Everything I've learned about solar storm risk and EMP attacks

A few months ago, I came across one of the most extraordinary papers I have ever read.

In testimony before a Congressional Committee, it has been asserted that a prolonged collapse of this nation’s electrical grid—through starvation, disease, and societal collapse—could result in the death of up to 90% of the American population.

According to the paper, the grid could be knocked out either by solar storms or by Electromagnetic Pulse (EMP) attacks. Solar storms are common. Serious storms like the 1859 Carrington Event are expected to happen about once every 150 years.

The paper continues:

HV transformers are the weak link in the system, and the Federal Energy Regulatory Commission (FERC) has identified 30 of these as being critical. The simultaneous loss of just 9, in various combinations, could cripple the network and lead to a cascading failure, resulting in a “coast-to coast blackout”.

If the HV transformers are irreparably damaged, they might not get replaced for 1-2 years, which would be devastating.

The great majority of these units are custom built. The lead time between order and delivery for a domestically manufactured HV transformer is between 12 and 24  months, and this is under benign, low demand conditions.

To manage this risk, we could stockpile these transformers. An order of 30 of them would cost only $300M. But due to regulatory failure and the financial incentives of the utility industry, the paper claims, we are not stockpiling these transformers.

When I read this paper, I was stunned. Is the risk of prolonged grid collapse really that high? And is it true that, just as the CDC failed to stockpile masks for a pandemic that we were all warned about, we are equally unprepared for a grid failure that could lead to societal collapse and mass starvation?

To answer these questions, I did some homework. I read congressional testimony, think tank technical reports, a book, academic papers, insurance company assessments, several industry technical reports, and multiple reports in the trade media. What I found was at times contradictory. Somewhat troublingly, both sides of the issue accused each other of bias from financial incentives. Overall, my view is that while some of the EMP and solar storm risk is overhyped, it remains a serious issue, and one of the main tail risks we should be preparing for.

I summarize my conclusions as a dialogue.

Dialogue

Q: If the power goes out for 12-24 months, will it really be so bad?
A: Yes, if the power goes out for that long — and that is a big “if” — the results would be catastrophic. Almost everything our modern society depends on, from water pumps to gasoline pumps to ATM machines, would all stop working. This means no water, no car, and no cash.

Q: Jesus. What could cause these outages?
A: The biggest risks are Solar Storms and EMP attacks.

Solar Storms

Q: Let’s start with solar storms. First question: What is a solar storm?
A: A solar storm is a temporary disturbance in the earth’s magnetic field, typically caused by coronal mass ejections (CMEs) from the sun.

Q: How can these storms disrupt the grid?
A: The electromagnetic pulse can induce a quasi-direct current in power transmission lines, which can cause transformers to overheat and be permanently damaged. If enough of these transformers go down, we could be in serious long term trouble.

Q: Are dangerous solar storms common?
A: Yes. The March 1989 geomagnetic storm caused a nine-hour complete outage of the Hyro-Quebec grid. The 1921 Railroad Storm was 10 times stronger than the Quebec storm. And the 1859 Carrington Event is estimated to be 10 times stronger than the 1921 storm. That is, the Carrington Event was 100 times stronger than the storm that took down the Quebec grid. Storms as large as the Carrington Event have a 10% chance of happening every decade, although some say this estimate is too high. Larger storms may be possible.

Q: This is terrifying! What hope do we have?
A: The good news is that we will probably have advance warning. A team at the Space Weather Prediction Center (SWPC) monitors activity on the sun and activity in the earth’s magnetosphere. They can alert utilities of a CME with several hours to days notice, depending on its severity. This puts the utilities on high alert to turn off their transformers if needed.

Figure 1: Monitoring solar storms at the Space Weather Prediction Center

Q: That makes me feel better. But how reliable is this system?
A: Not as reliable as you might like. One-third of major storms arrive unexpectedly, according to the SWPC’s own 2010 analysis. And that’s not just the small storms. According to a news article in Science, the SWPC might also be poor at identifying the characteristics of severe storms, since they are so rare.

Q: Wait, what? Did you say we miss one-third of the storms? And that we might even miss the big ones? We’re screwed!
A: Well, the good news is that even if the storm goes undetected, the transformers won’t fry up immediately. It will take several minutes for them to heat up. The utility companies can monitor them in realtime and turn them off if they get too hot (see NERC guidelines and Section 3.8.2 of this manual).

