Electricity stored as temperature difference | Hacker News Reader

Electricity stored as temperature difference

windpowerengineering.com

81 pointssimon_ks42 comments

Europe has already looked at industrial freezers as batteries.

Take them to -40 when electricity is cheap let them get to -20 during peak.

http://spectrum.ieee.org/energy/the-smarter-grid/swiss-wareh...

They call it a "virtual battery" but that seems misleading to me - I don't see any way to get electricity back out, it sounds more like it just saves electricity consumed at peak periods by using, but they couldn't store more than they could use in the first place?

One element of using intermittent or variable renewable energy is to take the approach of dispatching demand to meet your available capacity rather than dispatching capacity to meet demand.

Presently, electrical generation relies on base-load power -- plants which run very nearly all the time, typically coal or nuclear -- with "peaking" and matching capabilities -- generation which can be rapidly added or removed, with response times typically measured in minutes (grids can change fairly rapidly, but not that rapidly, and much demand is highly predictable based on weather, human behavior, and other patterns).

Some large power loads can effectively be banked, and heating and cooling loads are key among these. By providing a large thermal storage mass (either hot or cold), it's possible to "bank" energy when there's a surplus, or when generation is low-cost, and draw down that bank when energy's scarce. Other alternatives are running energy-intensive processes which can be cycled either rapidly or predictably as needed. Aluminium smelting (very electrically intensive), hydrogen electrolysis, and other industrial processes are candidates for this. It's also effectively how pumped-storage hydro works.

You may not be getting back electricity directly, but by offsetting a large cooling load, you're shifting demand.

Yeah... Well, you can consider a battery if you consider the load (in the grid) needs not to be constant

“With smart controls, we should be able to dynamically adapt the power consumption, to consume electricity when there’s enough power in the grid and not to consume when power is scarce.”

You could get electricity back with something as simple as a stirling engine, but there is no point to doing that, as you aren't storing enough to take yourself off the grid anyhow.

Goldman Sachs does something roughly similar with their air conditioning system (Create ice at night when it's cheap and plentiful, and then melt it during the day to keep their buildings cool)

http://www.bloomberg.com/news/2014-08-01/goldman-s-icy-arbit...

Australia, Portugal, and presumably other countries have systems that heat up Hot Water heaters at night when power is less expensive as well.

This is essentially the same system as Light Sail Energy:

http://www.lightsail.com/

Started by HN's own DaniFong (https://news.ycombinator.com/user?id=DaniFong).

Yes in that it's utilizing compressed gas for energy storage, but no in that in how they're addressing the issue of adibiatic heating and cooling of gas -- the fact that gasses heat when compressed and cool when expanded -- they're actually using this heat itself as the energy storage mechanism, with argon as the working fluid.

Light Sail's approach harvests the provided heat and cooling, but utilizes those for what is effectively a coordinated heating/cooling service. The heat itself isn't part of the energy storage/recovery process.

Both are interesting approaches to the heating/cooling problem.

If you don't allow for the heating and cooling, you run the risk of explosions or of excessive thermal loss from your compressed gas. Explosions are possible if you're storing large volumes of gas in underground former natural gas reservoirs -- injecting large volumes of high-temperature N2 + O2 atmospheric gas risks igniting the residual methane. You also have the problem of icing over your recovery apparatus. As a consequence many CAES (compressed air energy storage) designs incorporate a natural gas burner on the exhaust side which both heats the expanding gas and adds additional energy to the process. The resulting systems are a hybrid of natural gas generation plus energy storage.

The heat is certainly part of the energy storage in lightsail, if the heat generated by the compression wasn't being stored in water and then fed back in during expansion then that energy would be lost.

Is just that with lightsail it is a minor part of the storage, the majority being pressure. They are two very different solutions that approach two similar but differently structured thermodynamic problems, lightsail's system being high pressure/low temperature, and these folk doing a low pressure/high temperature store.

According to their website it looks quite early stage/small scale/experimental. Anybody know how far from feasible and commercial scale they are?

