twitter: @enochsmith23                              email:

Thursday, July 5, 2012

Forecast: Sunny, 12% Chance of Catastrophe

A solar storm could blackout electricity for months to hundreds of millions of people in the United States, with a 12% chance of happening virtually anytime within the next ten years. This could cause the deaths of millions of people in the United States. The storm could cause similar blackouts around the world. There is scientific evidence to support all these statements. (Update, July 2013: Congress is again considering a bill to address this problem - the Secure High-voltage Infrastructure for Electricity from Lethal Damage Act - please urge Congress and the President to support this act.)

Blackout regions for a repeat of 1921 solar storm.
A recent paper estimates there is a 12% chance of a "Carrington event" solar storm within the next 10 years. (Riley, 2012) In a 2008 National Academy of Sciences report, Kappenman [1] estimates a repeat of a 1921 solar storm  smaller than a Carrington event would knockout 350 extra-high voltage (EHV) transformers on the US electric power grid, and that this could blackout electricity for 130 million people, potentially for months because of the time it takes to replace a single EHV transformer. Kappenman estimates a repeat of the 1859 Carrington event could "take out the entire [US] grid ... for weeks or perhaps months on end" (Ferris, 2012; emphasis added). Such blackouts could also happen around the world because the power grids of other nations have the same vulnerability, and a solar storm is a worldwide event.

The electric power grid is a foundation of modern civilization.[2] It transmits power for cities to pump water and sewage, for gas stations to pump fuel into cars, for refrigeration of food, for telecommunications, lighting and air conditioning, and for computers and the Internet, which are now critical for banking and commerce. If these services were halted across the nation or a multi-state region to hundreds of millions of people for a period of months, millions of people could perish. A human being can survive only a few days without water. Cars could run out of gasoline before people could leave the region.

The good news is that it may be relatively inexpensive to protect the EHV transformers, to prevent or reduce blackouts from a Carrington event. The bad news is that there is little public awareness and apparently insufficient industry and government consensus to solve the problem. Although in 2010 the US House passed a bill to address this problem with recommendations from the National Academy of Sciences and the Department of Energy, the Senate did not pass the bill. (Pry, 2010; Behr, 2011) Instead, the Senate put more funding into "clean energy", which would not address this problem. (Unruh, 2010)

The goal of this article is to help generate enough public awareness of the problem to ensure it is addressed before a catastrophe occurs. This article will focus just on the Carrington event, rather than the 1921 storm, since that is what Riley estimates has a 12% chance of happening in the next 10 years. Riley's method would estimate a repeat of the 1921 storm has a higher probability. 

Relative size of Sun, CME, and Earth (JPL/NASA and ESA)
About every day or so, a solar storm releases a billion-ton burst of electrons and protons, called a coronal mass ejection (CME), which travels away from the Sun at speeds averaging about a million miles per hour. [3] If a CME happens to be directed toward the Earth, it can impact the Earth's magnetic field, causing a geomagnetic storm to release a trillion watts of power into the atmosphere, producing auroras. An especially strong CME can cause geomagnetically induced currents (GIC's) to flow in high-voltage electric transmission lines, which effectively act as very long antennas. A GIC can cause 300 amperes of current to flow from the ground into a high voltage transformer, melting its copper winding. (Kappenman, 2012)

A 1989 solar storm much less powerful than either the 1921 solar storm or a Carrington event damaged transformers and caused transmission lines in Quebec's power grid to fail, blacking out power to 6 million people for 9 hours. (Boteler, 2012; Koza, 2008)

1989 damage to Quebec power grid. (Boteler,2012)

According to Kappenman (2012) "A later surge in the storm destroyed a large transformer at a New Jersey nuclear plant and nearly took down U.S. power grids from the mid-Atlantic through the Pacific Northwest."

Solar storm damage to an EHV transformer. (Kappenman, 2011)

Damage to transformer in South Africa. (Ostenwald, 2012)
In 1972, a solar storm caused a power transformer in British Columbia to explode (Ferris, 2012; Ostenwald, 2012).

A solar storm in 2003 caused a blackout in Sweden, inducing currents as high as 330 amperes on a transformer. The ESKOM Network in South Africa reported 15 transformers were damaged by high GIC currents. (Ostenwald, 2012)

Carrington's 1859 drawing of sunspots and flares.
The most powerful geomagnetic storm in recorded  history was in 1859, before any electric power grid existed. This event was named after the British astronomer Richard C. Carrington who observed sunspots and solar flares linked with it. According to the 2008 National Academy of Sciences report, "auroral displays of extraordinary brilliance were observed throughout North and South America, Europe, Asia, and Australia, and were seen as far south as Hawaii, the Caribbean, and Central America in the Northern Hemisphere and in the Southern Hemisphere as far north as Santiago, Chile ... and telegraph networks around the world—the “Victorian Internet”—experienced major disruptions and outages." Sparking caused fires in some telegraph offices. "In several locations, operators disconnected their systems from the batteries and sent messages using only the current induced by the aurora."

