Part of Grid Network »

The Grid Professionals Group covers electric current from its transmission step down to each customer's home. 


Addressing the low-carbon million-gigawatt-hour energy storage challenge

Charles  Forsberg's picture
Principal Research Scientist MIT
  • Member since 2018
  • 8 items added with 944 views
  • Dec 13, 2021

Access Publication

Replacing fossil fuels is difficult because they serve two functions: (1) energy and (2) energy storage to enable energy to be provided to the customer when needed. Fossil fuels have very low storage costs; thus, it may be harder to replace the storage function than the energy function of fossil fuels. To meet the variable hourly to seasonal demand for energy today in the United States, we store a 90-day supply of oil, a 30-day supply of natural gas and a 100+ day supply of coal. We require several million gigawatt-hours of energy storage for our economy to function. That energy storage capacity was built based on real-world experience over a century, not models. Our recent studies  ( indicate that there are only four affordable options for energy storage at this scale: nuclear fuel, liquid biofuels, hydrogen and heat. Options such as batteries may be used for short-term storage but are unaffordable for longer-term storage because of capital costs.  

Charles  Forsberg's picture
Thank Charles for the Post!
Energy Central contributors share their experience and insights for the benefit of other Members (like you). Please show them your appreciation by leaving a comment, 'liking' this post, or following this Member.
More posts from this member
Spell checking: Press the CTRL or COMMAND key then click on the underlined misspelled word.
Jim Stack's picture
Jim Stack on Dec 15, 2021

  A simple statement but storage is only part of the need. QUOTE=Fossil fuels have very low storage costs; thus, it may be harder to replace the storage function than the energy function of fossil fuels. 

   The 4 big problems with fossil fuel and Nuclear energy is 1- They can't ramp up and down in time to meet the changing hourly not to mention daily needs of the users. So a Nuclear facility has to run at full power so it can meet the Peak Time of Day maximum need. But it can't ramp down as the night loads may only be half the Peak needs. What happens to all that extra power at the wrong time that can't be used? In our area my friends at the power plant told they they short it to ground. It gets wasted. This is the same for COAL and Older NG plants. Newer NG plant can ramp up and down a little faster to try and match changing loads but that makes they more inefficient and cost more to operate.    The 2nd problem with the east storage of energy at fossil fuel and Nuclear plants is the waste after use like coal slurry and spent nuclear material. 3rd we have the vast amount of water to boil to spin the turbines to make the electricity . Plus the added water to cool the Nuclear plants that can become critical.  4th the finite amount of fuel in COAL and URANIUM and NG. We hit Peak NG in 2014 and most now come from Fracking.  None of these 4 issues is sustainable. 

Charles  Forsberg's picture
Charles Forsberg on Dec 16, 2021

Nuclear power plants in countries such as France have been ramping up and down to match daily to weekly electricity requirements for over 40 years. In the U.S. they do not usually ramp up and down because they have very low incremental operating cost. Economics determines operating mode. Advanced reactors such as the TerraPower/GE Natrium reactor will include molten salt heat storage to enable base-load reactor operation with peak power output much larger than the baseload reactor power level as discussed in the paper and associated references. This is the same heat storage system used in Concentrated Solar Power plants. Dependng upon future electricity markets, heat storage may be backfitted onto existing nuclear plants. Large-scale wind and solar is only viable if have a low-cost way to provide electricity at times of low wind and solar output on an hourly to seasonal basis. That reality will probably make nuclear with heat storage the required low-carbon technology for large-scale use of wind and solar. Today the economic viability of wind and solar is dependent upon gas turbines to provide affordable electricity at times of low wind or solar output. 


Uranium resources is not an issue for thousands to millions of years depending upon assumptions. That is because the energy from a pound of uranium is about a million times that from a pound of coal. One can pay much more per pound of uranium. Today uranium is about 5% of the cost of nuclear energy. 


