This group brings together the best thinkers on energy and climate. Join us for smart, insightful posts and conversations about where the energy industry is and where it is going.


Despite the Hype, Batteries Aren't the Cheapest Way to Store Energy on the Grid

Tesla Powerwall installed in garage

Tesla Powerwall installed in a garage.

There are many different kinds of energy storage technologies, each with its own advantages and drawbacks. Lithium ion batteries are the most popular form of storage at the moment, but according to Roger Dargaville, Deputy Director of the Melbourne Energy Institute, they are not always the cheapest option. Nevertheless, lithium ion will probably be the dominant option, not because of economics, but because of human behavior. Courtesy of The Conversation.

Storage is the word of the moment in the energy industry. Since Tesla unveiled its Powerwall, politicians, commentators and industry have hyped storage – and particularly batteries – as the solution for getting more renewable energy into electricity grids and reducing our reliance on fossil fuels.

The concept of storage is simple. A storage system takes power off the grid or from a local generation source and puts it back onto the grid or uses it locally later. It seems like a good idea if you have too much energy, or it is cheap at some times of the day and expensive at others.

So could storage be the answer, and how much would it cost?

The costs of storage

Of course storage isn’t free. It comes with both a capital cost (buying it in the first place) and a running cost, which is related to the cost of electricity to charge the battery and the round-trip efficiency – how much power is lost in the charging and discharging cycle.

To be a sensible economic investment, the benefits have to outweigh the costs. In other words, the savings on your energy bill have to be greater than the capital costs plus the running costs.

There are many different kinds of storage technologies, each with different characteristics. Lithium ion batteries are attractive as they operate effectively at small scales, are lightweight and have good round-trip efficiency. But they are currently expensive per unit of storage capacity.

Pumped hydro at the other end of the scale operates at very large scales, has good round-trip efficiency and is very cheap per unit.

Flywheels (or rotors) have low round-trip efficiency and don’t store a lot of power, but are able to dispatch lots of power in a short time and can also contribute to frequency stability.

Other storage technologies include compressed air, cryogenic (liquid air) energy storage, flow batteries and hydrogen. Each has its respective pluses and minuses.

Dargaville 1

Each of these technologies will have an appropriate place in the grid to be installed. Lithium ion batteries are a logical choice for a small-scale distributed application, while pumped hydro will work best at the large scale for grid management.

Flow batteries, liquid air and compressed air are in-between technologies in terms of scale, and flywheels and capacitors are most useful at the substation level for voltage and frequency control.

Batteries versus hydro

Let’s focus on lithium ion batteries and compare them to pumped hydro storage.

Lithium ion batteries are coming down in cost at a significant rate. Bloomberg has plotted the costs of lithium ion alongside solar PV. This shows the two technologies share a similar cost curve gradient, with lithium ion reducing from US$1,200 per kilowatt hour to US$600 per kWh in five years (not including installation costs).

Dargaville 2

As more batteries are built, the price gets cheaper. Bloomberg New Energy Finance

So where does lithium ion need to get to be cost-effective? Imagine a home with a 4.5kW rooftop PV system and variable electricity rate (for instance off-peak cost of 20c, shoulder of 26c and peak of 40c, similar to this tariff).

In such a home a 7kWh battery needs to cost less than A$7,000 fully installed to actually save the homeowner money. In other words, the cost per kWh of storage should be roughly A$1,000 to break even. Currently, batteries cost A$1,000-3,000 per kWh, so they are on the cusp of being cost-effective.

However, there is an important catch here. Retail electricity rates tend to exaggerate the true range in costs between peak and off-peak. The difference in the wholesale market (where retailers buy their electricity) is around 5-10c per kWh, much less than the 20c range in current variable rates. If retailers begin to lose market share, they may respond by reducing or removing these variable rates. That would make peak rates cheaper and mean that batteries would need to be correspondingly cheaper to be cost-effective.

For instance, a flat electricity rate of 25c per kWh means that batteries would need to cost around A$300 per kWh to be cost-effective. That’s less than a third of their current costs.