Quoting Kappenman (2012):
"Of all the parts of the power grid, high-voltage transformers are among the most likely to fail in a geomagnetic storm and also among the most difficult to replace. If a big storm were to knock out several hundred transformers in one fell swoop, manufacturers wouldn't be able to supply replacements quickly—there is no global stockpile. EHV transformers, which can handle voltages of 345 kV or higher, weigh about 200 tons and cost about $10 million each. Building one requires exquisite, near-artisanal craftsmanship, including meticulously hand winding the paper-tape insulation around the copper winding at the transformer's core. One EHV transformer can take several weeks to assemble and test, and it takes years to train skilled assemblers. Even the largest transformer plants can build only about 30 to 50 per year. With the shortage of skilled labor and specialized materials that would likely accompany a prolonged blackout, simply maintaining that level of output would be a challenge, never mind ramping up new production."
A 2009 Metatech report stated:
  • "Manufacturing capability in the world for EHV-class transformers continues to be limited relative to present market demand for these devices. Further, manufacturers would be unable to rapidly supply the large number of replacement transformers needed should the U.S. or other power grids suffer a major catastrophic loss of EHV Transformers.
  • "Manufacturers presently have a backlog of nearly 3 years for all EHV transformers (230 kV and above). ...
  • "Only one plant exists in the U.S. capable of manufacturing a transformer up to 345 kV. No manufacturing capability exists in the U.S. at present for 500 kV and 765 kV transformers, which represent the largest group of At-Risk transformers in the U.S."
Thus it appears the U.S. would depend on shipment of most at-risk EHV transformers from manufacturing plants located overseas. Kappenman (2012) notes the transformers are becoming more vulnerable: 
"Also worrisome is that many transformers in the United States, Western Europe, and Japan are fast approaching the end of their useful lives. The average age of U.S. transformers is about 40 years. Over the decades, each of these devices has likely experienced at least minor overheating and other insults from GICs. An aged transformer that's been exposed to repeated injury is of course far more likely to fail than a brand-new transformer."
This implies transformers could fail as a result of smaller, more frequent solar storms. 

Realistically, local backup generators can replace only a small fraction of the energy delivered via the electrical grid. If the catastrophe occurs, then backup generators will eventually run out of fuel and it may not be possible to get more fuel, since diesel and gasoline pumps run on electricity, and electricity provides power for storage and control systems in the oil industry.[2] Even so, backup generators might help buy time for recovery to mitigate a catastrophe, if they were deployed in advance in much greater numbers. This is further discussed below.

Likewise, solar power and wind power are at best only of limited help. If the grid is down because EHV transformers are knocked out, then energy from power plants cannot be transmitted to users. It doesn't matter whether the power plant is a large solar farm or wind farm. People could still use locally generated solar power for their homes and to recharge electric vehicles, but this offers at best a very limited mitigation.

Scientists cannot predict reliably whether a Carrington event will occur in the next few weeks, months, this year or next year, etc. The best prediction to date appears to be Riley's estimate of a 12% probability within the next 10 years. Riley writes "By virtue of their rarity, extreme space weather events, such as the Carrington event of 1859, are difficult to study, their rates of occurrence are difficult to estimate, and prediction of a specific future event is virtually impossible."

Scientists apparently do not yet know what is the most powerful space weather event that the Sun can generate, at its present age. For example, recently scientists observed the most powerful solar X-ray flare ever, briefly making the Sun the brightest gamma ray source in the sky. (Chow, 2012) And recently it was reported that "Sometime between 1,237 and 1,238 years ago, the earth was inundated by a massive blast of high-energy radiation greater than any known to have occurred either before or since – but no one knows its source." (Myslewski, 2012) Daniel Baker of the University of Colorado's Laboratory for Atmospheric and Space Physics was quoted as saying "I would like to think about whether a strong CME moving directly towards Earth could have produced the intense proton population that impacted Earth's atmosphere". 

It appears several other scientists and engineers agree with Kappenman that a Carrington event, or similar great geomagnetic storm, could cause extensive long-lasting damage to the electric power grid. A recent Wall Street Journal article (Tracy, 2012) quotes Peter Vincent Pry of of the Task Force on National and Homeland Security as saying "We already know that the danger to society is great enough to warrant taking immediate action." Frank Koza of PJM Interconnect is quoted as saying "no one can provide sufficient evidence that an immediate large-scale investment by the assets owner and government would adequately address the risk", which indicates agreement that there is a serious risk, at least.

Pulkkinen et al (2011) write "it is generally understood that geomagnetically induced currents (GIC) causing half-cycle saturation of high-voltage power transformers are the leading mode for the most severe problems such as electric blackouts and equipment damage", citing papers by Kappenman (1996) and Molinski (2002). However, Tracy (2012) quotes Pulkkinen as not being certain that a Carrington event would have the impact predicted by Kappenman, saying "We need to carry out more detailed and more rigorous analysis before we know for sure."[4]

Since solar storms even half the magnitude of the Carrington event happen rarely, it might not be possible to get enough data about them to support a rigorous analysis all scientists would agree with. And scientists may not have time to reach a complete consensus, before another Carrington event occurs. In that case it would be prudent to protect the grid from the scenarios predicted by worst case estimates, if it is possible to do so.

Although Riley (2012) estimates a 12% chance of another Carrington event within the next 10 years, an analysis of ice core data indicates the 1859 Carrington event was the most powerful solar storm in the past 450 years. (McCracken et al, 2001)  Thus, some might think that solar storms on the scale of the Carrington event can only happen "once in 500 years" and that it is very unlikely there will be two Carrington events between, say 1859 and 2022, a period of 163 years.