Peter Farley's picture
Peter Farley on Jan 9, 2022

France does modulate the power from its nuclear plants but nowhere near enough to cover daily variation in demand. Typical daily variation in nuclear output is 4-5 GW, while demand variation is 15-25 GW. That is why gas varies from 2-8 GW and hydro from <2 to 14 GW+ and nuclear capacity factors in France are 72% vs 90% in the US. At maximum demand last week it was importing 10.3 GW out of 80 GW demand


The simplified analysis below ignores transmission constraints, energy efficiency, electrification, demand response. cooling water availability and siting restrictions all of which affect real investment decisions, but it shows that for the same investment a wind and solar dominant system needs less storage than a nuclear dominated system. 


Monthly US electricity consumption varies from 290 TWh to 410 TWh. Hourly demand demand varies from about 330 GW to 700 GW and averages 470 GW so allowing for 15% reserve margin and 95% availability of dispatchable plant, the US grid needs about 870 GW of capacity, assuming perfect transmission availability. At the moment it actually has 950GW of dispatchable capacity. Source EIA

In a perfect nuclear/hydro/storage grid, that would mean about 780 GW of nuclear/storage capacity. Nuclear in the US only supplies about 20% of annual demand and a peak of 96 GW less than 1/3rd of minimum, so once nuclear exceeds about 85% of minimum demand it will be curtailed. Thus capacity factor will fall from the current 90-92% to a more France like 72% so to supply say 80% of US demand from nuclear would require nominal annual capacity of (4,100 x 0.80/0.72) = 4,600 TWh. At 95% availability that would mean 550 GW of nameplate capacity with peak coincident output of about 520 GW. To meet peak demand with safe reserve margin combined with peak hydro there would need to be 870-70-520 =  280 GW of backup. 

Now in the peak month if we push 520 GW of nuclear up to 90% and hydro up to 50% they will supply 365 TWh with remaining renewables supplying about 50 TWh, so demand would be balanced with supply so no new storage longer than a month would be required. However over the peak three day period in August where demand averaged 580 GW, backup would have to provide about 5,200 GWh.

Even allowing for lower capacity factor, wind and solar are about 40% of the installed cost per annual MWh of nuclear, so instead of installing a notional 4,600 TWh/y of nuclear, for the same investment you can install annual capacity for 11,500 TWh of wind and solar. (roughly 5-6,000 GW).  In the worst 3 day period that would still provide an average of 650 GW so a wind solar grid built for the same investment needs less energy storage over three days. 


But let us be very clear, on the worst day wind/solar/hydro would still require backup for some hours because output can drop to about 7% of nameplate across the country for short periods. So when demand hits 700 GW renewable output + hydro may well be as low as 400-450 GW, so backup capacity will still need to be 280-300 GW, but for no more than 2 hours and then declining, so in terms of GWh it will be at most 1,200 GWh, significantly less than the backup investment for the putative nuclear dominant system.


In practice whichever near zero carbon route is chosen, there will be nuclear, renewables, storage and a lot more controlled demand because that will be cheaper than just providing enough generation and transmission to meet any conceivable peak demand. The question, which is largely a financial decision, which will be made by thousands of customers and utilities not by the federal government or Universities is "which is cheaper for a given level of reliability and emissions".  At the present time if SMRs can supply power at half the cost of Plant Vogtle or Hinckley Point while being able to flex up and down from 20-100% of power and be profitable at 60% capacity factor then they will be part of the mix.




Julian Jackson's picture
Julian Jackson on Dec 17, 2021

As far as nuclear power goes, it looks like modular Thorium reactors will be cheaper to build and run, and produce far less waste. The USA ran an experimental one in the 1960s for four years. I'm sure we could do better these days with more advanced technology. Seems like quite a few different designs are being worked on now.

Peter Farley's picture
Peter Farley on Jan 9, 2022

Apologies for the long post but this paper is so seriously misguided that it needs detailed rebuttal. 

The fundamental question is why do you need so much storage. 90 days of oil storage is in case of interruption of supply due to war or economic embargo, unlikely with wind and solar. Then much of the coal and gas storage is required because of seasonal, transport or processing constraints, most of which are either irrelevant or completely different to those in a a predominantly renewable energy system. 