You could argue that using batteries also reduces the cost of the network itself. By reducing loads at peak time, we can reduce or even remove the need for infrastructure upgrades (substations and additional power lines, for instance).

But this is only true if electricity demand is growing. If demand is flat or falling, then distribution networks will tend to be under-used. Therefore reducing peak demand will not result in any savings.

Overall demand in the National Electricity Market has declined significantly since 2009, so the benefits of storage on the grid will be negligible other than in high-growth corridors. Demand has rebounded in 2015-16 and it will be interesting to watch and see if this is a resumption of the steady increase or if the demand stays low.

Dargaville 3

Demand in Australia’s National Electricity Market has been falling.

Pumped hydro, on the other hand, is a relatively inexpensive storage technology (already at around A$100 per kWh) as it can store large amounts of energy using a very inexpensive material.

All you need is some water and the means to pump it uphill. So while it can’t be used everywhere, there are many places in the National Electricity Market where it is possible. There are already 1,500 megawatts of pumped hydro in the market (Shoalhaven, Wivenhoe and Tumut 3).

This would be a more logical solution – cheaper and easier to control by the market operator. But in the same way that rooftop PV has gained more popularity than large-scale solar (even though the latter should be cheaper), distributed storage in the form of lithium ion batteries may be the eventual winner, not because of economics but because of human behaviour.


This article was first published on The Conversation and is republished here with permission.

Dr Roger Dargaville (@rogerd70) is the Deputy Director of the Melbourne Energy Institute. He is an expert in energy systems and climate change, specializing in large-scale energy system transition optimisation, and novel energy storage technologies such as seawater pumped hydro and liquid air energy storage. He has conducted research in global carbon cycle science, simulating the emissions of carbon dioxide from fossil fuel and exchanges between the atmosphere, land and oceans as well as stratospheric ozone depletion. He leads a research group of PhD and Masters students working on a diverse range of energy related topics including disruptive business models, EROI, transmission systems, bioenergy, wave energy and high penetration rooftop photovoltaics systems. He coordinates the subjects Renewable Energy and Climate Modelling as part of the University of Melbourne’s Master of Energy Systems degree.

Original Post

Energy Post's picture

Thank Energy 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.


Bob Meinetz's picture
Bob Meinetz on Jan 12, 2017 4:49 pm GMT

Roger, your article begins with the implicit assumption storing grid energy is necessary – at all. Somehow, we’ve gotten by for a century by generating electricity without any storage. To generate clean electricity in 2017, we have two options:

1) Use nuclear fission, or

2) Embark on a costly, wasteful charade of storing intermittent energy from the sun and wind, filling in the gaps by burning fossil fuel, or cutting down trees and burning them.

The option that’s cleaner is also the one which has been proven to work.

Rex Berglund's picture
Rex Berglund on Jan 12, 2017 9:43 pm GMT

There’s a case for dispatchable renewables, and besides, the cost is coming down:

“The Kauai Island Electric Cooperative and AES Corp. announced plans to pair a 28 MW solar array with a 20 MW, 100 MWh battery system to deliver dispatchable renewable generation to the Hawaiian island.

KIUC will pay $0.11/kWh for power delivered from the solar-plus-storage system, according to a release, below the cost of oil-fired power that comprises the island’s current baseload generation.

The project is the second flexible solar facility for the small co-op. In 2015, KIUC signed a deal with SolarCity to pair a 13 MW solar array with a 52 MWh battery that will deliver power for $0.145/kWh.”

Paul O's picture
Paul O on Jan 12, 2017 10:37 pm GMT

Even if the costs are coming down, I still can’t see a cause for excitement in the Hawaii dispatchable solar that you have mentioned. So since power engineering is not my forte I wish to put out some questions.

On the face of it, a 20 MW array will require 5hrs to fully charge a 100MWH battery, the battery being meant to deliver dispatchable power. If Hawaii has 12hrs of daylight and allowing for maximum solar activity all those 12 hrs, and assuming we can’t be using the power for other purposes while the battery is being charged, this would imply that the solar array can provide 7hrs of useful power in the possible 12 hrs of daylight, before resorting to the stored 5hrs of power. In the end, 7+5 is still 12hrs, battery not withstanding.