Q: I dunno, I watched that documentary on Chernobyl, and it seems like in emergency situations people aren’t great at following instructions in the manual.
A: I have my doubts as well. But from what I can tell, the utility companies will probably be able to execute if there is another Carrington Event, resulting in some unpleasant but short-term disruptions, but not a catastrophic long-term collapse.

Q: I’m still not convinced.
A: Well, even if the utility companies completely fail to react, many engineers believe that the voltage collapse caused by a severe solar storm will ultimately save the transformers, since the circuit breakers will open automatically. But this is disputed.

Overall, the combination of human intervention and automatic grid self-protection makes me feel that catastrophic long-term collapse from a solar storm is unlikely but possible. We should take preparedness extremely seriously.

Should we move on to EMP attacks?

EMP Attacks

Q: Sure, what is an EMP attack?
A: EMP stands for “electromagnetic pulse”. Just as EMPs can be emitted during a solar storm, they can also be man-made, in what is known as an “EMP attack”. There are a variety of ways to generate an EMP attack, but the one that worries people the most is a nuclear weapon detonated at high altitude. The pulse generated from such an attack could reach almost everywhere in the continental United States. Major nuclear powers like Russia and China have the ability to launch such an attack. More recently, potentially irrational adversaries like North Korea have pursued this capability.

Q: Should I be more worried about EMP Attacks or solar storms?
A: Opinions differ. But the EMP Commission is more worried about EMP attacks.

Q: Why?
A: There are two reasons. The first is that EMP attacks are unpredictable. Unlike with solar storms, where we will have several hours to days of advance notice, EMP attacks will happen immediately, giving us no time to prepare.

Q: That makes sense. What’s the second reason?
A: The second reason is that EMP attacks contain a high frequency component called an E1 field, which can potentially cause permanent damage to smaller electronics. Both EMP attacks and solar storms have an E3 field, which can heat up transformers. But only an EMP attack contains an E1 field.

Q: That sound scary. We should prepare for an EMP attack.
A: I agree. And the nice thing is that much of what we do to prepare for an EMP attack, like making our transformers more resilient, will also help us with a solar storm.

Q: Given that this could potentially kill 90% of Americans and send the rest of us back to the Stone Age, why isn’t this our most important national priority?
A: Well, not everyone agrees that it’s that big a threat. Whereas the hawkish EMP Commission has been warning of prolonged national grid collapse, other people — often affiliated with the utility industry — have claimed that an EMP attack would only cause brief disruptions limited to a few states. The EMP Commission accuses the utility industry of minimizing the risk to avoid liability. But EMP skeptics speculate that the EMP hawks are themselves motivated to maximize the profits of their own EMP books and protection companies. I tend to believe both sides are motivated by genuine conviction, but that’s still a relevant backdrop to the debate.

The 2019 EPRI report and its critics

Q: It’s terrifying that the predictions are so contradictory! How do we know who is right?
A: We don’t know for sure. But in an attempt to settle the issue, the Electric Power Research Institute (EPRI) conducted a careful study over three years from 2016-2019. They concluded that while an EMP attack could do some damage to a few states, there was no risk of prolonged country-wide collapse.

Q: Wait a second… isn’t EPRI funded by the utility industry?
A: Indeed it is, a fact that EMP hawks are quick to point out. Veterans of the EMP Commission call the EPRI report “junk science”.

But I am not so sure. While the EPRI report does leave some unanswered questions, it was still the most sophisticated and careful report I have read on the topic. And that’s not just my opinion. Sharon Burke, a former assistant secretary of defense for operational energy in the Obama administration, speaks highly of the report. “When you are doing documented research on physical systems, it is still solid evidence, no matter who paid for it. This is not someone’s opinion.” As for accusations of bias, it’s worth pointing out that the report was done in close consultation with leading experts from the DOE and national labs. Much of the data came directly from the government.

Q: So it sounds like the industry put together a high quality report. I’m still a little skeptical of it though. Could you give me some more detail about what they found?
A: Sure, I’ll start with their simplest and most optimistic finding, but then I’ll get into some of the more mixed (and interesting!) results.

