The Future of Energy: Will 'Cheap as Dirt' Batteries Transform the Grid?

Jesse

De-centralization is in many respects a fashion. A century ago electricity grids were highly de-centralized. If I take Britain as an example (I have the data lying on my computer). In 1924 Britain had 491 power stations, but had only 3.7 GW of total capacity. By 1990 Britain had 75 GW of capacity, 20 times more than in 1924. Yet, it had 163 power stations. This is an incredible increase in centralisation.

The long term trend for conventional power plants has been greater and greater centralisation. Why would renewables be any different? Offshore wind, so far, is following the centralisation trend. We are now talking about GW scale plants. Onshore wind seems to be following this trend in America. In more densely populated European countries it is probably more socially challenging to centralise onshore wind. Finding suitable space for wind farms bigger than 100 MW in Britain or Germany is much more difficult than in the US. But where there is space, such as the central plains of America or Western China (though not its eastern provinces) there are now huge numbers of plants greater than 500 MW in size. In fact almost all of the world’s >200 MW wind farms are in the US or in China’s western provinces.

We are also seeing some trends towards big solar plants in the US. And cheap batteries aren’t likely to change this. The economics of grid disconnection are not likely to move beyond the dubious. You just have to ask yourself how you get through Christmas with a solar panel and battery, and no grid access. That option only seems to exist in the minds of some DG advocates. The other issue is why cheap batteries would favour de-centralisation. I find it hard to see how one big centralised battery cannot be far more efficient than a bunch of de-centralised things in people’s basements.

The other, more fundamental reason, why DG won’t take over is power density. Factories, office blocks and multi-story residential buildings all consume energy at a rate of more than 100 watts per square metre. Rooftop solar simply cannot supply the energy needs of modern cities.

Centralised facilities will not be going away.

Batteries will never, ever be used for load-balancing the grid. Why not? Because batteries are already as cheap as dirt. And that’s still too expensive!

It’s hard to imagine a battery being much cheaper than a conventional lead-acid battery of today. Lead is a very common material and it’s already dirt-cheap. Ditto for sulfuric acid, the other major component. But let’s suppose that we’re missing something, and that some bright guy from MIT figures out how to make a battery from stuff that’s even cheaper than that. Half the price, let’s say. And let’s further assume that the bright guy figures out how this very cheap battery will last twice as many cycles as a lead-acid battery does. Will that be cheap enough?

No it won’t. A lead-acid battery today costs $150/kWh, and lasts for about 500 full cycles. Our hypthothetical dirt battery would therefore cost $75/kWh and last for 1000 full cycles. Lead-acid has a round-trip efficiency of 80%, and let’s assume that our dirt battery is the same.

The capital cost is $75/kWh, so each full cycle (of the 1000) costs 7.5¢/kWh. We lose 20% of the energy on each round trip, so add another 20% to the cost. That means electricity from the battery will increase the cost of electricity by 9¢/kWh when we use it, over and above the cost of generating the electricity that went into storage in the first place.

It gets pretty hard to imagine a scenario where pulling electricity from a battery, at x¢ + 9¢, will be cheaper than generating that same electricity on demand for x¢. Especially when you consider that for most technologies, x is somewhere between 4 and 7, more or less: so even when you’ve got this hypothetical dirt-cheap battery, it would double or triple the cost.

And then there’s the EROI consideration. Batteries will add to the sunk energy of any generating technology without increasing the amount of energy generated. Thus any scheme with batteries will decrease the EROI of that generating path. Renewable generation right now is pretty close to the edge for economic viability on EROI, and Weißbach et al. (2013) has shown that when you add storage for buffering (any storage, even ultra-cheap pumped hydro), the EROI of the renewble system drops below the point of being economically viable. Which means you’re not going to be using batteries for renewables. And for dispatchables, you don’t need them in the first place.

Weißbach, D., Ruprecht, G., Huke, A., Czerski, K., Gottlieb, S., & Hussein, A. (2013). Energy intensities, EROIs (energy returned on invested), and energy payback times of electricity generating power plants. Energy, 52, 210-221.

Jesse, I agree with Keith below who makes many good points, to which I’ll add a few of my own.

Batteries would not even be part of the discussion were they not being pushed as a solution for the intermittency of renewables; there are many good reasons for that, even given the generous assumption they’re cheap as dirt. With connections and management, they’re comparatively inefficient; we’re not close to having enough renewables to charge them; even when we do, we’ll still need a dispatchable infrastructure (running out of electricity is not an option); most people have no idea how many of them we’d need to power a grid (a lot); they don’t last forever, and even if they’re cheap as dirt the maintenance to replace them is not.