The whole premise of this report is fundamentally flawed.  For example the claim that battery costs will flatline at $500/kWh when battery manufacturers are already claiming $300.  The estimate of current storage installations is way out. According to the DOE 14.5 GW of grid scale storage is scheduled to come on line from 2021 to 24. That is seven times the rate claimed in the paper.

The claim that most energy will go through storage is incorrect. Most power is used during the day so over a year at least 50% of energy use can be supplied direct from East West or tracking solar. Further many activities that are currently scheduled for night time such as water heating and municipal water transfer can just as easily and more cheaply done during the day if you build enough solar, that means that on most days hydro will be preserved for evening and morning peaks and wind and existing nuclear will supply most of the nighttime load, leaving new storage with 0-20% of the daily load.

The analysis appears to have significantly underestimated existing hydro storage because the output of conventional hydro-electric dams can be modulated to suit demand so some of the storage in hydro dams has to be added to pumped hydro. Further, according to the EIA there is about 1,200 MW of untapped hydro potential on existing dams while many existing hydro installations can be reconfigured for increased power at the expense of shorter duration.

There is 79 GW of conventional hydro in the US that runs at an average 40% capacity so hydro output can be expected to vary between 16 and 70 GW. A 50 GW spread for 24 hours is 1,200 GWh added to the existing 550 GWh so we reach 1,700 GWh of available storage already. By reconfiguring existing installations and powering unpowered dams that can easily be increased to 1,800-2,000 GWh with very little new pumped hydro.

The US currently uses 10,900 GWh per day of which nuclear can be expected to supply 2,100 and hydro/pumped hydro on a good day about 2,000 and dispatchable renewables like biomass, waste to energy, geothermal etc. about 200 GWh for a total of 4,300 GWh of dispatchable low emissions generation.

Lets assume that the US  eventually installs 500 GW of onshore wind 80 GW offshore, 500 GW of utility solar and 700 GW of rooftop solar. That should supply an average of 11,700 GWh per day but on windy sunny days some of it will be curtailed, so over a year probably only about 10,000 GWh/d.  Now an examination of the Australian NEM, which is much less geographically diverse than the US grid, shows that the worst wind and solar day is at least 60% of the average day, but let us be really cautious and say that it is only 50% of average output or 5,000 GWh in a high renewable US scenario. So very bad wind and solar day new backup is 11,000 GWh - 4,300 - 5,000 = 2,700 GWh. 

Now the current US coal and gas generation fleet has never exceeded a combined annual capacity factor of about 65% so instead of installing 1,800 GW of wind and solar perhaps it could install 2,300 GW, that would mean more curtailment but is that a problem. Currently coal plants average 50% capacity, CC gas plants about 55% and OC gas about 20% i.e. they are frequently curtailed. For example  According to the EIA the US has about 530 GW of gas capacity. In the last 30 days, peak gas output was 223 GW and minimum 84 GW. One can say that curtailment of gas was between 46% and 84% so if an average 30% of wind and solar was curtailed over a year what is the problem. So by increasing generation capacity by 30% and minimum day output to 6,200 GWh and therefore reducing storage, the extra generation capacity may well be justified.

Now while wind and solar can be very low for a day and low for a week, over a week average output will be higher than the worst day. For example in Germany in the last year, lowest renewable day was 1/3rd of average but lowest week was 60% of average. So over a week storage requirements are only about 2.5-3.5 times the worst day not seven times.  So for the current electrical demand a zero carbon system with 1,800 GW of wind and solar can maintain supply with 6-9,000 GWh of new backup. With 2,300 GW of wind and solar additional storage requirements would be cut to 4-6,000 GWh This is almost two orders of magnitude less than the paper claims

Then there is a claim by many that there is a need for seasonal storage yet an examination of EIA data will show that monthly renewable output doesn't vary much and the lowest periods are in July. Given that the US solar output is only about 40% of the output from wind, that situation is easily remedied by increasing the share of solar vs wind installations as is now happening.  So within about seven to eight years the seasonality will have all but disappeared from generation data. 