Q: OK, what was their simplest and most optimistic finding?
A: Remember how, unlike activity from solar storms, EMP attacks create something called an E1 field? One of risks of an E1 field is that key electrical devices known as “digital protective relays” (DPRs) can be damaged, which could cause major outages. To test this hypothesis, the EPRI report authors subjected 17 different types of DPRs to E1 fields of different intensities, ranging from 0 kV/m to 50 kV/m. They found that none of the DPRs were permanently damaged under any of these intensities. Some of the devices became temporarily disabled, but they could be easily restarted with a power cycle.

Figure 2: Setup for EPRI’s free field tests on DPRs. [EPRI]

Q: OK, so these DPR devices could withstand EMPs up to 50 kV/m, but how does this compare to the intensities that could come from an actual EMP attack?
A: Even with a giant 1000 kT nuclear weapon, the peak field once it reaches the ground is only 25 kV/m. The EPRI report tested a field twice as high.

Q: What were some of the other findings in the EPRI report?
A: So here’s where it gets a little bit more pessimistic, but still nothing close to prolonged country-level grid collapse. EMP pulses are risky not only because they can do direct damage to DPRs. They can also induce voltage surges, which can in turn cause damage to electric equipment.

To test the amount of damage caused by surges, the authors injected direct voltage into the inputs of the devices…

Q: Wait, why did they inject the voltage rather than using an EMP to naturally induce the voltage?
A: That would require radiating the EMP over a large area, which they said was “not practical”. I’m not an expert, but this seems to be a standard practice for EMP testing.

Anyway, in their voltage injection tests they found that under sufficient voltage, the inputs to the DPRs could become permanently damaged. Some devices required voltages as high as 80 kV to be damaged. Others required voltages as low as 5 kV.

Q: Did you say kV? Weren’t you earlier talking about kV/m?
A: Yes, for these direct voltage injection experiments, the relevant unit is the voltage measured in kV. For the earlier experiment, which measured the impact of the free field pulse on the DPR, the relevant unit is kV/m.

Q: If these device inputs can be permanently damaged by a 5 kV surge, does this mean they will be damaged by a nuclear EMP attack?
A: It depends! This brings us to the coolest part of the paper. The amount of voltage induced in a wire depends on a lot of things. First, it depends on the strength of the electromagnetic field. This in turn depends on the distance from the detonation site. So even if a giant nuclear weapon is able to create a peak field of 50 kV/m, the field will be weaker at more distant locations.

Figure 3: The magnitude of the E1 field decreases with distance from the detonation site. [EPRI]

But even more interestingly, the induced voltage also depends on the polarization of the field, the incidence angle (psi), and the azimuthal angle (phi). Depending on where the input power line is located and how it is oriented, the actual induced voltage might be much lower than one would expect under maximal conditions.

Figure 4: The induced voltage in a wire depends on its orientation relative to the incoming electromagnetic field. [EPRI]

Based on the known locations of substations throughout the US (and assuming random orientations), the authors were able to estimate the distribution of induced voltages in these substations, assuming a very strong nuclear attack. In most cases the induced voltages would be less than 10 kV, but some would be even higher.

Q: 10 kV? Isn’t that above the 5 kV damage threshold for some device inputs?
A: Yes, for some of the devices.

Q: So given that different devices will experience different voltage surges and given that different devices will have different voltages thresholds, is there a way we can calculate the percentage of device inputs that would be damaged?
A: Yes! With simulations. While the authors don’t know the type and orientation of each device out in the world, they can make reasonable assumptions that these are uniformly distributed over known substation locations. For each device the authors sampled an induced voltage from the distribution of induced voltage (this depends on the device’s location and on a randomly sampled orientation). They also sampled a damage threshold from a distribution they constructed from their empirical voltage injection tests. For each device in the simulation, if the sampled induced voltage was greater than the damage threshold, they marked the device as damaged.

Figure 5: To estimate how many DPRs would be damaged by voltage surges, the authors drew independent samples from their simulated distribution of voltage surges (Stress PDF) and their empirically measured distribution of damage thresholds (Strength PDF). [EPRI]

Q: That seems vaguely reasonable. What were the results?
A: It depended on how much protection they assumed and the intensity of the attack, but the fraction of damaged devices ranged from 1% to 20%.

