Blotting Out the Sun
Many smart people are curious about the risk of sunlight dimming due to volcanic eruptions for societies that heavily rely on solar energy.
The answer is short - there isn't much risk. The key is the interaction between the sulfur plume and the likely configurations of solar-dominated grids.
Tambora as a Guide
Mount Tambora erupted in 1815 in the largest volcanic eruption in recorded human history. 1816 was known as the "Year without a Summer" due to eruption-induced cooling.
Releases like this have two stages. The first is the ash and soot that surrounds the immediate eruption site. Ash blocks 20%-95% of sunlight near the eruption site, but usually subsides within 1-2 days. The second stage is the release of sulfur dioxide that spreads around the globe and peaks several months after the eruption. Sulfur dioxide is an aerosol that dims the sun and causes cooling.
Tambora was a VEI-7 on a scale of 0-8. Like earthquakes, it is a log scale, with the measurement being volume of magma. The scale does not predict sulfur release or sunlight dimming. Estimates from ice cores are that Tambora released 3.5 times more sulfur than Pinatubo (VEI-6), but only had double the radiative forcing effect. There are feedback loops that limit dimming. For instance, higher concentrations of SO2 form larger particles that are less effective as aerosols and drop out faster.
Estimates for the peak impact on radiative flux for a VEI-7 event are between 5 and 10 W/m2 out of the average daily figure of ~330 W/m2. The peak impact was between 1.5% and 3%. The peaks are relatively short-lived, 1-3 months, with much less effect on the shoulders that stretch 1-2 years.
One hypothetical model of a sulfur release 100x Pinabtubo (imitating VEI-8) estimated a 15%-18% decrease in flux, but more recent studies suggest that self-limiting likely keeps flux reduction somewhere in the 5-20 W/m2 range, or a 1.5%-6% decline.
The rough estimates for frequency are 50-100 years for VEI-6, 500-1000 years for VEI-7, and more than 50,000 years for VEI-8.
To summarize:
- The immediate eruption site and surrounding area will face a 1-2 day 20%-95% reduction in sunlight.
- Eruptions that happen on human timescales reduce sunlight by 0.5%-3% at their peak for a few months.
- The impact of "once-in-humanity" super eruptions is uncertain, but possibly only slightly worse than more common events. Worst-case estimates are a 15% reduction.
Features of Solar-Dominated Grids
Brian Potter has a recent series on the cost of solar at various market shares. A key feature is that solar capacity is overbuilt at high market share. Solar panels are cheaper than batteries. The economically rational path is to buy enough batteries to get through the day and extra solar panels to produce enough electricity on cloudy days. The last 5%-10% of generation comes from natural gas power plants or a similar technology.
The relevant takeaway is that there is double-digit extra solar capacity for 80%-90% of hours in solar-dominated grids.
Meshing the Models
Initial Disaster Area
The initial area is subject to extra cloudy weather for 1-2 days. Air conditioning demand should decrease. Solar output will be greater than zero, and flexible generation will step up to charge batteries.
At worst, electricity prices will need to increase substantially for a few hours, or there will be a few brownouts if markets are not in place to balance supply and demand. The impact should be similar to cold snaps or heat waves.
Wider Global Impact
I did some rough modeling to bound the effects. Here are the assumptions:
- Solar's market share is 90%.
- One day's worth of batteries.
- Constant daily demand.
- Baseline hourly solar output based on average US weather patterns.
- The remaining 10% of generation is a fuel like natural gas.
The constant daily demand assumption is decent because peak summer demand and solar output match. Growing demand categories, such as EVs and data centers, flatten demand.
The way sulfur impacts sunlight and weather means that solar PV output will not fall as much as radiative flux. The drop will be 1/3 to 1/4 less severe. Lower temperatures and a shift towards the redder part of the spectrum help solar PV production. Technologies that rely on direct sunlight, such as concentrated solar power, would fare much worse. Solar PV utilizes scattered light well. Most models also predict drops in rainfall, suggesting cloud cover might be less.
The excess capacity in solar arrays means there is no loss in net output in the vast majority of hours. Dimming sunlight a few percent, like a VEI-7 might do, barely affects grid operation. The extra fuel burn during cloudy days is of the same magnitude as a cold winter, with net energy demand increasing by ~1%. Prices may need to increase temporarily to balance fuel supply and demand, but this is not unprecedented.
Grid operations won't be affected on most days, even under the most extreme estimates of 15% sunlight reductions (10%-12% PV reduction). 85% of hours still have excess solar and battery capacity, only slightly worse than usual. The situation would be especially manageable if the eruption peak cooling impact were during the Northern Hemisphere summer. Even a winter peak would likely be less severe than the recent European fuel shortage following Russia's invasion of Ukraine in 2022.
Space heating is a special case. First, only about 25%-30% of the global population needs heating. It seems unlikely that electric air-source heat pumps would dominate heating with only daily storage. In real life, there would be massive amounts of storage. It could be any mix of geothermal, hydro, fuels, thermal, nuclear, electrochemical, insulation, etc. Heating is also the easiest energy consumption to adjust down comfortably if you "Jimmy Carter" it. The heating strategy will depend heavily on the context of whatever system is in place. Today's system, which is fuel-based, would fare fine.
A final key point is that the better markets work, the easier it will be to manage the effect. The options will be better if the economy continues to electrify. Moving EV charging to a sunnier day, increasing AI token pricing on cloudy days, or adjusting the thermostat down a few degrees to save money in winter might be all it takes, even during a 50,000-year event.
Conclusion
Solar PV overcapacity in solar-dominated electricity grids is much larger than the impacts on sunlight and solar PV output from massive volcanic eruptions. There is little reason to plan for them outside of actions that should be taken anyway, such as developing flexible electricity markets for supply and demand. There is no reason to favor other technologies in fear of volcanic risk. These results make sense in hindsight; humanity would have starved if volcanoes dimmed sunlight much more.