Possibly we’re doomed to repeat the mistakes of the past, but the grid we enjoy today is the result of a century of not making things deliberately more complex and sophisticated, but as simplified and centralized as possible. Many doubters who try generating their own electricity soon understand why sharing that responsibility can be a good thing.

Hi Jesse,

Thanks for your reply. Okay, let’s use the published results from Sadoway’s Nature paper, here:

Wang, K., Jiang, K., Chung, B., Ouchi, T., Burke, P. J., Boysen, D. A., … & Sadoway, D. R. (2014). Lithium-antimony-lead liquid metal battery for grid-level energy storage. Nature. doi:10.1038/nature13700

The paper indicates 10 years of daily charge-discharge cycling degrades performance to about 85%. (This is about 20 times worse than the .0002%/cycle figure you cite; decimal point error?) which is 3650 cycles. The paper also indicates, based on the current prices of lead, antimony, and lithium, a cost of $375/kWh.

But let’s assume 20 years of cycling instead, for 7300 cycles. At $375/kWh capex and 7300 cycles, that’s 5 cents per kWh-cycle. The paper also indicates 73% roundtrip efficiency, so mark that up to 7 cents — and we’re still doubling the cost of electricity. Further, since this battery operates at 450 C, we would have to generate a lot of heat to use it. It’s completely unclear how that heat could be generated, especially in a fossil-free world, or what the cost of that heat would be. 

Storage makes sense only when it’s cheaper than generation on-demand (almost never), or where on-demand is impossible (rarely). So there has to be some economic advantage involved somewhere. But what are we getting, even from Sadoway’s battery, that justifies doubling the cost? Where’s the economic gain from that?

The growth in the battery market for grid applications will surely be coupled to the growth of electric vehicles. Battery costs are the primary variable keeping EV’s expensive, and the knockon effect of lowering battery costs for EV’s (i.e. Tesla’s Gigafactory) will be an expansion of lowcost lithium-ion batteries for stationary applications, particularly as used batteries are traded in for new ones. Tesla is actively pusuing this market.

The real benefit of distributed battery storage is not that it enables renewables to replace centralized power plants, quite the opposite, in a world of growing energy demand we need all the clean energy sources we can get, conventional, nuclear and renewable.

Distributed battery storage, especially if it can be coupled with real time pricing data, enables time shifting and price arbitrage by buffering power for a few hours from periods of high supply/low price to other times of day when the prices are better. It should also enable greater grid reliability and fewer outages.

Just as the internet changed broadcast media from a one-way hub and spoke model to a two-way web model, distributed power generation offers a corollary. Big power plants aren’t going away, but they will be cohabitating the power grid with power produced from the edges. The key to making this work nicely is buffering capacity (batteries) and pricing data that allows participants to know when to store and when to release power. 

Keith,

Lithium-ion battery prices are less than $200 per kWh and dropping. The industry is aiming for $100 per kWh. The market for second hand batteries coming out of EV’s could be very attractive for distributed storage. Ultimately the batteries can be completely recycled and remanufactured.

The real value for batteries is time shifting from one time of day to another depending on pricing. You can see this with solar PV in particular. If you project forward to a time when there is high penetration of rooftop solar then all the PV in a region is producing power at the same time, flooding the grid with supply and presumably lowering real time prices. Distributed storage allows those electrons to be buffered and released at more favorable times. 

The intermittency of wind and solar offers an obvious use for battery storage, but production from natural gas could also get in the game as spot prices fluctuate and market participants seek opportunities for price arbitrage.

Real time power pricing data is a critical missing link.

Jesse, I agree that batteries would be able to effectively load-shift nuclear and make gas peakers unnecessary. However, it’s more likely that solving the issues with making nuclear completely responsive to load will prove more tractable and cost-effective than designing, building, and maintaining an entirely separate storage infrastructure.

The idea that nuclear can’t load-follow is already a bit of an anachronism – France operates all reactors in load-following mode, and they can ramp from 50-100% in under 10 minutes.

 Since most of the currently used nuclear power plants have strong manoeuvrability capabilities in their designs (except for some very old NPPs), there is no or limited impact (within the design margins) of load-following on the acceleration of ageing of large equipment components. However, load-following does have some influence on the ageing of certain operational components (e.g. valves), and thus one can expect an increase in maintenance costs. Moreover, for older plants some additional investment could be needed, especially in instrumentation and control, in order to become eligible for operation in load-following mode.