Magnitude of solar events from 1561-1994. (McCracken et al, 2001)
However, until scientists understand the complex dynamics of the Sun well enough to predict when such events will occur, it is more cautious to estimate probabilities of events within time periods, than to jump to a conclusion an event can only happen once in 500 years. Using a simple probability calculation, if a Carrington event has one chance in 500 to happen independently each year, and it happens in 1859, then it has a 28% chance of happening again within 163 years.

Mathematicians would object to applying independent probabilities retrospectively to the period from 1859 to 2011. If we estimate that the chance is 1 in 500 for a Carrington event to happen independently each year (which is a reasonable first estimate, based on the ice core data) then a simple probability calculation gives a 2% chance for a Carrington event to happen within the next ten years.

Riley's estimate of a 12% probability within the next ten years is much more scientific and mathematically sophisticated. [5] He writes: 
"By showing that the frequency of occurrence scales as an inverse power of the severity of [a space weather] event, and assuming that this relationship holds at higher magnitudes, we are able to estimate the probability that an event larger than some criteria will occur within a certain interval of time in the future. For example, the probability of another Carrington event (based on Dst < 850 nT) occurring within the next decade is ~12%."(Riley, 2012)
Riley makes it clear that 12% is only the best estimate he can give at this point as a space weather forecast. This is not a situation where one can calculate a definitive, exact probability. The science of solar physics is incomplete, the data is imperfect and at best supports approximate estimates.

From a policy and engineering perspective it does not matter whether the probability is 2% or 12%. We would not tolerate a design of an elevator that had one chance in a thousand of killing its occupants. If our electric power grid has vulnerabilities to a Carrington event that can cause millions of deaths, and trillions of dollars in losses anytime within the next 10 years, then we should take immediate action to prevent the deaths and damages, even if the preventive cost is several billion dollars. The risk may be highest in the next year or two, corresponding to expected solar cycle peak sunspot activity.

The bottom line is that betting the fate of millions of people against the probability of another Carrington event in the next decade is, in Kappenman's words, "playing Russian roulette with the Sun" (Ferris, 2012). For all we know, a Carrington event could happen two weeks from now, next month, or anytime afterwards.
To avoid or mitigate a catastrophe from a Carrington event, three strategies are possible:
  1. Space Weather Monitoring & Prediction: NASA presently monitors solar storms and issues warnings about CME's. Power companies could prevent damage to transformers by pre-emptively taking them offline if a CME is predicted to be extremely dangerous, intentionally causing short-term blackouts in order to prevent longer-term blackouts as a result of damaged transformers. While monitoring and prediction are important and worthwhile, any predictive method has the risk of false positives and false negatives, either of which could result in substantial costs. This also relies on  human judgment for the dynamic protection of the electrical grid, with the potential for human errors, compounded by possible previous false positives or negatives. Hence, monitoring and prediction should not be the only form of prevention or mitigation. 
  2. Protective Devices. Designs exist for protecting EHV transformers by placing resistors or high-powered vacuum tubes at the neutral to ground juncture. (Kappenman, 2012; Ostenwald, 2012). The benefit of such designs would be to continuously mitigate or prevent damage to EHV transformers, caused by geomagnetic storms. Kappenman estimates the cost to implement such designs in the U.S. to be on the order of $1 billion. (Stein, 2012)
  3. Backup Devices. Spare transformers could be manufactured and placed onsite at power plants for immediate use if necessary.  At $10 million each, 350 transformers would cost $3.5 billion.[6] The U.S. should develop the capability to manufacture these onshore, to reduce lead time and ensure supply. A year's supply of diesel fuel should be stored with the backup generators at each US nuclear facility, along with spare parts for the generators, to protect against Fukushima explosions. Kappenman estimates less than $1 billion for this. (Stein, 2012) Gasoline and diesel fueled backup generators could be an important line of defense to mitigate a catastrophe, if most gasoline stations had backup generators running on gasoline to power their pumps, if banks had backup generators to continue running computers, if the Internet had backup generators for key computers, if there were adequate backup generators for water and sewer pumps, etc. The economic and technical feasibility of this would need to be evaluated in the emergency program recommended below.
To provide maximum protection against a catastrophe, all of the above strategies should be combined. The initial estimates for items 2 and 3 total $5.5 billion. According to the 2008 National Academy report, the US financial impact of a solar superstorm could be 1 to 2 trillion dollars in the first year. If there is a 12% chance of a Carrington event within the next 10 years, then the mathematically expected impact is on the order of hundreds of billions of dollars, which more than justifies the cost of these protective measures. Of course, such calculations cannot comprehend the human suffering and loss of life from a multi-state or national blackout lasting for months. 

The protection of the electric grid should not be addressed by introducing new regulations and gradually phasing in changes, which is the present direction being considered by the Federal Energy Regulatory Commission (Tracy, 2012). That could be reasonable if it were certain a Carrington event would not occur within fifty years, but is not wise for a 12% chance of a catastrophe within ten years.

