Executive Summary

This initial investigation begins to examine strategies to mitigate and respond to risks posed by high-impact geomagnetic events, which can severely damage electrical infrastructure. This is split into four sections. The first set are ordered from least promising to most promising targets for investment; recovery, adaptation, and withstanding events. Finally, risk reduction is noted as a highly speculative but high value-of-information area for research.

Recovery approaches involve replacing broken equipment and infrastructure after an event, and focus on the logistical and financial challenges of replacing key infrastructure, notably the costly and rare Ultra-High Voltage (UHV) transformers. Adaptation strategies, including the use of Ground-Induced Current (GIC) blocking devices, are identified as viable and potentially partially adopted means for system operators to prevent damage and reduce restoration costs. Next, withstanding involves protocols for what power systems can do during geomagnetic events. This emphasizes proactive grid shutdowns and sectional isolation, which work by leveraging the short warning period before a CME reaches Earth, or plausibly in advance of nuclear war. These are very promising, but should be pursued by governments vand industry. Finally, risk reduction options are currently limited, but potential exists for longer term, highly speculative approaches for minimizing geomagnetic vulnerabilities at the global level.

Specific high-impact or very impactful mitigations are not explored, but because private companies have some incentive to address the risk, policy approaches involving insurance and regulation are noted. Further work could address this, and will be outlined in the conclusion.

Background

The electrical grid is critical infrastructure, and if electrical systems were destroyed at a national or global level, it could plausibly be or lead to a global catastrophe, especially given the fragility and interconnectedness of other systems. A brief (8 minute) video overview from Kurzgesagt from 2020 explains the risk of solar storms. Given that the lack of sufficient backup transformers was recently highlighted by J.D. Vance on Joe Rogan’s podcast, I wanted to double check my current understanding of the risk and mitigations available. Rather than focusing on risk estimation, which has been done before, I’ll provide a very brief summary of the risk, then focus on risk mitigations, and highlight what is possible, and what has been done.

In 2015, David Roodman wrote an in-depth, 56-page investigation into solar storms for Open Philanthropy, concluding “the probability of catastrophe is well under 1% per decade, but is nevertheless uncertain enough, given the immense stakes, to warrant more serious attention.“ Note that this was limited to solar storms, not electromagnetic pulses from intentional acts, which would adversarially target the weakest or most vulnerable aspects of a system - but would be at a national, rather than global, level. (And given that they would require nuclear attacks, would be part of broader nuclear war risk, which is not the current focus.) Roodman’s investigation was limited to probability and possible impact, not mitigations, so I view the current (much shallower) investigation as a continuation or extension of that.

This is not to say that this is novel - a number of exercises have been done, including international ones. One such report is here, and it notes that “Some countries have successfully hardened their transmission grids to space-weather impact and sustained relatively little or no damage due to currents induced by past moderate space-weather events,“ but “the vulnerability of the power grid with respect to Carrington-type events is less conclusive“

What can be done?

At a very, very high level, based on previous work I did (in a different domain), there are three different ways to make a system more resilient; withstand, adapt, and recover. There is also risk reduction, which can be critical, and prior to resilience. I have not reviewed legislation on the topic, but my understanding is that there hasn’t been progress. (Note that I have not reviewed the NDAA for past years or infrastructure bills to see if they include relevant provisions.)

Recover

I’ll address recovery first, since it has received the most prior attention - notably reflected in Vance’s suggestion that we need to have backup transformers. CMEs would damage transformers by inducing current in long-distance wires, which then damages the transformers. Recovery from failure would require rebuilding whatever portion of the grid was destroyed. Replacing the entire US electrical grid could cost $5 trillion (USD, 2017) per Joshua D. Rhodes, a UT Austin Research Scientist, but this estimate includes replacing the power plants themselves, which would not be destroyed in our scenario. The transformers, which are at high risk, would cost a “mere” $600b in current dollars, and the largest ones are more likely to be destroyed in an event. This analysis presumably overestimates actual costs if the system were replaced more intelligently, but more critically, it understates the cost and ignores the likely impossibility of doing so quickly if it needs to be done in an emergency scenario.

The components most at risk from even a moderate event are Ultra-High Voltage transformers. These are very, very expensive ($100m for the Three Mile Island transformer!) and relatively few exist. On the other hand, China’s largest transmission line evidently uses 28 of them. (Each is rated for about as much power as the 3MI plant.) I don’t have a breakdown of transformers in the US electrical grid into ultra-high versus high versus relatively smaller units, nor it is clear to me what proportion would be at risk in various sized events. However, larger events would create additional risks, including destroying smaller transformers. Less likely, high-voltage power lines could be badly damaged if there was a very extreme space weather event - I am uncertain if this is a significant risk, and would require further analysis. 