Load Following with Nuclear Power Plants

 One possiblity for improving responsiveness would be to integrate a small modular reactor, used solely for load-following, with two or more baseload reactors.

I quite agree about the importance of electric vehicles as drivers for new battery technologies and cost reductions. See Eric Wesoff’s article on how Tesla’s Powertrain group is driving the grid-scale energy storage business. I also agree with Edwards other points.

Batteries needn’t be “as cheap as dirt” to have a sizeable impact on the grid. There’s a continuum of applications for battery storage, all influenced by the cost of storage. For example, some people already choose to live in remote areas “off the grid”. The electricity they use is extremely expensive by grid standards, mostly due to the high cost and short lifetimes of available batteries. But those people judge the cost to be worth the benefits they get from off-grid living. They represent an already-viable market for off-grid power systems.

If cheaper, longer lasting batteries come along, the market for off-grid systems will certainly grow. When coupled with some other on-going trends, the growth could be explosive. The other trends include:

• falling prices for PV panels and supporting equipment
• high bandwidth, location-independent internet connectivity via networks of low orbit satellites
• energy-, water-, and labor-efficient indoor farming systems
• the exhorbitant cost of land and home ownership in most urban and metropolitan areas

The current market for off-grid systems is small enough that several doublings would still leave it too small, in itself, to seriously impact the grid. But other applications that would be advanced by cheaper battery storage would also be stirring. E.g., the EV market, commercial buildings with battery-backed, non-interruptable DC power distribution, and who knows what else.

The bottom line is that battery storage is foundation technology; advances in that area will have big effects on the shape of the economy well before it becomes cheap enough to be used for large-scale load shifting.

Thanks for an interesting article, Jesse. 

A few points:

1. An example of a flow battery (similar characteristics to liquid metal battery) with suitable economics for price arbitrage (~$50/MWh spread between buying and selling) would be the following: 40 year lifetime, CAPEX of $60/kWh, BOP costs of $10/kWh, OPEX of $3/kWh/yr, 75% efficiency, and 1 cycle per day with average 50% depth of discharge. The CAPEX, OPEX and BOP costs are all more than one order of magnitude lower than today and the high daily depth of discharge will be applicable only to balancing solar PV in sunny regions close to the equator. See this battery storage article and the associated Excel sheet for more info. 

2. Distributed generation with solar PV faces fundamental limitations both in terms of economics, overall deployment and regional applicability. Power density limitations mean that distributed solar cannot deliver more than 10% of global energy. Seasonal variations mean that solar PV will have a very low value at higher latitudes where a lot of economic activity takes place. The high BOP costs and the low capacity factors related to distributed solar will keep it more expensive than utility scale alternatives. Battery storage can do nothing to alleviate these fundamental challenges. More detail can be found in a previous article. 

3. If climate science is right, we will need large negative emissions from the power sector starting from mid century even under the completely unrealistic assumption that the world embarks on a well coordinated climate mitigation pathway starting tomorrow. Battery storage and renewables cannot achieve this purpose. In fact, large quantities of storage will substantially increase the overall energy demand (and associated carbon emissions) due to thermodynamic losses and embodied energy associated with this new additional grid service which has not been necessary before. 

Based on this information, it appears as if battery technology will be limted to low-volume applications like frequency regulation, peak load in sunny regions, transmission investment deferral in isolated regions and hybrids/PHEVs. Don’t get me wrong, there is a lot of value to be added here, but I cannot see batteries transforming everything as many advocates claim. 

It’s also worth noting that there are other good options for supporting load following from a predominantly nuclear energy economy. Storing surplus output in battery banks or wasting it are hardly the only possibilities. One can hold the thermal power from the reactor constant, achieving optimum utilization of the resource, but vary the amount used to generate power.

Heat not used to generate power can be sent into a geothermal heat store. That heat store could then be tapped as needed to provide peaking power beyond the capacity of the reactor. It could even be drawn down to supply power for intervals when the reactor was down for maintenance.

Roger, I agree that the market for off-grid systems could be explosive with cheaper batteries. Whether that’s a positive development for the environment is something else.

I can’t find reliable statistics for the percentage of off-grid Americans who rely on diesel/firewood/propane to augment their generation, but my guess is it’s substantial. An off-the-grid Maine resident who, even with diesel generation, uses 81% less electricity than the U.S. average:

The bottom line is that our solar production provides only 46 percent of what we need, and what we need is minimal.