In the case of the OP, even they are at least 5 to 6 years out according to the founders.

Last summer LightSail announced they had completed a full-pressure test of their containment structure at 200 atmospheres. [1] And Danielle Fong tweeted that their pressure tank survived the 'gunfire test'.[2]

The last update I saw showed complete cycle pilot tests in 2015 with deliveries to customers EOY 2015 / early 2016. They should be to market much faster than the OP version since LightSail is working on ~.5MW - 1MW installations and OP is working on much larger versions.

[1] - https://twitter.com/DanielleFong/status/474369529326219264 [2] - https://twitter.com/DanielleFong/status/523188555988746240

Thats actually an impressive efficiency as the maximum theoretical efficiency of this "carnot engine" is 85% at these temperatures.

For anyone curious, Carnot efficiency is the maximum efficiency of extracting work from a temperature differential:

Carnot efficiency = 1 - Tc/Th = (Th - Tc)/Th <- I find the last form easier to remember

Where temperatures are in Kelvin. So in this case 500 C and -160 C are 773 K and 113 K so:

Carnot efficiency = (773 - 113)/773 = 85.4%

It’s a handy approximation for the most efficiency that could be expected from other cycles. So for example an internal combustion (Otto cycle) engine running below the temperature of boiling water would have an expected efficiency at room temperature of about 68 F or 20 C or 293 K of:

Carnot efficiency = (373 - 293)/373 = 21.5%

Whereas jet (Brayton cycle) engines may have a temperature differential of 1000 C:

Carnot efficiency = (1293 - 293)/1293 = 77.3%

Things are a bit more complicated than this because with active cooling it’s not just the temperature of the engine’s components, but the temperature of the exhaust gasses, efficiencies of valves, compressors, turbines, etc. So modern internal combustion engines may reach 25% efficiency and turbines may reach 45% efficiency but I still find the Carnot cycle good for guestimation.

So for example, I remember research in the 90s for making ceramic internal combustion engines lubricated with graphite or exhaust that would run at a higher temperature and have an efficiency closer to jet engines. There were also working Stirling engine cars that would have gotten significantly better mileage because it’s more practical to approach the Carnot limit with the Stirling cycle than the Otto cycle:

https://www.youtube.com/watch?v=H_Vnxapd5fs

The main tradeoff is that there hasn’t been as much research in high compression Stirling engines so they tend to have a higher volume than internal combustion engines at the same power output. But since Stirling engines have significantly fewer moving parts and use external combustion (meaning they can run on any fuel), I could never quite figure out why they were never mass produced. Perhaps if they had been, we would have seen industrial sized Stirling engines with Argon as the working fluid decades ago.

Then again, before the web and Wikipedia it would have been hard to make these kinds of points at a Thanksgiving dinner table.

The maximum theoretical efficiency of any battery is 100%. This one is no different.

What you are missing is that while the electricity generator has a maximal efficiency that is smaller than 1, the thermal pump has a greater than 1 efficiency. Multiplied, they are always 1.

Ok, so this makes a lot of sense for renewable sources of energy like wind and solar - production facilities that are distant from bodies of water can store surplus energy during peak production times (windy or sunny days) and then use that stored surplus when production is stagnant (calm days or at night).

The current practice is to use surplus energy to pump water into a reservoir during peak production, and then converting it to hydro-electric power later when production is stagnant. The Isentropic advantage is that you don't need a reservoir to store the energy, the storage system can be build regardless of geography.

I have two questions:

1. What kind of insulation does it take to keep the gravel hot or cold enough to store the energy for a long time? Is the energy lost due to natural thermodynamic processes comparable to energy lost due to evaporation in a pumped-hydro storage system?