Are Volcanoes a Risk to Solar Dominated Grids?
2025 July 16 Twitter Substack See all postsIt seems grids will fare fine because of non linear affects.
Blotting Out the Sun
Many smart people are curious about the risk of sunlight dimming due to volcanic eruptions for societies that heavily rely on solar energy.
The answer is short - there isn't much risk. The key is the interaction between the sulfur plume and the likely configurations of solar-dominated grids.
Tambora as a Guide
Mount Tambora erupted in 1815 in the largest volcanic eruption in recorded human history. 1816 was known as the "Year without a Summer" due to eruption-induced cooling.
Releases like this have two stages. The first is the ash and soot that surrounds the immediate eruption site. Ash blocks 20%-95% of sunlight near the eruption site, but usually subsides within 1-2 days. The second stage is the release of sulfur dioxide that spreads around the globe and peaks several months after the eruption. Sulfur dioxide is an aerosol that dims the sun and causes cooling.
Tambora was a VEI-7 on a scale of 0-8. Like earthquakes, it is a log scale, with the measurement being volume of magma. The scale does not predict sulfur release or sunlight dimming. Estimates from ice cores are that Tambora released 3.5 times more sulfur than Pinatubo (VEI-6), but only had double the radiative forcing effect. There are feedback loops that limit dimming. For instance, higher concentrations of SO2 form larger particles that are less effective as aerosols and drop out faster.
Estimates for the peak impact on radiative flux for a VEI-7 event are between 5 and 10 W/m2 out of the average daily figure of ~330 W/m2. The peak impact was between 1.5% and 3%. The peaks are relatively short-lived, 1-3 months, with much less effect on the shoulders that stretch 1-2 years.
One hypothetical model of a sulfur release 100x Pinabtubo (imitating VEI-8) estimated a 15%-18% decrease in flux, but more recent studies suggest that self-limiting likely keeps flux reduction somewhere in the 5-20 W/m2 range, or a 1.5%-6% decline.
The rough estimates for frequency are 50-100 years for VEI-6, 500-1000 years for VEI-7, and more than 50,000 years for VEI-8.
To summarize:
Features of Solar-Dominated Grids
Brian Potter has a recent series on the cost of solar at various market shares. A key feature is that solar capacity is overbuilt at high market share. Solar panels are cheaper than batteries. The economically rational path is to buy enough batteries to get through the day and extra solar panels to produce enough electricity on cloudy days. The last 5%-10% of generation comes from natural gas power plants or a similar technology.
The relevant takeaway is that there is double-digit extra solar capacity for 80%-90% of hours in solar-dominated grids.
Meshing the Models
Initial Disaster Area
The initial area is subject to extra cloudy weather for 1-2 days. Air conditioning demand should decrease. Solar output will be greater than zero, and flexible generation will step up to charge batteries.
At worst, electricity prices will need to increase substantially for a few hours, or there will be a few brownouts if markets are not in place to balance supply and demand. The impact should be similar to cold snaps or heat waves.
Wider Global Impact
I did some rough modeling to bound the effects. Here are the assumptions:
The constant daily demand assumption is decent because peak summer demand and solar output match. Growing demand categories, such as EVs and data centers, flatten demand.
The way sulfur impacts sunlight and weather means that solar PV output will not fall as much as radiative flux. The drop will be 1/3 to 1/4 less severe. Lower temperatures and a shift towards the redder part of the spectrum help solar PV production. Technologies that rely on direct sunlight, such as concentrated solar power, would fare much worse. Solar PV utilizes scattered light well. Most models also predict drops in rainfall, suggesting cloud cover might be less.
The excess capacity in solar arrays means there is no loss in net output in the vast majority of hours. Dimming sunlight a few percent, like a VEI-7 might do, barely affects grid operation. The extra fuel burn during cloudy days is of the same magnitude as a cold winter, with net energy demand increasing by ~1%. Prices may need to increase temporarily to balance fuel supply and demand, but this is not unprecedented.
Grid operations won't be affected on most days, even under the most extreme estimates of 15% sunlight reductions (10%-12% PV reduction). 85% of hours still have excess solar and battery capacity, only slightly worse than usual. The situation would be especially manageable if the eruption peak cooling impact were during the Northern Hemisphere summer. Even a winter peak would likely be less severe than the recent European fuel shortage following Russia's invasion of Ukraine in 2022.
Space heating is a special case. First, only about 25%-30% of the global population needs heating. It seems unlikely that electric air-source heat pumps would dominate heating with only daily storage. In real life, there would be massive amounts of storage. It could be any mix of geothermal, hydro, fuels, thermal, nuclear, electrochemical, insulation, etc. Heating is also the easiest energy consumption to adjust down comfortably if you "Jimmy Carter" it. The heating strategy will depend heavily on the context of whatever system is in place. Today's system, which is fuel-based, would fare fine.
A final key point is that the better markets work, the easier it will be to manage the effect. The options will be better if the economy continues to electrify. Moving EV charging to a sunnier day, increasing AI token pricing on cloudy days, or adjusting the thermostat down a few degrees to save money in winter might be all it takes, even during a 50,000-year event.
Conclusion
Solar PV overcapacity in solar-dominated electricity grids is much larger than the impacts on sunlight and solar PV output from massive volcanic eruptions. There is little reason to plan for them outside of actions that should be taken anyway, such as developing flexible electricity markets for supply and demand. There is no reason to favor other technologies in fear of volcanic risk. These results make sense in hindsight; humanity would have starved if volcanoes dimmed sunlight much more.