The result is that we run the diesel generator every few days. In four months, we’ve had 89 hours of noise and smoke and burned 45 gallons of diesel. Not what I expected.

http://green.blogs.nytimes.com/2012/01/18/the-pleasures-and-pitfalls-of-off-the-grid-solar/

That $375/kWh figure was for an older liquid metal battery (using magnesium/antimony).

The Nature paper gives a figure of $68/kWh for the lithium/lead-antimony battery (or $65/kWh at a larger scale – see Extended Data Table 3).

Those figures are for the electrode material only, the cost of electrolyte and manufacturing are not accounted for (“Detailed cost calculations for the cells are presented in Extended Data Tables 1–3. A balance of system and salt costs is not included because the technology has yet to be fully developed on the commercial scale, and so there is no accurate basis for such estimation.”).

Assuming that total battery cost was $85/kWh, capable of 7300 cycles and with the 73% round-trip efficiency, that makes the cost about 1.6c/kWh.

“Further, since this battery operates at 450 C, we would have to generate a lot of heat to use it. It’s completely unclear how that heat could be generated”

Sadoway’s view is that the energy lost from daily charging/discharging (27%) would be sufficient to keep the battery at its operating temperature. (The earlier Mg||Sb battery had an operating temperature of 700 C, which would probably have required external heating.)

“Most grids upto around 65-70% [nuclear penetration] can operate their fleet at 100% output all the time.”

Yeah, the thing people forget is that modern nuclear plants have an average capacity factor around 90%, but it’s really 98% during the summer and winter peaks, and 80% the rest of the year (due to scheduled maintance and refueling). So nuclear is really better than pure baseload, as its output is positively correleated with the seasonal demand variations.

You’re bashing a straw man, Roger. Seasonal storage is a very different beast than daily load smoothing.

Of course Li-ion batteries will never be a viable solution for seasonal storage. The only good solutions for seasonal variations in demand are what you mentioned — stored fuels for seasonal power generation — or load balancing with planned seasonal industries (e.g., synthetic fuel production), or simply overbuilt capacity with seasonal shut-in of unneeded capacity.

Geothermal heat storage would also work.

“We would need batteries to go to an ultra-low carbon nuclear-heavy system…”

Batteries are not needed in a nuclear-rich grid which also has large-scale dispatchable fuel synthesis (ultra-low carbon must include low carbon fuel, not just electricity).  Water electrolysis can be throttled up and down about as fast as batteries (much faster than steam or gas turbines), thus it can provide frequency regulation and the other ancillary services you mentioned.

Of course one key to making fuel synthesis practical is that it must make a high value fuel (e.g. transportation fuel), rather than simply replacing the low-cost/low-value fuels which are used for electricity production.  In this way, the poor round-trip efficiency of power-to-fuel-to-power will not matter.  The other key is do the fuel synthesis in countries like China and India which have cheap sustainable energy.

“...batteries could simply be a viable alternative to gas … peaking power plants that operate today with very low capacity factors…”

Well no, if you operate batteries with very low capacity factors, you’re back to very expensive electricity coming out of the batteries.  If you use Lithium-ion batteries, they’ll wear our due to age before you get your money’s worth out them, and if you use liquid metal batteries (which have longer calendar lives), you’ll have to buy energy to keep them warm when you’re not using them.  

In contrast, if a few of your fuel synthesis plants are only shutdown a few weeks per year (to free-up electric capacity), then so much the better.  And of course in synfuel plants that have a hydrogen intermediate product can store that in tanks which hold a few hours worth in order to run the rest of the process at a constant rate (for process simplicity) while providing daily load mirroring.

I think the best hope for grid-scale batteries is in isolated places like Maui, where grid demand is too small for nuclear, and so sustainable energy necessarily means a large solar and wind contribution, and where little geographic smoothing is possible.

“There is simply no way to do seasonal storage“

Certainly not with batteries, but it is clearly possible with liquid syn-fuel, for example ammonia.  Ammonia (NH3) is an unpressurized liquid when chilled to -30C.  For large tank sizes (building sized), the per-BTU cost of insulating the tank is negligible, as is the refrigeration cost.

While it is unlikely that ammonia made from electricity will every be as cheap as coal, it does seem feasible for it to be competitive with imported oil (which many countries buy).