Protection of the nation's EHV transformers should be achieved by an emergency program of the federal government, moving with well-coordinated, deliberate speed to avert a national catastrophe as rapidly as possible. If possible, protective devices should be installed on key EHV transformers before the peak of solar activity predicted in early to mid 2013, or within the next year or two at most, since we do not know how much time we have.[7]

If research and engineering are needed to develop solutions for protecting the grid, then this work should be conducted with the centralized speed of a Manhattan Project, enlisting the best engineers and scientists in the nation. In parallel with developing solutions to protect the grid, researchers should continue to analyze the potential impact of a Carrington event on the grid, to try to reach a scientific consensus on its severity.

In strengthening the electric power grid to withstand a Carrington event, the emergency program should also strengthen it to withstand a nuclear EMP attack. This should be part of a larger, emergency effort to protect diverse national technologies (Internet, computers, electronics, etc.) against EMP attacks (viz. Foster et al, 2004;  Casti, 2012).[8]

The emergency program should also evaluate the economic and technical feasibility of backup devices and rapidly introduce them if possible. Spare transformers should be manufactured and placed onsite at power plants for immediate use if necessary. Engineering should be expedited to configure nuclear plants to perform cooling with electricity generated onsite during electric power blackouts, and not depend solely on backup generators to avoid Fukushima explosions.[9] If feasible, the program should ensure cities have backup generators to power water and sewer pumps, provide backup generators for the Internet, etc. There should be regulations requiring gas stations to have gasoline-powered backup generators enabling them to continue pumping fuel during an electric grid blackout;  requiring banks to have backup generators able to run their computers; etc.

The emergency program should be federally funded, since it involves the interstate electric power grid and the welfare of the nation. We cannot expect the states or the individual power utilities to independently cover the costs, in time to avert a potential catastrophe. The costs to protect EHV transformers from solar storms could not have been expected by the utilities when power plants were built. It's appropriate for the federal government to bear this burden, to protect the grid as rapidly as possible.

Knowledge should be shared with other nations. It will be to the benefit of each nation, if all nations can experience a Carrington event as just a brief blackout. If other nations require help to avoid a catastrophe, this would be an appropriate use of the budget for foreign aid. 

One hundred years ago, a ship was launched that its owners said was "designed to be unsinkable". Thinking the ship would remain afloat if it struck an iceberg, it did not carry enough lifeboats for everyone aboard. On its maiden voyage, it hit an iceberg and sank, with the deaths of 1,514 people. The ship had been designed to stay afloat even if four of its compartments were flooded, but the blow of the iceberg caused five compartments to flood. Its naval architect was on board and realized quickly the ship was doomed, after the fact. He went down with the ship, trying to help save as many lives as possible.

Over the past one hundred years, engineers have constructed a vast electric power grid for "spaceship Earth". Virtually every aspect of human civilization now depends on this grid. Only recently have we learned that spaceship Earth's electric power grid is vulnerable to a blow from the Sun, which in the worst case could leave human civilization across large geographic areas "adrift at land", eventually without water, without food, and without fuel, for months on end.
Some owners of the electric power grid may believe it is unsinkable. Some in industry and government may believe a solar storm strong enough to take down the grid for months on end is unlikely to happen. There is reluctance to spend money on lifeboats for the EHV transformers that are vital to the grid, though several scientists and engineers warn they are at risk. 

In a recent PBS Nova program, James Green of NASA Planetary Science, who discovered the 1859 reports of the Carrington event, said:
 "It has the potential of knocking out power grids. And if it burns out transformers that are hard to replace, we may be without electricity in many areas for a very long period of time. ... It takes several months to make them, and if you burn out half of them, we're going to be shooting squirrels and chopping wood out the back yard for a long period of time, just to survive."
Of course, most people cannot survive by shooting squirrels and chopping wood. This was a way of saying the situation would be dire. To support this article's statement that millions of lives could be lost, an "absolute worst case scenario" is presented in note [10]. 

Study the available information. If you agree this is a problem worth solving, building awareness is the first step - Please tweet or email a link to this article, asking people to read it:
Also, please tell people whom you know about the problem, and urge Congress and President Obama to support an emergency program to solve it.

You should also try to prepare for survival if a catastrophe occurs, of course.

Enoch Smith is a concerned US citizen, writing as a civic duty in the constructive tradition of Publius and the Federalist Papers. These writings should be evaluated solely on their facts and logic. They were not written by anyone mentioned in them. 

[1] John Kappenman is a power engineer and consultant who has studied the effects of solar storms on electrical power grids for thirty years. The references section below cites some of his published papers. According to his web page he has "provided presentations to the US Presidents’ Commission on Critical Infrastructure Protection on the Potential Impact of Geomagnetic Storms on Electric Power System Reliability ... served on the Science Advisors Panel for the NOAA Space Environment Center ... [been] one of the principle investigators under contract with the Congressional Commission to Assess the Threat to the United States from Electromagnetic Pulse ... presented testimony before the US House Science Committee in October 2003 on the importance of geomagnetic storm forecasting for the electric power industry ... was a principal investigator examining the Vulnerability of the Electric Power Grid for Severe Geomagnetic Storms for FEMA under Executive Order 13407 ... was also one of the Principal Contributors to the 2008 US National Academy of Sciences Report on Severe Space Weather Events"

[2] Quoting Kappenman (2003a):