Isn’t there some redundancy here? Why have the battery at all if the number of hours you can get power is the same with or without the battery?

Am I being too uninformed? Please elucidate me?

Darius Bentvels's picture
Darius Bentvels on Jan 13, 2017 6:21 am GMT

During day-light the solar installation loads the battery + supplies inhabitants.
In the evening and night no solar, so then the battery supplies the inhabitants.
Assuming night consumption is very low, the installation can supply inhabitants 24hrs.

Darius Bentvels's picture
Darius Bentvels on Jan 13, 2017 6:53 am GMT

Nuclear is far more expensive and hence also emit far more CO2, as the costs of economic activities and CO2 emissions are closely related.*)

The emissions of the full fuel cycle (mining, enrichment, fuel rod production, guarded waste storage during thousands or yrs) as well as decommissioning is often forgotten.

*) Except when fossil fuel is used to generate electricity. As fossil contains carbon which deliver high extra CO2 emissions when burned.

Paul O's picture
Paul O on Jan 13, 2017 7:00 am GMT

Thanks for the comment, but I am hard pressed to imagine how a 20 MW array can charge a 100MWH battery, and still deliver useable power to the populace, given that the sun is not at its peak in early morning and at dusk and that the entire power output of the array will have to be given over to charging the 100MWH battery if the battery is to become fully charged in 5 hours.

Or this is how I expect it to be.

Roger Arnold's picture
Roger Arnold on Jan 13, 2017 7:43 am GMT

Certainly batteries can’t allow any solar array to deliver more energy in one day than it can deliver in one day. That seems to be what you want it to do. You’re of course right that a 20 MW array can’t charge a 100 MWh battery bank in less than 5 hours, but so what?

The array splits its output, delivering a part to the grid and the rest to the battery bank. Then when the array is not producing, the battery bank has stored energy to meet the demand. That demand must be well under 20 MW. In fact it must average only 8 MW over the course of the 12 hours or more when the solar array won’t be producing. If we assume that demand also averages 8 MW during the time the array is producing, then there would be 12 MW available for charging the battery bank for the times of non-production.

Those numbers are unrealistic, of course. Demand varies over the course of a day, and the array will only produce it’s full rated 20 MW for an hour or so around noon. But the point is the same: the purpose of batteries — or any other type of energy storage — isn’t to magically increase the amount of energy that the system can produce; it’s to allow the energy that it does produce to be stored for later use as needed.

That’s pretty basic, and if you don’t get that, I don’t know what to tell you.

Roger Arnold's picture
Roger Arnold on Jan 13, 2017 8:25 am GMT

The figure in the article omits several of what I consider the most viable grid-scale energy storage technologies. Two are variants of pumped hydro that involve deep excavation: pumped hydro between a surface reservoir and a deep underground reservoir, and gravity power on the model espoused by Gravity Power LLC.

Another promising candidate is based on thermal energy storage as proposed by Isentropic Energy in the UK. There’s also quasi-isothermal CAES, as pursued by Lightsail Energy. Oh, and there’s CO2 plume geothermal, as advanced by TerraCOH.

There are a couple of other good candidate technologies as well, but they don’t have convenient web sites I can point to that explain them.

Unfortunately, in the absence of a meaningful price on carbon emissions, no energy storage technology can compete with cheap natural gas burned in simple combustion turbines. The latter are not very efficient. They emit a lot of CO2, but they’re cheap enough that intermittent operation at an average capacity factor of 20% or so that high penetration of renewables dictate isn’t too much of a financial burden on the electrical system.

The burden to life on the planet? Well, that’s another matter.

Darius Bentvels's picture
Darius Bentvels on Jan 13, 2017 9:07 am GMT

From Rex’s comment:
The solar array is 28MW, not 20MW.
The battery (un)loading speed is 20MW.
This is in line with normal design rules.