2. If the storage system is geographically independent, could it be moved in an energy efficient manner? The problem with almost every energy source we have right now is that it has to be close to the population it serves. On the other hand, if we could generate everyone's electricy as solar energy in Nevada or nuclear energy hundreds of miles from populations, and then move it to distribution plants, then we could answer a lot of energy questions.

re: #2 - the challenge with power systems and their locality to the population is the cost of Transmission System. The closer they are, the less real-estate you have to dedicate to massive transmission towers, as well as less lost energy due to heat in the transmission lines.

Placing the Isentropic center near a population center wouldn't give you any advantage there.

I read 2. as about moving the storage unit itself. Charge the unit at on place then move the unit and generate power at another place.

Yes, this was what I meant.

As a completely non-technically-inclined person, can someone explain how this is intended to yield electricity as output?

From what little I do understand, the enormous surface area of the crushed gravel acts as a very efficient heat transfer mechanism, so it can cool(or heat) the Argon, depending on which chamber one is looking at.

The pistons are there to cycle the Argon between high and low pressure.. Presumably to keep the gas flow going. But can't this be done with just the pistons and without the pressure gradient? (edit: nvm this part; the heat energy is obviously generated by this pressure gradient).

Then comes the time to get the heat energy out for actual use. The article mentions that the gas flow can be reversed. So presumably the gravel that was previously being cooled, will now be heated. Isn't this just exchanging the heat between the gravel and gas? How does this yield net energy output?

Isentropic PHES Technology Explained https://youtube.com/watch?v=sIxt6nMf-IQ

Thank you! It seems my confusion arose from assuming this is a power generator, while it is a storage device.

Sounds like the molton salt systems that are already deployed. I assume this is better?

If that pumped gravel heat exchanger works that well, why don't they create phase change reservoirs and save most of their capital costs?

Phase change systems like steam engines follow the Rankine cycle, which cannot reach as high an efficiency as systems like this which follow the Carnot cycle.

Not a phase change heat engine, but a phase change heat reservoir. Yes, they'll lose some efficiency, but if they can exchange heat as well as it's implied (from their efficiency figures), they may lose very little.

> a per hour storage cost of $103/kWh

Ouch!

I think the confusion is that the $103/kWh is the capital cost to build the storage rather than the operating cost to actually recover a kWh of energy. So to provide 10kW for 10 hours (for instance, overnight storage for a large house), they'd need a 100kWh system at a cost of ~$10,000, where they're claiming that a pumped hydro system would cost 20% more. The benefit of course to pumped hydro is that you can have hundreds or thousands of MW in storage.

These units would be consistent with other reporting for grid-scale battery systems costing ~$250/kWh to construct:

http://cleantechnica.com/2013/12/18/utility-scale-battery-st...

As for operating costs, I'm assuming that the nameplate storage capacity is the output. So if you have a 100kWh system with ~60% throughput efficiency, you'd need 160kWh of input at your standard energy rates and then the energy coming out would be 'free' -- of course there'd be costs for maintenance and depreciation too though.

> So if you have a 100kWh system with ~60% throughput efficiency, you'd need 160kWh of input at your standard energy rates and then the energy coming out would be 'free'

I think the idea is that this is storage for renewables. As such, you wouldn't need more "input at your standard energy rates" than what it might take to charge the system for overnight use.

Right, I guess I was getting at the 'cost' of energy would be he opportunity cost of not selling it back to the grid which is typically the retail price in areas with net metering. Although in off-the-grid situations, you're right that there wouldn't be any incremental cost to charging the storage.

That makes much more sense. My unstated assumption was that a decimal place had been forgotten -- thanks for doing my homework for me and clearing things up :)

As others have noted, this is capital cost per kWh of capacity.

Compare against Beacon Energy's 25kWh flywheel storage units, which have a cost of $10,000 per kWh.