Now obviously the aim is to be net zero but if we initially aim for 95% zero carbon then in the bad week out of 80 TWh total demand when wind and solar drop to 50 TWh while  nuclear and hydro, biomass etc. provide 20 TWh and gas has to provide 10-15 TWh or perhaps 150 TWh over the year is that not a dramatic result. 

Then there is the question of energy use. The US uses about 40% more energy per $ of GDP than Europe and in both the US and Europe absolute energy use is declining. So even if you say you need X weeks storage in 10-15 years time that is likely to be 10-20% less in absolute terms than today.

Much of the storage will be integrated into the system. Many homes and businesses will install smart controls on water heaters and EV chargers and then batteries as costs fall, This will be to save unused solar, slash peak demand tariffs, provide backup power or even just as a technological fashion statement. Similarly wind and solar farms will more and more be faced with negative prices or be constrained off.  So even now in high renewable regions, it is becoming financially viable to add storage rated at 15-20% of peak power for 3-4 hours to move production away from low prices to higher prices. Assuming that 80% of the 2,300 GW of wind and solar has storage attached that is another 1,200-1,500 GWh.

There are about 150m grid connected premises in the US with an average daily demand of 78 kWh. If 15% of those customers install thermal or battery storage equal to 1/2 a days demand that is another 900 GWh.

By 2035 there will probably be 150m EVs on the road in the US. Typically they will have 75 kWh of storage that on average will be 60% charged, smart charging coupled with weather forecasting will allow the average state of charge to vary between 45% and 75% on any given day so there will be a 3,300 GWh potential variation in vehicle storage.

So by 2035 there will be at least 15,000 GWh of storage across the system just by allowing normal market forces to operate. There is little evidence to show that that won't be enough.

Finally the cost per MWh of storage is far less than the paper's claim, because most of the storage is either already in the system, as hydro or EV batteries or much of it will be thermal storage which can be 1/10th the cost of batteries. The EIA figures for batteries are historical averages and they are still falling rapidly with iron, sodium and redox batteries having much lower costs per MWh than Lithium ion so rather than assymptoting out at $500/kWh as you suggest $100-200/kWh is more likely.  

In summary your estimate for the cost of storage to back up a rationally sized grid with 25-40% over capacity in generation as we have always had, is out by between 50 and 200 times the eventual cost.




Charles  Forsberg's picture
Charles Forsberg on Jan 11, 2022

I discuss energy for the total U.S. economy--not just the 18% of the energy consumed by the final customer in the form of electricity. In a single day, the U.S. total energy consumption is 78,000 GWH. If you get a cold wave over the central United States with a diameter over a 1000 miles, that energy consumption is much higher thanks to heating loads. If you get low-wind conditions that last for a week, wind is out. Europe has had several low-wind events of several weeks. At the same time of peak winter demand have minimum solar output.  


Can't bet on hydro. The problem is that every decade or so, have some dry years and that storage capacity disappears. The current bail out for dry years is fossil fuels. 


People did not build all that storage because they like energy storage. They built it because they learned the hard way that bad things happen if you do not have storage. That is why energy storage in the real world is measured in units of millions of gigawatt hours.  I live where the temperature is near 5F. Most of the energy in my house today and for the next few days is coming out of gas storage facilities.


A note on battery costs. The U.S. Energy Information Agency has been tracking battery facility costs that are leveling off near $500/kWh. That includes batteries, AC/DC converters, land, fire control, connections to power lines, taxes, insurance and all the other real-world costs, EIA reports complete project costs, not the hype of battery advocates that at most includes the battery cost on the factory floor. If you build a house, you will discover that lumber is a fraction of total costs. Same true of a battery costs in battery storage systems.  


Bottom line. Must build for bad days and there is no free lunch. 