Major infrastructure dependencies. (Kappenman, 2003a; Source: US DOE)
"As this illustrates, electric power supply is central to the sustained operation of most of the nation’s other critical infrastructures. Only a small portion of these infrastructure facilities have emergency on-site generation of sufficient capacity that allows them to continue operation in the face of a blackout event. Water treatment and pumping require enormous amounts of electric power and as result very few of these systems have redundant power supply options. Loss of pumping in time will lead to drop of city water pressure, as storage tanks and reservoirs cannot be recharged for residential distribution. In large high-rise buildings, city supply water pressure needs to be supplemented with electric pumps to lift water to upper floors for water distribution. Therefore within a matter of a few hours potable water distribution in many locations can become a serious concern. Perishable foods are generally at risk of complete loss within 12 hours or less. ... nearly all refueling operations from underground storage tanks require restoration of electric power supply. ... backup generation at a few critical hubs ... generally have around 72 hours of available fuel. Therefore power grid outages of longer durations would be highly problematic in that refueling may be logistically impossible in all situations."
The situation related to backup generators is illustrated by the following table:
US energy consumption in Petawatts. (Source: Wikipedia, US EIA)

Theoretically, if the entire oil consumption of the US were diverted to local backup generators fueled by diesel or gasoline, then it could replace the power generated by nuclear, coal, and hydroelectric energy sources while the electric power grid is down. However, this would be logistically impossible during a blackout, and leave little fuel for transportation.

[3] The frequency of CME's ranges from about once every five days during solar minima, to about three times a day during solar maxima. The 1859 Carrington event was actually two CME's hitting the Earth within a period of a few days (August 28 to September 2), with the second CME causing the problems observed with telegraphs. So, a Carrington event may be more likely to occur during a solar maxima period, such as the Sun is now entering.

Xie et al (2005) studied solar storms lasting longer than 3 days. They found about 65% were caused by successive CME's, and that the most intense storms involved multiple CME's interacting with each other and with high-speed solar wind streams. The 1921 solar superstorm was described at the time as "a protracted series of magnetic storms" from May 12-21 (Cortie, 1921). Boteler (2012) comments that the 1989 solar storm causing the Quebec blackout was due to two CME's. The "Halloween solar storm" involved multiple CME's causing three geomagnetic storms between October 29 and November 2, 2003. During the current solar cycle, a series of four large CME's impacted Earth between August 4-5, 2010. On August 10, 2011, three consecutive CME's produced auroras as far south as Oklahoma and Alabama.

This means that EHV transformers need to be protected against a multiple whammy of CME's happening within a few days to a few weeks. If feasible, the protection technology needs to be robust, able to handle any number of large CME's happening in series. We don't want to repeat the Titanic error of protecting against N events and then sinking because of N+1.

[4] In assessing the potential impact of a 100-year geomagnetic storm, Pulkkinen et al (2011) discuss information about the geomagnetic field recorded during the Carrington event:
"...from the viewpoint of the analysis in this paper, perhaps the most significant observation during the event comes from Rome, Italy ... the observation indicates that the maximum expansion of the auroral current system may have been about 20 degrees more southward than during the March 1989 or October 2003 storms. Although this may sound somewhat fantastic and the single data point was not based on modern scientific instrumentation, one cannot simply disregard the Rome observation. ..  there is no direct means to quantify the likelihood of such occurrence. ... we are inclined to hold our conjecture that the threshold at about 50-55 degrees of geomagnetic latitude holds for most major and extreme geomagnetic storms - possibly also for 100-year events. ... However, it is emphasized again that more extreme geomagnetic storm data is required for more definite conclusions regarding the 100-year location of the threshold geomagnetic latitude."
If the geomagnetic field can be affected at lower latitudes, this would indicate a broader region of potential impact on the electric grid. They consider different explanations including the possibility the Rome observation was erroneous. It does not appear they consider an explanation due to the Carrington event being two CME's: perhaps the first CME weakened the geomagnetic field, allowing the second CME to affect lower latitudes. As noted previously, during the Carrington event auroras were observed at latitudes as far south as Hawaii, the Caribbean, and Central America in the Northern Hemisphere, according to the 2008 National Academy report.