You can’t get 30% conversion efficiency from the sensible heat in hot water — even if it’s pressurized and heated to well over 100 C. You might get 30% from the first few degrees of extraction, but with each degree of heat extracted, the conversion efficiency drops a bit lower. The heat cycle to obtain even a decent 15% net conversion efficiency is quite complex.

Thermal stores can provide large-scale energy storage, and they can even do so with relatively high round-trip efficiencies. But you have to go with quasi-isentropic heat pump/engines and separate hot and cold thermocline heat stores in graval columns. In the UK, Isentropic Energy, Ltd is promoting that type of system.

There is no need to re-invent the wheel.  The US Dept of Energy’s NREL seems to do studies on energy storage every few years as part of their solar energy program.  Here is one that focused on trough solar thermal plants.  

They don’t discuss hot water storage, presumably because the high pressure would make it uncompetitive with other solution (remember that the pressure vessel for nuclear reactors is made in a giant forge, but the steel containment for the reactor, as well as the thermal energy tanks in the report are welded together on-site). 

I don’t advocate using power-to-fuel-to-power, due to the poor round trip efficiency and poor economics compared to continuing to burn fossil fuels.  Rather, make synfuel for use as transportation fuel, and use the synfuel plant as a dispatchable load to match electrical supply and demand.  

Build the grid to have enough nuclear, geothermal, and hydro capacity that no peaking plants are needed; then use the exess capacity for fuel production.  This will likely not quite make enough fuel to satisfy all transportation demand, hence there is still plenty of room for battery electric vehicles.

If you have nuclear/renewable capacity that is curtailed over 30% of the time, there is no economic justification to discard the energy rather than converting it to fuel. (However, there is inertia which makes people want to continue using polluting hydrocarbon fuels, which are harder to synthesis).

Another article on energy storage which deliberately omits to mention that proven and perfected PHES offers vastly greater energy storage from a synchronous unit capable of providing five times more grid stabilizing system services than a coal, gas, or nuclear plant. System services are the grid’s Achilles heel when accommodating destabilising asynchronous wind and solar generation facilities, and the ability of batteries to deliver vital synchronous inertia is zero. The issue of highly toxic materials used in batteries of limited lifetime is also very relevant, in comparsion PHES plants are built of concrete, steel and some copper, and have a potential lifetime measured in centuries while the energy storage is a function of gravity acting on water. Batteries are fantastic in an iPad, but not for the TWh level of energy storage combined with system service provision required in a wind / solar dominated grid.

It works because the system as historically constituted had the sheer mechanical inertia of many large rotating machines, both generators and motors.  We still have this, mostly.  The system as envisioned by most “renewables” advocates has little or none, and they haven’t a clue about the need to replace it let alone how to do it or how much it will cost.

That’s a new one for me.  Thanks for the heads-up.

After looking up the article, I found that the battery systems proposed have extremely limited capabilities:  one to six hours at rated power.  This would indeed take care of frequency regulation on the scale of seconds to minutes, as AC Propulsion determined more than a decade ago.  Perhaps they can also serve as spinning reserve.  But such batteries are far too small to buffer daily and longer ebbs and flows of wind and solar power.

this would eliminate the notion of “must-run” generation capacity altogether.

To replace it with what, at what cost?

If your 1-hour battery bank is allowed to go down to 25% charge before backup is started, your backup can take up to 15 minutes to start.  There are plenty of plants that can come to full load in 15 minutes, including the LM100 gas turbine.  But these are expensive-to-run peaking plants, and if they’re to replace coal most of them will be new (so amortization must be paid).  What you’ll wind up doing is using batteries as buffers to allow cold-starting of peaking plants, and the peaking plants as buffers while cheap coal-fired plants are started.

The Greentechmedia article claims that the batteries “can break down one of the last barriers to 100% renewable energy.”  Nothing could be further from the truth; in any energy deficit greater than the batteries’ capacity, thermal (either fossil or nuclear) plants must pick up the load.  The batteries are grossly inadequate to supply capacity for the grid, so the existing plants will remain.  The batteries could be used to minimize the use of gas-fired peaking generators, while carbon-free sources like nuclear handle the base load, hydro handles the remaining load-balancing, and wind and PV add something when and where they are useful.  But I’m sure the game plan is to use the batteries to declare “must-run generation is passé, so all you nuclear plants shut down” and turn the market over to the natural-gas interests at a huge increase in both cost and CO2 emissions.

The so-called Greens are quite open about their agenda to replace nuclear power with “renewables” plus gas.  They should all be considered ecocidal criminals.