Peter Farley's picture
Peter Farley on Jan 13, 2022

Nobody is disputing that you need storage or claiming that we can get something for nothing. It is a question of duration. Electricity is far more than 18% of final effective use of energy in the US, because for example 3/4 of the energy in a barrel of oil is lost before the wheels of a vehicle. The whole fossil fuel industry uses about 10% of US energy consumption, 80% of that will disappear. There will still be biofuels or synthetic methanol/ethanol/ammonia for air and sea transport and some backup power. There is far more than enough storage already for that.

Even in cold climates over a season, heatpump heating is twice as efficient as gas. So, with no change in comfort or utility, an electrified US economy will only use 2.5~3.5 times as much energy as current electricity demand. Further my estimate of storage included ground transport, so absolute worst case my calculation of storage requirement is under by a factor of four so let's say the US needs 60,000 GWh of which about a quarter will be in hydro/ pumped hydro and batteries in EVs, so the new storage is at most 45,000 GWh about 1.5% of your estimate.

Most of the storage in the system is for fossil fuels, 90 days is an arbitrary number dreamt after the Middle Eastern oil embargoes. It is obvious to everybody that there are periods of low wind and solar, but they are never zero, they don't last anywhere near 90 days and certainly they cannot be disrupted by a foreign country, Nuclear systems have about three years of storage because that is the length of the fuel cycle. The solar cycle is a day, and the wind cycle is about a week. On top of that you have seasonal cycles but the combined wind and solar variations are quite small. If the US generated as much energy from solar as wind, the seasonal variations of combined output would be minimal. You can test that statement by going through the EIA monthly data and comparing it to annual data. Wind is stronger in winter and at night solar the opposite, but because wind supplies more than double annual solar output renewable share falls in summer. 

If you laboriously go through the EIA data, combined wind and solar share varies by no more than 15% month to month. i.e. if you build enough wind and solar to supply 130% of typical monthly demand it will still cover the bad month. I agree even with a 130% annual supply capacity, there will be hours where existing hydro and nuclear cannot even provide half the energy required so yes there will be 300~500 GW of new dispatchable energy in net zero system. But it doesn't need to run at that rate for  a single day let alone 90 days.       


You also assume the US will continue its prodigious waste of energy where it uses 40% more energy per $ of GDP than most of the rest of the OECD outside the arctic. It has reduced specific energy consumption by about 15% since 2008.  And it is just getting started, electrified transport uses roughly 1/5th of the energy and electrified heating 1/2 that of a fossil fuel system. Building codes and energy efficiency standards are being tightened so the US waste of energy will decline.  


Re batteries I am not disputing the EIA figure, as I said they are based on the historical data. I am disputing your contention that they will asymptote out at $500. It may even be true that inverters, transformers, switchgear won't fall much in price, but if the batteries themselves fall, long duration storage will still fall in cost.  For example, if a two-hour battery today is $600 in infrastructure and $500 in batteries i.e. batteries $250/kWh for a total of $550/kWh. Then batteries themselves fall to $150/kWh (already achieved in automotive practice) and you put in an 8-hour battery the infrastructure might rise to $750 because switchgear, transformers and inverters remain the same and battery cost at 8 x $150 is $1,200 for a total of $1,950.  $1,950/8 is  < $250/kWh.


Your analysis is like saying in 1890 that we used about 1 lb of hay per person mile.  To account for bad seasons, we need about two years storage of hay.  In the US in 1990 there will be 300m people travelling 25,000 miles per year, so we need storage for 7.5 billion tons of hay.  








Charles  Forsberg's picture
Charles Forsberg on Jan 19, 2022

Based on real world experience, the paper concludes that a low-carbon energy system needs about 6 weeks of storage—versus currently 90-day supply of oil, 30-day supply of natural gas, 100+ day of coal and 9 to 12 months of nuclear fuel in reactors. What Mr Farley is proposing is that the U.S. energy system needs less than a day of storage.


Systems that claim to need little storage is needed are based on the belief nothing will go wrong if electrify the economy with a massive electrical grid to balance out demand with supply across the country. There will be no winter events, hurricanes, earthquakes or solar flares that take out part of the gird for weeks. The catch is that the electric gird is more sensitive to these events than pipelines or railroads. Pipelines are underground. Railroads have existed for 150 years where storms, floods and earthquakes have identified the weak points that have been reinforced. Super grids have additional failure modes such as solar storms and complexity that go up as the size increases.