[5] Riley's paper analyzes four measures of severity for space weather events, noting that many others could be investigated in future research. For two of these measures (CME speed and strength of geomagnetic storms measured in the 'disturbance-storm time' index, or Dst)  he derives an estimate of 12% probability for a Carrington event to occur within the next 10 years. He notes this agreement may be a coincidence. For a third data set, regarding strength of X-ray flares, he finds the data is insufficient to estimate probability of a repeat Carrington event. The ice core analysis of McCracken et al (2001) provides a fourth data set, from which Riley derives a 3% probability of a Carrington event to occur in the next 10 years. He writes:
The value of [the ice core] data lies in their long time span, going back more than 400 years; however, they are not without caveats. First, while the nitrate spikes are generally believed by space physicists to be a record of large, historical space weather events ... ice core chemists are skeptical [Wolff et al., 2008]. They posit that no viable mechanism exists by which Solar Proton Events could be imprinted within the ice, suggesting instead that high concentrations of sea salt provide a simpler and more consistent explanation for the deposition of aerosol nitrates. Second, there are only 70 events spanning the 450 years for which we have data. ... with such a limited number of events, the statistics of the fit and the resulting probability estimates will be more prone to error.
Riley discusses alternate methods for estimating probabilities of rare events, explaining why he uses power law distributions for rare space weather events. His method depends on the probabilities involved not changing over time, i.e. having "time stationarity". He writes:
For the Sun, this is clearly an approximation that requires careful assessment. On the scale of a decade or so, the solar activity cycle modulates many solar parameters ... The largest 2% of geomagnetic storms (the so-called “super storms”), for example, tend to occur shortly after the maxima of the decadal-scale solar activity cycles, occurring most often near the equinoxes ... any predictions made over say the next decade would necessarily be solar cycle-averaged predictions, and the actual probability of an extreme space weather event at some point in the cycle may be different.
For the Dst data set, he notes:
Another potentially important trend is the tendency for the strongest storms to become stronger over the last four solar cycle maxima. Whereas the top 5 storms around 1970 lay between 200 and 300 nT, by 2005, the five most intense storms lay between 350 and 450 nT. If such a trend is real it suggests a basic nonstationarity in the data on the same scale as the duration of the data set, and that predictions of future events may underestimate their true probability.
For the ice core data set, he writes:
Although it is not possible to show rigorously, because of the limited sample size, there is no obvious trend in the distribution of event sizes or temporal clustering to suggest that the time series is obviously nontime stationary.
 More generally, he writes:
Over longer time scales, there is ample evidence for nonstationarity. Solanki et al ...argued that compared to activity over the last 11,000 years, the last 70 years have been a time of exceptional activity. Steinhilber et al ...also found that solar activity now is stronger than 85% of the time over the last 9,300 years. ... If the time series were not stationary in the past, they are no more likely to be stationary in the future. ... Abreu et al ...estimated that the current grand maximum is only likely to persist for another 15–36 years. Similarly, Lockwood et al ...argued that solar activity rose during the first half of the 20th century, peaking in years 1955 and 1986, and subsequently declined, with the grand maximum ending within the next 20 years or so. Lockwood et al...further estimated that there is a 10% probability that activity on the Sun will decrease producing grand minimum conditions during the next 40 years. Obviously, if such conditions do ensue, probabilistic forecasts based on more active solar conditions may be less accurate.
Finally, he notes that his analysis does not cover the time-nonstationarity of solar storms that involve multiple CME's happening in sequence. So, there is much more work that can be done to refine future estimates for likelihood of a Carrington event. Riley's estimate of a 12% chance within 10 years is the best estimate he can give at present, based on his analyses.

[6] Behr (2011) reports, quoting Richard Lordan of EPRI:
The Department of Homeland Security has funded EPRI's design of a modular replacement transformer that is now being tested. It can be adapted to the range of substation configurations around the grid, and shipped in three pieces by truck to wherever it was needed. It will be installed for field testing in 2012. But there are a host of  unanswered policy questions before the replacement transformers could provide effective backup...beginning with how they would be paid for. "What would be the appropriate deployment strategy? How many are needed? Who owns them? Who maintains them? And who determines when an event is severe enough to warrant deployment?" he said. "These conversations are going on."
Rather than shipping replacement transformers after failures occur, the nation would recover more quickly and more surely from a loss of many transformers if replacements were already onsite, ready for use. In any case, it appears that development of an adequate supply of backup transformers will necessarily be a longer-term activity. Installation of protective devices on existing transformers appears to be the most feasible way of quickly preventing or mitigating the problem.

[7] Boteler (2012) describes an edge effect, which indicates the key transformers to protect may be those at the start and end of long networks.
Source: Boteler (2012)
It seems that if the outer transformers are protected, then the edge effect could move inward to the next transformers, and so on - this is a topic for engineering design to protect the grid.

NASA predicts the peak sunspot activity of the current solar cycle to occur in early or mid 2013. The current solar cycle may have the least peak sunspot activity in the past 100 years, which is one reason for optimism.

Source: Ostenwald (2012)
On the other hand, the 1859 Carrington event was in a solar cycle with peak sunspot activity lower than most of the cycles over the past sixty years.

[8] In theory, a single nuclear missile launched from a ship in the Gulf of Mexico and causing a nuclear explosion 200 miles above Kansas, could generate an electromagnetic (EMP) pulse that could potentially knock out the electric grid and virtually all electronics in the US (all computers, all cars that have electronic ignition, ...). This is a major reason why the US is at risk of attack from rogue nuclear nations like North Korea or potentially Iran. Some ideas the Manhattan Project could investigate include: Building Faraday cages into new automobiles, to protect their electronics; Building Faraday cages around critical computers for banking, the internet and telecommunications, as well as around ATM's, PC's, TVs, etc; Developing and applying extreme-high spike protection for the E1 component of EMP, to protect devices connected to the electric power grid. It may be much more expensive to protect the nation from an EMP attack than from a Carrington Event, but it should be done, if feasible.

[9] That is, normally nuclear generators should perform cooling using electrical power from the grid, as they currently do in the US, so that they can be cooled by the grid during maintenance. During an electric power blackout, a reactor should be cooled with the electricity it generates onsite. Backup generators should only be needed for cooling if a nuclear plant must be shut down for maintenance during a blackout. This strategy would reduce the risk of Fukushima explosions for current nuclear plants during an extended blackout. New technologies for nuclear power generation, such as pebble bed reactors, are designed to be passively safe and be cooled by natural circulation. Because of its efficiency in generating large amounts of energy, nuclear power remains necessary to support US energy needs and economic growth.