Q: Is that good or bad?
A: I can’t get a clear answer on that. The tone of the EPRI report implied that this wasn’t a big deal, but a rebuttal from the Electromagnetic Defense Task Force (EDTF) said this was quite serious:

Relay malfunction during a HEMP attack would likely cause other electric grid systems to fail, resulting in large-scale cascading blackouts and widespread equipment damage. Notably, E1 effects on protective relays are likely to interrupt substation self-protection processes needed to interrupt E3 current flow through transformers.

Q: Interesting. That reminds me, you’ve talked a lot about EPRI’s analysis of the E1 field, both directly via free field and indirectly via voltage surges. But what about the E3 field? If I recall, that was the thing that could heat up and damage transformers, right?
A: Yes, the authors looked into this as well. While they didn’t physically test any transformers, they were able to simulate the effects of E3 fields on about a thousand transformers, using information they had about the age and quality of each transformer. Of the thousand transformers they simulated, only 3 to 21 would experience damage. These damaged transformers would be geographically dispersed.

Q: 3 to 21 transformers would experience damage? Couldn’t that be quite bad? Didn’t you say earlier that 9 critical transformers could take down the whole grid?
A: I can’t get a clear answer on this either, but I think it’s ok. The EPRI report made this sound like not a big deal, and the most comprehensive critique of the EPRI report did not bring it up in its list of concerns.

Q: What are some criticisms of the EPRI Report?
A: The Electromagnetic Defense Task Force (EDTF) includes dozens of critiques, but a few that jump out are:

Q: Now I’m confused and don’t know who to believe!
A: I agree, it’s not great. Apparently, EPRI did not coordinate with the Congressional EMP Commission to compare results and methodology. Many of the test results are inconsistent with the EMP Commission’s own test results from 2008, and I don’t believe anybody has resolved the discrepancies. So there is a troubling lack of communication here.

Q: I’m feeling overwhelmed with all this information about EMPs! Can you summarize what we’ve learned?
A: Sure.

  1. The EMP Commission believes EMP attacks are more dangerous than solar storms, because they are unpredictable and because they include an E1 field, not just an E3 field.
  2. The industry-sponsored EPRI report, which seems to be carefully done, claims that an EMP attack will at most cause temporary disruptions in a few states. The E1 field itself will not cause any permanent damage to DPR devices, although a small fraction of devices may be damaged by E1-induced voltage surges. The vast majority of transformers will survive the E3-induced temperature increases. The few transformers that may fail will be geographically dispersed.
  3. However, a rebuttal report from the EDTF raised many concerns with the EPRI report, including concerns about cascading failures in complex systems. The discrepancies between EPRI and EDTF, as well as the discrepancies between EPRI field tests and previous tests by the EMP Commission, have not been fully resolved.

The chances of an EMP attack

Q: I’ve heard a lot from you about how dangerous an EMP attack could be. But how likely is it that anyone will actually try to attack us with an EMP?
A: Unlike the EMP Commission, most national security experts view EMP attacks as a second rate threat. While perhaps some small terrorist groups or rogue nations might launch a localized EMP attack that might take out a substation or two, it’s unlikely that any country capable of launching a large-scale EMP attack would actually do so.

Q: Why not?
A: Because to launch a large-scale EMP attack, a country would need a large nuclear weapon. And if a country was planning on using a large nuclear weapon, it would make more sense — in the morbid logic of war — to conventionally drop it on a city than to launch an EMP attack which would at most cause some brief power disruptions in a few states. As physicist Yousaf Butt put it, “A weapon of mass destruction is preferable to a weapon of mass disruption”.

Q: So while EMP attacks could be serious, they are less dangerous than advertised, and it’s unlikely that anyone would ever want to launch a major attack.
A: What’s especially interesting is that because EMP attacks are unlikely, and because they will probably only cause transient disruption even if they do happen, some experts believe that the bigger threat is solar storms! The EMP Commission, from this perspective, had their priorities backward.

What are we doing about all of this?

Q: What is the United States doing to protect itself against solar storms and EMP attacks?
A: According to the EMP Commission, not nearly enough. The EMP Commission members are very critical of the regulatory environment, which they view as dysfunctional. FERC, the government agency that regulates utilities, does not have the power to make the utilities protect the grid. All that FERC can do is ask the industry umbrella group (NERC) to propose an EMP protection standard, but the NERC standard is determined by industry representatives and is therefore too weak.

