm. Maybe, "Quantum Statistics." The principle being that yes indeed; what you know about the past affects the present. :-)

Incidentally, the present rating of effects of the current solar X5 "event" have been bumped up by NOAA to "strong" from "moderate" - updates are available at swpc.noaa.gov , but be aware they're not well indicated; you have to actually read the notice, which may look identical to the one a few hours ago- but isn't. "Finally, a Solar Radiation Storm is also in progress and levels are currently above the S3 (Strong) threshold. Region 1429 remains potent and subsequent activity is certainly possible. Updates here as conditions warrant."

So there's this explanation of the G3 geomagnetic storm rating that NOAA is predicting: http://www.swpc.noaa.gov/NOAAscales/#G3 , but spaceweather.com is also saying:

"Analysts at the Goddard Space Weather Lab, who prepared the CME's forecast track, say the impact could spark a strong-to-severe geomagnetic storm. Sky watchers at all latitudes should be alert for auroras."

I guess the predictions are just predictions. Hoping for an aurora sighting in the teeny tiny strip of sky visible from my Brooklyn back yard tonight!

Now it's updated again (at spaceweather.com), saying the CME will arrive around 11:25 pm tonight (plus or minus 7 hours!) "possibly triggering a strong-to-extreme geomagnetic storm".

They really don't know what happens next, do they?

jerah- my impression on the predictions so far is they're on a par with our local tv weather reporters when a snow storm is moving in- they tend to be pretty enthusiastic about the possibilities. Usually reality is a little less than the enthusiasm- except sometimes- they underestimate, of course; we've wound up with 16" when they were predicting 5".

It's an uncertain world. :-)

Thanks for the reassurances, all.

We have grid feed (instead of a generator, we can draw from the grid) to our inverter - so I suppose if I know a pulse might be coming, it would be prudent to make sure that link is shut off? Or does the very fact that the wires are there provide enough antenna effect to be problematic - even if the switch is flipped to the 'off' position?

So, Sharon blogs about Solar Flare induced EMP, and lo and behold, a big solar flare pops up. Please do NOT write anything about a major subduction earthquake in the Pacific Northwest.

Marrowstone Island
Jefferson County
Washington State

So what does a X5.4 flare, like the one that just happened, mean? Some satellite disruption? Possible ground current disruptions at higher latitudes?

No, it probably just means that there's a higher than average chance of aurorae, and they might be visible at lower latitudes than usual. The biggest flare of the last cycle was (I think) around X25, and nobody other than aurora watchers noticed much of anything.


The topic came up in class because on some level the idea that toss #9 should be 50/50 does not agree with "common sense"; nor with the way any sane gambler would bet.

But sane gamblers do understand that independent events have independent probabilities. The irrational gamblers, who are naturally much more numerous, are the ones relying on common sense. They adhere to the notion that "the odds even out", or that tossing heads somehow "reduces" the available number of heads for future tosses, or that coins have a "memory". It's called the gambler's fallacy, and casino profits are almost entirely because of it.

One reason I asked about the precise probability is that any answer other than 50/50 leads to paradoxes and oddness. For example, suppose I am about to flip a coin exactly twice. Is the sum of the odds of H-T and T-H somehow greater than the sum of the odds of T-T and H-H? If the gambler's fallacy is correct, then this should be the case, in contradiction to both probability theory and massive repeated trials. Also, does it matter whether I flip the same coin twice, or instead flip two different identical coins once each? The experts generally agree that it should make no difference, but the gambler's fallacy cares about the distinction.

If you want to get technical, something like the opposite of the gambler's fallacy is more likely to be true, to a limited extent. That's because for any given coin or roulette wheel, there's a chance that it's biased for one side/color more than the other. So if you see someone flip a hundred tails, you should bet on tails because of the increased evidence that the coin's physical makeup is such that it simply doesn't land on heads. (Maybe it's a double-tailed counterfeit coin.) But nearly all real coins have no such bias, so there's usually little reason to consider the possibility.

That said, plenty of non-independent probabilistic events can indeed be such that earlier events change probabilities for future events. For example, whether or not there will be a major California earthquake in 2013 is inherently linked to geological forces which are in turn connected to all the geological events in California of 2012, and 2011, etc. It's not independent. But when you specifically used the example of coin tossing, that caught my attention.

Greenpa, with respect, why do you adhere to a branch of statistics that you don't know the name of, nor the nature of the research being done by statisticians whose names you also don't know? Personally, before I claimed to adhere to a school of thought at minimum I'd want to know what evidence there was for it, besides intuition... all the more if it goes against the prevailing thinking in the discipline as a whole.

Q: Is it possible that Greenpa is poorly recalling (and therefore inaccurately describing) Bayesian statistics?

Note: Personal attack neither explicit nor implicit here. The list of things that that I would fail to describe accurately due to poor recall is lengthy.

Kerrick- see the first sentence of comment 23. :-) Why? Because statistics is a very secondary field for me; when I need full formal statistics- I get someone else to do it. My focus is- the hypothesis, and the experimental outcomes. So a good working "feel" for it; has been more than plenty, for decades. And 90% of my research is always counter to the discipline as a whole; that's what I do.

NJ- yep, that rings all the bells. Thanks! See? Somebody right near by had the answer!


Greenpa â Sorry if I seemed to rant a bit, earlier there. Anyway, I think I can understand how Bayesian statistics (which is indeed a respectable field, and maybe even larger now than when your professor told you about it) could be confused for gambler's fallacy type thinking. (Among other things, a Bayesian is likelier to consider the possibility that a given coin is biased, and thus that a string of tails technically does tell you something about the next flip's odds.) It's even possible that your professor was confused about it; it wouldn't be the first time teachers have made such mistakes.

Lenoxus - no problem! :-) In spite of my apparent casual attitude, I do spend time paying attention to the USES of statistics, and the effect on our cultural trajectories. One of the things that happens frequently is the MISapplication of one form of statistics for another- without anyone in the audience realizing it; and frequently with the perpetrator unaware, also. That has horrific consequences, and was the basis for my initial grumblings here. If anyone is still interested in my thoughts, I've actually dealt with it a bit on my own blog, two topics:



Both of those are, I think, important considerations about statistics that the scientific community is NOT dealing with; and mostly has not noticed. I keep the comments open on those things- yours would be welcome.

There is also a bit of painful statistics news from the NYT today, having to do with police corruption and hiding and deleting numbers -