Following is a description of what could conceivably happen as a worst case, not a prediction of what will happen. Hopefully governments will take steps to protect electric power grids so that a future Carrington event causes just brief blackouts, and few if any deaths occur.  If no such steps are taken, then perhaps a Carrington event will not cause such widespread damage. At this point there is not a complete scientific consensus, but at least one expert has said the entire US grid could be down for weeks or months on end. This scenario extrapolates that prediction to the rest of the world, since a Carrington event will be a worldwide geomagnetic storm:

A Carrington event occurs, and humanity is not prepared to prevent or reduce its impact. It takes down electrical grids around the world. Collectively, nations around the world need to replace thousands of EHV transformers. Every continent except Antarctica is blacked out. The problem is so severe that no nation could possibly restore electrical power for several months.

Highway jam due to hurricane Rita, 2005.
Several technologies are shut down immediately when the Blackout occurs. People no longer have air conditioning or refrigeration. Food in homes and stores begins to spoil. Gasoline stations cannot pump fuel. Electric cars cannot be recharged, except where solar power is available off the grid. After a few weeks almost all cars have run out of fuel, and highways are jammed by vehicles with empty tanks. The railroads halt due to lack of fuel and electricity. Air travel halts due to fuel shortages. What food or fuel exists cannot be delivered. Cities without backup generators cannot pump water or sewage for the public. Those with backup generators stop when they run out of fuel. Markets quickly run out of food because they depend on just-in-time shipments. Farming is virtually halted in developed nations, where farms depend on electricity and fuel. Ocean transport is halted, due to lack of fuel. There is no ability to ship replacement EHV transformers even if they were available. Solar and wind power cannot be transmitted because the grid is down. Where solar power is available locally in homes, it provides little relief, nor could it even if every home had solar power: people need food and water. Rioting and looting become widespread. Sewage problems lead to outbreaks of dysentery.

Cell phones, smart phones, the Internet and GPS all stop working when the Blackout occurs. Land-line telephones continue to work, running on batteries and backup generators, but are jammed by an overload of calls. It takes hours to connect a phone call. There is no electricity for television and radios. People can hear emergency radio messages if they have battery-powered radios. Eventually the land-line telephones and battery-powered radios run out of energy.

Because the Internet is down, and there is no electricity to run computers, commerce stops. People cannot get cash at ATMs. Bank tellers cannot lookup account balances to give cash to customers, or perform transfers between accounts. Stores cannot process credit cards. They are out of stock anyway, and can't get new shipments. Manufacturing is halted, because plants don't have power, cannot communicate, and depend on just-in-time shipments of new stocks. New EHV transformers cannot be manufactured, even by hand.

One of  the four Fukushima explosions. Unit 3, March 14 2011.
Nuclear power plants begin running backup generators to keep their cores cooled when the Blackout starts, because they have not yet been configured to use the electricity generated onsite in a blackout. Due to the Blackout, delivery of new fuel is prevented in this worst case scenario. When the backup generators fail or run out of fuel, the reactor cores and spent fuel rods begin to melt down. After several weeks, 104 nuclear reactors in the US explode like Fukushima, exposing up to 1/3 of the US population to radiation. Around the world a total of 450 nuclear reactors explode. (Kappenman, 2012; Stein, 2012)

After some weeks of the above conditions, without water, food, fuel or electricity, civilization around the world has collapsed. Within a few months, every nation directly affected by the Blackout has lost almost all of its population, due to dehydration, starvation, violence, disease and radioactive poisoning. The worldwide loss of lives goes from millions to billions. Developing and under-developed nations are also decimated, because of their dependence on the developed nations, and their dependence on electricity and oil.

In such unstable conditions, there is a greater chance of nuclear or conventional war. Because communications were halted by the Blackout, the leaders of some nations may not know what happened, or may not believe it was a solar storm. They may think they suffered a nuclear EMP attack, and decide to retaliate against supposed enemies. Aggressive, power-hungry leaders may calculate their military forces can survive the Blackout if they attack other, weakened nations.

Even if war does not occur, in this absolute worst case scenario the loss of life is so great, and the breakdown in manufacturing and transportation so widespread that months turn into years and the electrical power grids are never restored.

A small percentage of people around the world survive, those in remote locations who are able to live self-sufficiently without electricity and gasoline. Humanity is reduced to a pre-industrial state, perhaps to the Stone Age. The knowledge necessary to rebuild the current level of technology may be lost for centuries or millennia. If and when it is regained, people may no longer remember what happened to this civilization, and may not know enough to avoid a repeat collapse. 

Behr, P. (2011) This Week's Solar Flare Illuminates the Grid's Vulnerability. The New York Times, June 9, 2011.

Boteler, D. (2012) The Impact of Space Weather on the Electric Power Grid.

Casti, J. (2012) X-Events - The Collapse of Everything. HarperCollins Publishers.

Chow, D. (2012) Massive solar flare in March broke sun storm, June 16, 2012.

Cortie, A. L. (1921) The Sun-Spot Group and the Magnetic Disturbances, 1921 May 8-21. Nature, June 2, 1921.