Germany and California are learning that lesson as Germany promises to keep lignite plants open for decades and California buys more gas turbines—because those systems have cheap large-scale energy storage. Germany has the highest electric rates in Europe and is the bad guy of western Europe in terms of carbon dioxide emissions. California has among the highest electric rates in the United States—along with the highest poverty rate in the United States when you adjust for the cost of living that are partly driven by high energy costs.


There is also the reality of peak demand. When a cold front from Canada arrives in January, it arrives at the time of lowest solar output—short days with the sun to the south. It covers more than half the U.S. In really cold weather, wind goes to zero. In my youth I did winter camping in Minnesota—as the temperature drops you can hear a person walking in the woods a mile away because the usual background noise from the wind is gone. The Texas grid nearly collapsed because of the collapse of wind output as the temperature went down for several days. They cut off a lot of customers. At the same time, the instantaneous heating demand climbs to several time average energy demand. Air-based heat pumps go to resistance heating mode because they do not operate in very cold weather with electric heating demand going up because of the combination of colder weather and less efficient heating system. The power demand goes up from all-electric vehicles to provide heat for the passenger compartment. In short, massive increase in energy demand for multiple days while wind and solar decrease. Such normal events alone require many days of storage.


The consequences of a large energy system without large storage are catastrophic. In an all-electric system, if power failure the rescue workers have to deliver food to everyone in a couple of days because electric car batteries go dead so people can’t travel to food distribution. At the same time have to repair a system that does not allow for fast repairs if loose big lines (weeks) or transformers (months to years). If cold climates, people begin to freeze to death in hours—which is one reason why gas or propane-fired fireplaces that operate without electricity have become popular in cold climates. The gas system has a 30-day supply to run so one expects in the future any serious hydrogen system will have about the same storage capacity. Storage depends upon what else is in the energy mix. The question I have, where I do not have the definitive answer, is six weeks storage enough? 


I have advocated that major energy conferences should be held in January or early February at the University of Minnesota Duluth Campus that overlooks Lake Superior. The weather would be a reminder that when energy systems fail, a lot of people will die in a relatively short period of time if no energy storage. The conferences could have outdoor demonstrations of maintenance operations in cold weather--an extremely labor-intensive slow activity in cold weather. Same problem if earthquake, hurricane or solar flare (where we are overdue for a very bad day that takes out long-distance lines). 

Charles  Forsberg's picture
Charles Forsberg on Jan 19, 2022

Added note about national electrical grids--we do not have anything close to a national electricity grid. Power plants were originally built in the middle of cities to serve customers. Longer transmission lines were built to transfer electricity from hydroelectric dams and nuclear plants to nearby cities. Added transmission lines were built to improve reliability so if one plant failed, other plants could pick up the load. While there are nice wall charts showing the grid, very little electricity goes hundreds of miles from the power plant to the customer.


When wind and solar advocates talk about a national grid to address the storage challenge, they are proposing something that never has been built--a transmission system that moves massive amounts of electricity long distances reliably. There is a continuing debate in the electrical engineering community among grid experts whether a reliable national grid is technically feasible. The Chinese grid is broken into three parts because of some of these technical problems.  


What is clear from decades of experience is that it would take massive federal presence to force building such lines over the opposition of state governments, local governments and individuals. China is the only major country in recent years that has been able to build a significant number of long-distance power lines with significant electrical capacity. 

Get Published - Build a Following

The Energy Central Power Industry Network is based on one core idea - power industry professionals helping each other and advancing the industry by sharing and learning from each other.

If you have an experience or insight to share or have learned something from a conference or seminar, your peers and colleagues on Energy Central want to hear about it. It's also easy to share a link to an article you've liked or an industry resource that you think would be helpful.

                 Learn more about posting on Energy Central »