Darius Bentvels's picture
Darius Bentvels on Jan 13, 2017 9:16 am GMT

In fact the demand must be 8MW during ~6hrs, as:
– battery capacity will reduce gradually; and
– fully unloading such battery shortens its life much faster.

So a decent designer over-sizes the battery capacity substantially; often doubles it in order to have no problem for the next 10years or so.

Roger Arnold's picture
Roger Arnold on Jan 13, 2017 9:27 am GMT

Nuclear is far more expensive and hence also emit far more CO2

(Sigh!) I consider myself a patient and tolerant person, but even I get really tired of seeing you parroting that absurdity over and over.

In ERoEI analysis, the overall average ratio between GDP and energy consumption is sometimes used to attribute energy inputs that are too difficult to estimate accurately. But any analyst with any sense recognizes that it’s at best a crude assumption that should never be used for anything beyond minor miscellaneous inputs. Otherwise, it muddies the whole analysis to the point that it’s useless. Tell me, did that Van Gogh painting that sold a while back for $50 million (or whatever) really emit tens of thousands of tons of CO2 in its production?

A correction: I said that the analysis would be rendered useless, but I suppose there is one use: to render a covering gloss of scientific appearance to a blatent piece of propaganda. Your heroes Storm van Leeuwen & Smith were force to resort to the supposed cost-carbon equivalence in order to find some basis for asserting that nuclear power had a higher carbon footprint / lower ERoEI than wind and solar. Nobody but you and your anti-nuclear co-religionists take that study seriously, but like the Energizer Bunny, you just keep going.

Hops Gegangen's picture
Hops Gegangen on Jan 13, 2017 10:13 am GMT

We keep having this argument about nuclear versus renewables. I see them competing on their various merits. There will be an equilibrium somewhere of nuclear, solar, wind, storage, and methane fill in. All evolving over time and in different mixes in different places.

Torrey Beek's picture
Torrey Beek on Jan 13, 2017 2:54 pm GMT

So a decent designer over-sizes the battery capacity substantially; often doubles it in order to have no problem for the next 10years or so.

…not any designer that is trying to reduce project costs.

Bob Meinetz's picture
Bob Meinetz on Jan 13, 2017 3:49 pm GMT

Roger, TEC is currently reviewing policy with respect to Ideological Bunnies. I would encourage you to contribute your input, and help prevent billionaire environmentalists from burning their Van Goghs:

Bob Meinetz's picture
Bob Meinetz on Jan 13, 2017 5:15 pm GMT

Hops, inclusion of renewables in a grid mix very much works to prevent nuclear alongside it.

At this moment PG&E, in filings in support of its decision to abandon Diablo Canyon (nuclear) Power Plant, is diligently attempting to dismiss the plant’s ability to make anything but a “baseload” contribution to California’s grid mix. Though DCPP is perfectly able to load-follow grid demand, it’s important to make a distinction between “load-following” and “supply-balancing”: intermittent renewable sources occasionally introduce supply transients in excess of ±5GW/hr which nuclear, due to a phenomenon known as xenon poisoning, is incapable of effectively balancing. So natural gas turbines step in to balance renewables, and prevent wires from melting from poles.

What’s the answer? There are two which make sense:

1) Allow DCPP to remain open and help meet California’s baseload consumption, which only rarely dips below 15GW in the dark of night. Permit CCGT natural gas turbines to handle load-following and balancing of whatever renewables are available. Or:

2) Send all solar panels, wind turbines, and natural gas turbines to recycling pits. Keep DCPP, build more nuclear to meet demand, and enjoy the benefits of a 100% carbon-free grid.

PG&E is pursuing neither rational option. Instead, the company is advocating for the abandonment of DCPP at ratepayer expense, with a misty-eyed promise of renewables, efficiency, and storage, sometime, somewhere, to take its place. As Californians saw only six years ago with San Onofre (and Germany with the current collapse of its Energiewende) that doesn’t work. It will be replaced by burning natural gas.

That option is also obscenely profitable for both PG&E and Royal Dutch Shell, both of which are parties to the action, so both are performing due diligence on behalf of their shareholders and against public interest. With Trump and his Fossil Fuel Mafia descending on the nation’s capitol next week, few question which side will prevail.