Furthermore, there is confusion about which government agency is responsible for EMP protection. From George Baker’s testimony before Congress:

When I ask NERC officials about EMP protection, they informed me we don’t do EMP, that’s DOD’s responsibility. The Department of Defense tells me, EMP protection for civilian infrastructure is DHS’s responsibility. And then when I talk to DHS, I get answers that the protection should be done by the Department of Energy, since they are the infrastructure’s sector-specific agency. So we have EMP and GMD protection as finger-pointing exercises at present.

Q: That sounds bad.
A: Agreed. The good news is that in March 2019 the EMP hawks got their wish, when President Trump signed an executive order putting the White House in charge of national EMP preparedness, rather than relying on scattered federal agencies. EMP Commission chairman Peter Pry hailed the executive order, describing it as an “excellent first step”.

Even EMP skeptics expressed support for the executive order. Former staff member of the Senate Armed Services Committee Gregory T. Kiley wrote:

EMPs are by no means one of the top-tier national security challenges, nor the most pressing concern for the safety of our electrical grid. A careful and reasoned plan put forward, like what we saw last week from the White House, makes sense.

Q: What’s happened so far with the executive order?
A: According to Politico, the process has been disrupted due to the departure of several key advocates from the administration. Peter Pry warns of “inevitable opposition from recalcitrant lobbyists and bureaucrats”. That said, the National Defense Authorization Act passed by Congress this year includes some laws regarding EMP protection.

Q: How much will it cost to better prepare us for EMPs and Solar Storms?
A: It depends what the plan is. The bare minimum is to spend \$200M to protect the extra-high-voltage transformers. The EMP Commission proposed a \$2B plan to protect all the transformers and generators. George Baker has a much more expensive plan at \$30B. That might sound like a lot, but if you amortize it over several years, it would come out to a \$2-3 charge in your monthly utility bill.

Q: Getting back to how you started this blog post, you mentioned that some key high voltage transformers take 1-2 years to produce, and that we should therefore stockpile about 30 of them, at a cost of about $300 million.
A: I was surprised to hear that Peter Pry, of the hawkish EMP Commission, does not support this plan. According to him, it would be much cheaper to just put surge arrestors on our active transformers. Furthermore, many transformers are custom-built for individual substations, so it will not be easy to produce stockpiled interoperable transformers that can work at any substation.

Conclusions

Q. What’s your overall take on this?
A: The ‘mainstream’ belief is that solar storms and EMP attacks are concerning but overhyped. While almost everyone agrees that we should make our grid more resilient, many say the doomsday scenarios are extremely unlikely. Most national security experts think that countries capable of launching a major EMP attack will not want to launch one, although smaller attacks from rogue actors may be possible. Consistent with the reasonably careful EPRI report, the mainstream view is that the impact of even a major EMP attack will likely be short-term and regional, rather than long-term and national. As for solar storms, the utility companies will have plenty of advance warning to respond.

Other people, especially those who have worked on the EMP Commission, are much more worried about these risks. They point to the fact that hundreds of grid components have not been thoroughly tested, and that problems in one part of the grid can cascade to other parts in unpredictable ways. The media has at times portrayed these people negatively. In an article that I thought was irresponsible, Slate called the EMP concern “right-wing fretting” and a conservative “fixation”. Wired says that the EMP arena is filled with a lot of “hype” and “fearmongering”.

My own view is that while the ‘mainstream’ view is probably correct, and while there certainly has been some fearmongering, I am philosophically aligned with the alarmists. The mainstream belief at NASA in 1986 was that the Challenger was safe. The mainstream belief at Chernobyl in 1986 was that the reactor core could never rupture. The mainstream belief on Wall Street in 2007 was that mortgage-backed securities were safe. In all of these cases, the mainstream belief was wrong and the alarmists were right.

Now that we have seen our preparedness level for Covid-19, who are you going to believe: The people saying “Don’t worry, we have this unpredictable and complex system under control” or the people waving their hands and shouting “correlated risk!” I’m with the people shouting “correlated risk!”, even though they’ll probably end up being wrong.