Adapt

A number of approaches exist to adapt to this risk. First, there are existing design considerations which reduce vulnerability. Further work could enhance the ability of the grid to adapt. Roodman did a background research interview which noted “ground-induced current (GIC) blocking devices are the best option for protecting against the threat to the grid posed by geomagnetic storms,“ and “installing GIC blocking devices in transformers around the US would cost one billion dollars.“ This is in contrast to the earlier tends of hundreds of billions for replacing some or all of the transformers. Another approach is GIC-resistant transformer design; it is unclear to what extent this occurs, but requiring future transformers to have such designs, or incentivising it (perhaps via insurers, who cover the risk,) could be a useful policy intervention.

There are also systems for sharing the (limited) stock of replacement transformers, so that moderate levels of transformer failure can be addressed. This exists within the United States, but almost all transformers are built internationally, so that replacing supply during a more severe global event, when other countries will prioritize their own recovery, seems infeasible. I have not looked at whether international cooperation has been explored, or whether other countries have similar plans.

Switching to smaller scale microgrids could reduce the impact of certain risks, so that the ongoing transition to local solar is a plausibly significant trend - if these systems can themselves withstand damage. I am uncertain about the robustness of these systems to large solar storms, which may be critical, but they should at least have less exposure to the induced current than transformers connected to long-distance transmission lines.

Withstand

Withstanding an event would require that the electrical system not fail, or fail to a lesser extent, during an event. Thankfully, we have hours of warning for solar storms, and there is significant data collection and research on the impacts on the power system. Roodman highlighted that storms seem to damage transformers slowly, rather than causing immediate failure - but larger events would presumably cause more immediate damage. To prevent that, a number of short-term adaptations would allow power systems to proactively shut down or isolate sections of the grid to minimize damage. There is work on this, (including internationally,) though it is unclear to me to what extent such methods have been adapted. If such actions are undertaken, failures could be minimized and localized, making recovery easier, or reducing the extent to which adaptation is needed.

Risk Reduction

Risk reduction approaches include prevention, and reducing hazard^[1]. Prevention is often a better approach, but in this domain we aren’t (currently) able to change the likelihood of Coronal Mass Ejections, nor is preventing nuclear war in scope for this writeup.

Hazard reduction is in theory possible, but it is unclear how tractable it is. Most critically, a weakening geomagnetic field would increase the hazard experienced by the grid. Current weakening is probably a precursor to a flip, which will happen in the coming couple centuries. It is unclear to me, but during such a flip, there would be greatly increased vulnerability to solar storms. Preventing a flip seems infeasible at present, and the risks when it occurs are critical; this seems to argue for more investment in other mitigations, but also more research.

Somewhat related, initial analysis and speculation, which have been questioned, indicate that building megaconstellations like Starlink could exacerbate the risk. Ensuring the Earth’s geomagnetic field isn’t (further) weakened is a plausible risk-reduction mitigation, and is worthy of some attention. This could reduce the amount of damage that solar storms would do.  Additional medium-dive investigation into the hazard from a flip, and from satellites, and whether these can be feasibly mitigated, seems valuable, at the very least to better understand how valuable other mitigation pathways are.

Conclusions

It seems that the “recovery” options such as backup transformers, while simple, would not prevent disruptions and are easily the least cost-effective. Highlighting the lack of backup transformers is therefore largely a red-herring, even though it highlights that other methods are not fully able to address the risk.

Adapt and withstand approaches, on the other hand, are both feasible, and already pursued in research and by industry. At the same time, they are not currently adopted to an extent sufficient to withstand the most extreme events - but could plausibly be made so with the right regulatory policy and economic incentives. Research into the costs and feasibility of proactive shutdowns and grid isolation, and how it might work to complement other grid resilience measures, is high value. Similarly, it seems clear that there is room from important policy work on how to  motivate such measures, and which ones are most compatible with extant regulatory and engineering requirements.

Lastly, risk reduction is the most speculative and uncertain, but because of that, further investigation would be of high value - as long as it does not replace or delay investments in adapting and withstanding the risk.

1. ^{^}

   I will consider vulnerability reduction, rather than hazard reduction, to be resilience. (I’m not going to be careful about distinguishing hazard reduction and vulnerability reduction, though they do conceptually count as risk reduction. For example, things like reducing exposure by creating microgrids reduces vulnerability, but I consider it adaptation below instead.)