Hops Gegangen's picture
Hops Gegangen on Jan 13, 2017 7:39 pm GMT

Nuscale claims their SMR can do demand following and co-exist with renewables. They are planning for that. I assume they are not kidding.

Engineer- Poet's picture
Engineer- Poet on Jan 13, 2017 7:47 pm GMT

Actually, the array appears to be 28 MW (nameplate, somewhat less due to temperature) so the theoretical charging time for the battery is closer to 3.5 hours (probably 4+ with losses), not 5.

I do not see how this system can provide power so cheaply.  Figuring battery packs at $450/kWh ($45 million) and 20 MW of inverters at $180/kW ($3.6 million) those two components cost $48.6 million.  Amortizing over 10 years at 7% costs about $6.8 million/year.  Dividing this by maybe 5.5 MW average delivered, just the batteries and inverters cost $0.14/kWh.  The PV itself, maintenance and the essential backup generators will cost money too.

Well, maybe with 60% subsidies….

Roger Arnold's picture
Roger Arnold on Jan 13, 2017 9:11 pm GMT

Bob, it’s true that there’s an issue with mixing renewables and 2nd generation nuclear on the grid. But it’s more complex than just what can or can’t be done — economics are involved — and there are more and better options for resolution than the two you mention.

The presence of relatively inflexible baseline generation on the grid “cuts the bottom” off the demand curve of what must be supplied by other resources. It increases the relative penetration by renewables of the “virtual grid” that serves what’s left over after baseload is subtracted. It’s that higher relative penetration by intermittent renewables that’s the root of the problem. We’re agreed on that much, I presume?

The problem is compounded by a regulatory environment that makes it more profitable for PG&E to address the variability problem by shutting down a facility that has zero carbon emissions and very low marginal cost of electricity produced, in order to be able to utilize its fossil-fueled assets at a higher capacity factor. It’s a deliberate retreat from the nominal goal of reducing carbon emissions, motivated by the need to cut the losses that the system is forced to swallow when revenues that would otherwise go to those fossil-fueled assets go instead to renewables. The fossil-fueled assets that are hurting can’t be shut down, because they’re still needed at times to meet demand when wind and solar are under-producing. But in a pro-renewable regulatory environment, nobody is willing to acknowledge the operation of fossil-fueled assets at low capacity factors as part of the cost of renewables.

We probably agree about that as well.

Where we appear to differ is that the only resolution you see is to at least rein in the deployment of renewables. What I see is a need for a combination of near term measures that include deployment of more grid-scale energy storage, use of opportunities for load-based regulation, upgrading a portion of existing fossil-fueled generation capacity with efficient flexible generation capacity, and rate reform that distributes costs more honestly and transparently.

If the right compensatory measures are adopted, I don’t see clean baseload capacity as incompatible with renewables. With the right compensatory measures, the two even become complementary. Those measures do involve significant capital expenditures, and someone will have to foot the bill.

michael pettengill's picture
michael pettengill on Jan 14, 2017 9:02 am GMT

The electric grid has always depended on storage. All hydro is storage. Normally rain does the “pumping” but dams store the water in ponds until needed and then the sluice is opened to draw off water to generate power.

Whether woolen mills or hydro electric, peak power would drop the water level. Then the peak passes and flow hopefully restores the water level.

Bob Meinetz's picture
Bob Meinetz on Jan 14, 2017 9:12 am GMT

Hops, in October I had an engaging conversation over lunch with Lenka Kollar, NuScale’s Director of Business Strategy. Though at the time their testing relied on sophisticated modeling only, she was confident their 30MW mini-reactor would be able to scale up and down to supply-balance any challenge thrown at it by renewables.

I recall thinking it would be an excellent small solution to small problems, in developing countries which needed failsafe power generation to work in tandem with existing solar arrays. But a feat of synchronization among several linked reactors would be required to balance a 5GW/hr drop from a larger solar array, like Topaz, when a cloud front moves in.