Ferris, T. (2012) Sun Struck - The space weather forecast for the next few years: solar storms, with a chance of catastrophic blackouts. Are we prepared? National Geographic, June 2012, pp. 36-53.

Foster, John S. et al (2004) Report of the Commission to Assess the Threat to the United States from Electromagnetic Pulse (EMP) Attack. Volume 1: Executive Report.

Jaggard, V. (2011) As Sun Storms Ramp Up, Electric Grid Braces for Impact. National Geographic Daily News. August 3, 2011.

Kappenman, J.G. (1996) Geomagnetic Storms and Their Impact on Power Systems. IEEE Power Eng. Rev., May 1996, 5.

Kappenman, J. G. (2003a) The Vulnerability of the US Electric Power Grid to Space Weather and the Role of Space Weather Forecasting. Prepared Testimony before U.S. House Subcommittee on Environment, Technology & Standards Subcommittee Hearing on “What is Space Weather and Who Should Forecast It?”. October 30, 2003.

Kappenman, J. G. (2003b) Storm sudden commencement events and the associated geomagnetically induced current risks to ground-based systems at low-latitude and midlatitude locations. Space Weather, 1(3), 1016, doi:10.1029/2003SW000009.

Kappenman, J.G. (2006) Great geomagnetic storms and extreme impulsive geomagnetic field disturbance events An analysis of observational evidence including the great storm of May 1921. Advances in Space Research, 38, 188199/

Kappenman, J. G. (2008) The vulnerability of the U.S. electric power grid to severe space weather events, and future outlook. In Severe Space Weather Events--Understanding Societal and Economic Impacts. Workshop Report. National Research Council, National Academy of Sciences.

Kappenman, J. G. (2011)  Space Weather and US Electric Power Grid Vulnerabilities: An Overview of the Risks to this Critical Infrastructure and Research Necessary to Assess Vulnerability and Mitigate Impacts. Presentation at NASA Workshop on Space Weather Risks and Society, Oct. 16, 2011.

Kappenman, J. G. (2012)  A Perfect Storm of Planetary Proportions. IEEE Spectrum, February 2012

Koza, F. (2008) Space Weather Impacts on the Electric Power System. In Severe Space Weather Events--Understanding Societal and Economic Impacts. Workshop Report. National Research Council, National Academy of Sciences.

McCracken, K. G., G A. M. Dreschhoff, D. F. Smart, and M. A. Shea (2001) Solar cosmic ray events for the period 1561-1994. Journal of Geophysical Research, Vol. 106, No. A10, pp. 21,599-21,609.

Metatech (2009) An Overview of the National Academy of Sciences Report on Severe Space Weather and the ulnerability of US Electric Power Grid. Press Release, January 11, 2009.

Molinski, T. (2002) Why utilities respect geomagnetically induced currents. Journal of Atmospheric and Solar-Terrestrial Physics, 64, 17651778.

Myslewski, R. (2012) Earth bathed in high-energy radiation from colossal mystery blast - Around 1200 years ago, that is. The Register, June 5, 2012.

Ostenwald, S. (2012)  Space Weather - Impacts, Mitigation and Forecasting.

PBS Nova (2012) Secrets of the Sun - With new tools, scientists are striving to better grasp our star and its potentially widely destructive solar storms. Aired April 25, 2012.

Pry, P. V. (2010) Senate Fumbles EMP Protection. WorldNetDaily, August 19, 2010.

Pulkkinen, A., E. Bernabeu and J. Eichner (2011) Generation of 100-year geomagnetically induced current scenarios. Space Weather, Vol. 10, S04003.

Riley, P. (2012) On the probability of occurrence of extreme space weather events. Space Weather: The International Journal of Research and Applications, Vol. 10, S02012, 12 PP., 2012, doi:10.1029/2011SW000734

Smith, T. R. (2012) Key nuclear plant safety order likely to be delayed, NRC tells Senate. The Washington Post, March 15, 2012.

Space Studies Board (2008) Severe Space Weather Events - Understanding Societal and Economic Impacts. Workshop Report. National Research Council, National Academy of Sciences.

Stein, M. (2012) Four Hundred Chernobyls: Solar Flares, Electromagnetic Pulses and Nuclear Armageddon. Truthout, March 24, 2012.

Tracy, R. (2012) Here Comes the Sunstorm - Electric Grid Is Vulnerable to a Big Solar Blow; Officials Spar Over What to Do. Wall Street Journal, May 14, 2012.

Unruh, R. (2010) U.S. warned of threat worse than Katrina, plague, WWII. WorldNetDaily, September 27, 2010.

Viereck, R. (2007) Space Weather: What Is It? How Will It Affect You? NOAA Space Environment Center.

Wolff, E. W., A. E. Jones, S. J.-B. Bauguitte, and R. A. Salmon (2008) The interpretation of spikes and trends in concentration of nitrate in polar ice cores, based on evidence from snow and atmospheric measurements. Atmos. Chem. Phys., 8(18), 5627–5634, doi:10.5194/acp-8-5627-2008.

Xie, H., N. Gopalswamy, P. K. Manoharan, A. Lara, S. Yashiro, and S. Lapri (2005) Long Lived Geomagnetic Storms and Coronal Mass Ejections. Journal of Geophysical Research, submitted 2005.