There are trade-offs. Beacon's system is very highly responsive (flywheels can pretty much take up or deliver as much energy as you want, with response in the second to sub-second level) directly receive and deliver electrical energy, and have extremely high round-trip efficiency (90%+, approaching 99%). As an alternative to spinning reserve they have benefits.

http://www.beaconpower.in/html/technology.html http://beaconpower.com/

The problems are the limited capacities of individual units (25 kWh isn't that much power at grid scale), engineering problems, notably precession (the Earth's own rotation about its axis is a concern for the units), and interesting failure modes (preferred deployment is burying the units in below-grade concrete containment with massive lids -- you neither want systems flying apart nor wandering about the neighborhood at 16,000 RPM should the come unmoored).

Flywheels are among the more expensive energy storage options available, but at least at first glance the cooling option looks reasonable.

You don't have to get to grid scale for 25 kWh to seem like not much.

Your average, underfunded community or educational theater has 150-200 lighting dimmers, each at 2.4 kW. I can consume half a megawatt (200 circuits x 20A x 120V = 480kW) in 6 keystrokes.

To be fair, Beacon gangs its units -- you wouldn't have a single 25 kWh module, but an installation of them. The company's demonstration project is a 20 MW capacity plant with 200 flywheels.

http://www.energycentral.com/generationstorage/energystorage...

The other factor is that when you're operating at grid scale, it's overall changes to the grid flow that you're concerned with. A large city might have a peak load of 2000 MW, and the plant here could handle a fluctuation of 10% -- given the law of large numbers, that's a lot. Sure, you might be turning on a slew of light banks in any given period, but someone somewhere else could be cycling down a cooling plant, or resistance heater, or the like.

I suspect large rapidly cycling loads might also be at issue -- with electrified transit, light rail and trolley buses make high instant demands on the grid, and then return energy through regenerative braking. I don't know how such loads are managed, but they're substantial and I do know that utilities tend to segregate these from other residential and commercial circuits.

You need some LEDs man...

It's capital cost for providing 1kWh of storage for an imaginary 1 hour. I don't think that makes much sense, since you wouldn't be able to scale down the capital costs in proportion to the storage time. $103/kWh*6h = $618/kW which is on Isentropic's own site describing a turbine rather than pistons:

"With a capital cost of only $618/kW for 6 hours of storage ($375/kW power machinery and $40/kWh stored energy) and a round-trip efficiency in excess of 90% this is the cheapest and most efficient form of energy storage."

http://www.isentropic.co.uk/Energy-Storage-Systems

That's got to be a typo, or this story is burying the lead. Is pumped hydro really 30% more inefficient than even that figure? Because that would mean it's almost entirely a waste of money for anyone to store energy by this or other means.

If it's $103/GWh, on the other hand, then we're talking about only ~10% overhead on costs.

> Is pumped hydro really 30% more inefficient than even that figure?

The article states 30% cheaper, not 30% more efficient.

> If it's $103/GWh, on the other hand, then we're talking about only ~10% overhead on costs.

The article states $103/kWh, not per GWh. To calculate overhead costs you would need to know life cycle costs which were not disclosed.

It's not so bad. Less than pumped hydro and conventional batteries and at grid scale. Combined those are big pluses as pumped hybro can't be built anywhere and batteries are not suitable for grid scale.

The question is, what are the life cycle costs? If it's on par with pumped hydro, then we are talking.

In a storage context, quoting a price per kwh is somewhat ambiguous. Does it cost an additional $103 every time 1 kwh i stored, or does 1 kwh of capacity cost $103 with a per-usage cost of ~$0?

There's probably a standard interpretation, but it's not obvious to a lay reader like me.

Somewhere in the middle - the capital costs are $103/kWh - So, if you want to store 1 MWatt for 1 hour (Large Solar / Wind plant with variable sun/wind, and fixed size upstream transmission), it will cost you $103,000 fixed costs. OpEx is a function of maintenance, operation, life time, and duty-cycle of the plant.

I was wondering what that means too - maybe the strange unit ($103 kWh per hour) is a clue to what they are talking about. Remember this is temporary storage for wind or solar energy, presumably to be extracted when the sun goes down or the wind stops turning the blades.

Batteries are currently at $250, best case. Up to $500, depending on who you talk to.

Reply on news.ycombinator.com