And the train of logic always seemed to lead back to the same point: if we’re accepting nuclear and can afford it, why do we need renewables, with their intermittency and land-use issues, at all?

Helmut Frik's picture
Helmut Frik on Jan 14, 2017 10:46 am GMT

Well, because Land use is not large of renewables, especially when einther deserts, unareable land or roofs are used for solar, which is usual practice today.
And space occupied (restriting use for other purposes of wind is not higher than land use by nuclear.
So its the question is : is nuclear with all insurance and waste storage costs cheaper than tenerwables+ grid expansions? if you look e.g. at the prices with which dominion calculates it does not look like this. As well nuclear would come too late with too little capacity, due to long construction times and mostly inexistent supply chain.

Helmut Frik's picture
Helmut Frik on Jan 14, 2017 10:52 am GMT

Imm last time I talked to inverter manufacurers in MW scale they were well below 5ct/W. so 20 MW would be 1 Million or less. Your battery pack price might also be way of the price they pay, Amortisation, 7% costs are also guesses, so seems they calculate in a different way and have different costs than your assumptins. Since it’s their second projects the first guess would be that the other project works so far.

Paul O's picture
Paul O on Jan 14, 2017 11:43 am GMT

I was of the impression that the array was supposed to supplant Oil or other co2 emitting sources, and would be providing or expected to provide 100% of the power requirement of the grid.

It just didn’t seem like that could be done whilst simultaneously charging the battery.

Helmut Frik's picture
Helmut Frik on Jan 14, 2017 11:46 am GMT

Well wen Solar and wind power have been expensive, nuclear fans used exactly this argument to claim high CO2 emissions for renewables. Now it “works” the other way round – I did not accept it more early, and I do not use it now. Still nuclear is uneconomic.

Rex Berglund's picture
Rex Berglund on Jan 14, 2017 2:53 pm GMT

Lazard’s LCOE V10 states PV+storage is $92/MWh. Note d on pg 3 gives a plethora of assumptions, notably 20 year life, not 10.

Roger Arnold's picture
Roger Arnold on Jan 14, 2017 7:48 pm GMT

Rex, thanks for the link to that Lazard report. It appears to be a high quality reference — at least in terms of scrupulously documenting their assumptions and methodologies.

Unfortunately, there’s not really much information or analysis about storage options. Just the one large footnote for PV + storage where they state their assumptions about battery storage. Also nothing in depth about the costs of intermittency — though at least they acknowledge that there are costs.

Roger Arnold's picture
Roger Arnold on Jan 14, 2017 8:04 pm GMT

Yes, MW scale inverters are different beasts than the inverters for PV and battery backup systems. The difference that makes them much cheaper (per kWh) is that they’re converting HVDC to balanced 3-phase HVAC in long distance power transmission systems. They sit upstream of multiple levels of step-down transformers that will filter out harmonics. Not clear whether that class of inverter could be used for PV backup in Hawaii.

Rex Berglund's picture
Rex Berglund on Jan 17, 2017 7:54 pm GMT

Roger, Lazard also released their Levelized Cost of Storage V2.0, with good news for cost trends.

Given solar at 4.6¢/kWh and storage at 23¢/kWh, 75% of supply from the panels and 25% from storage yields 9.2¢/kWh. Some of the tech. and use cases in the LCOS 2.0 have prices in that range, though even the cheapest lithium ion is still a bit more expensive.

The studies I’ve seen say that long distance HVDC is less expensive than storage at the moment. Still, there are those who say the learning curve will soon drive down costs far enough for renewables + storage to be cheapest, we’ll see.

I noticed in another comment you mention the Lightsail tech., and I agree that does seem promising. Even better though would be the take on a Lithium-Oxygen battery in which a team from MIT, Peking University, and Argonne National Laboratory created a new variation of Lithium-air batteries which could be used in a conventional, fully sealed battery, promising similar theoretical performance as lithium-air batteries, while overcoming most of their drawbacks. It remains to be seen if this will make it out of the lab but if so that’s quite a game changer.

BTW, on a separate note thanks for your post on Cruising to Vegas, that is creative imagination at its best.

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 »