Engineers at MIT have developed a grid-scale energy storage concept known as the ‘sun in a box’, which they claim would be half the cost of pumped hydro storage.

The system relies on the transfer of molten silicon between two cylindrical graphite tanks, each about 10m wide. Silicon in the ‘cool’ tank would be stored at around 1,900°C, then pumped through a network of heated pipes into the second tank, reaching temperatures close to 2,400°C. The elements heating the pipes would be powered by excess energy from the grid, theoretically from renewable sources such as wind and solar.
To feed the stored energy back to the grid, molten silicon from the second tank – now glowing white hot – is passed through a system of pipes that emit light. This light is then captured by highly efficient multi-junction solar cells. Though commonly referred to as the ‘sun in a box’, the system’s official name is Thermal Energy Grid Storage-Multi-Junction Photovoltaics, or TEGS-MPV. The research is published in the journal Energy & Environmental Science.
“One of the affectionate names people have started calling our concept, is ‘sun in a box,’ which was coined by my colleague Shannon Yee at Georgia Tech,” said corresponding author Asegun Henry, an associate professor at MIT’s Department of Engineering. “It’s basically an extremely intense light source that’s all contained in a box that traps the heat.”
The system was inspired by concentrated solar plants that use molten salt as a storage medium. However, molten salt can only be heated to around 550°C before it becomes corrosive, meaning its efficiency is limited. Silicon, one of Earth’s most abundant elements, can operate safely at a much higher range. Key to the overall concept is a pump that can withstand such temperatures, and last year the team developed a pump that holds the Guinness World Record for the highest heat tolerance. Another area of concern was how the silicon would react with the graphite tanks. Testing on a miniature tank with silicon heated to 2,000°C for 60 minutes, the team found that a layer of silicon carbide formed, but actually acted as a non-reactive protective layer.
According to the researchers, a scaled-up version of the ‘sun in a box’ system using 10-metre tanks would allow a small city of 100,000 homes to be powered by renewable energy. Furthermore, it would not rely on natural geography in the way that pumped hydro does, as well as costing significantly less, claim the team. Pumped hydro is currently the cheapest form of grid-scale storage in widespread use.
“This is geographically unlimited, and is cheaper than pumped hydro, which is very exciting,” said Henry. “In theory, this is the linchpin to enabling renewable energy to power the entire grid.”
well, plenty of engineering challenges to go at, on this one! What temperature do the heating elements need to run at, to put heat INTO 2,400°C molten silicon? And isn’t most of the radiation emitted at this temperature infra-red (Wien’s law) c.f. surface of the ‘real’ sun 6000°C?
Sun temperature at surface is about 5500 deg. C, just saying. Trust me, though, you do not want to stare directly at a 2400 deg. C light source!. Solar peak wavelength is about 501 nm, and 2400 C source might peak at 1208 nm or thereabouts according to Wien’s displacement law. This does not mean that plenty of available radiation is not available in the 400-700 nm band of visible light. Nor does it mean there are not PV devices that are efficient in the near infrared. Also, with enough light pressure emitted, perhaps some of the wavelengths could be frequency doubled, although I think that would not be highly efficient use of the available radiation.
heaters are ~2500C, probably tungsten since graphite evaporates too quickly at that temperature.
yes mostly infrared, but near-IR unlike more TPV, so higher bandgap cells are used
Another great engineering accomplishment on the road to 100% renewable energy, a road that we must shorten urgently. You go, Tim the Beaver!
The storage battery is, in my opinion, a catchpenny, a sensation, a mechanism for swindling the public by stock companies. The storage battery is one of those peculiar things which appeals to the imagination, and no more perfect thing could be desired by stock swindlers than that very self same thing. … Just as soon as a man gets working on the secondary battery it brings out his latent capacity for lying. … Scientifically, storage is all right, but, commercially, as absolute a failure as one can imagine.“ (London, February 1883, “The Electrician”)
Right then! No less right today!
The argument is flawed, and a bit blind. Utility companies routinely store off peak demand energy, and withdraw it during peak when price is more favorable, and customers need the power. It is a win-win formula. It is also (in this form) very fast to dispatch compared to prime mover (turbine).
I see three critical parameters, but the article only gives one of them.
The first is the cost of building the plant, which is given as half the cost of Pump Storage, so that’s good.
The second is the power density. The biggest limitation for pump storage is that it takes a massive volume to store a relatively small amount of energy. I found another article on the web that says the storage density (of “sun in a box”) is 36 times that of lithium batteries (so that’s probably one thousand time the density of pump storage) which is excellent.
The third parameter is the overall energy efficiency of the system. Solar cells can be up to 33% efficient, but there will be other losses in the system, so I think they would be lucky if they could get back 15% of the energy they put in, which is a weakness.
If we had so much green energy that we were throwing it away when there was a surplus, this system would be ideal.
You missed the statement about multiple layer PV, which are projected to reach the actual limit of 83% efficiency in the proximate future. Already, these efficiencies are in the 70%+ range (very akin to pumped storage). IF waste heat is recovered and utilized (and I suspect it must), then the numbers only adjust upward in this case, as the heat would be “high grade”.
Yes, this system is designed for the case that excess electricity is available, which will occur if renewables grow. And renewable will grow if there is a way to store it… chicken and egg.
Check out the paper for details. the energy density (what you meant by power) is not as high as you found, and efficiency is ~50%, although not demonstrated yet.
https://pubs.rsc.org/en/content/articlelanding/2019/ee/c8ee02341g#!divAbstract
Wouldn’t any inefficiencies in the system just be more heat?
So why would there be any lose of energy in this closed system?
i.e. Why wouldn’t it be >99% efficient?
round trip efficiency ???
50%,
https://pubs.rsc.org/en/content/articlelanding/2019/ee/c8ee02341g#!divAbstract
This is the one I have been waiting to see more of, and I surmise the PV cells (multichroic) are ready for the real world, and can take the heat. I suppose they might also remove some heat from the PV cells if necessary, and this might be “high grade” heat, allowing a closed Brayton cycle (SCC) to operate off that heat, even more energy conversion. This is distinct from the molten silicon system being deployed now in Australia (1414 C).
If the box they are describing is 1000 cu. m then 1 degC of temperature change will involve 2×10^9J (as 1cc takes 2J of energy/deg so 1 cu m takes 2 x10^6)
And 500 degC will involve 10^12 Joules; which, upon dividing by 3.6×10^6, gives about 278000 Units (kWhr) of electricity.
I think this is a pretty good figure and shows he efficiency of sensible thermal storage.
As the average household uses (roughly) 10 units/day (and 38 for gas) this gives 27800 houses or 6000 houses (if all energy taken into account).
I must admit I am not so sanguine about the use of photovoltaic cells and suspect that a gas turbine (using nitrogen instead of air – otherwise goodbye graphite) and possible combined cycle steam turbine might have better efficiencies – and be off-the-shelf technology.
Interestingly a downsizing approach might be more useful – most houses could have a 1 cu m box or two in the attic! (can you get a 10 kW gas turbine generator?)
I should mention that the corrosiveness of molten salt (and lead) is an issue for nuclear industry – but they are going ahead at much higher temperatures, than those mentioned in the article – at a guess using SiC composites.
I did follow up the link to the paper and think that their arguments for a liquid storage medium- as opposed to a solid (cheap) one were interesting – but only valid for the photovoltaic approach.
Still I think this shows thermal storage to be a potentially good and affordable energy storage option (and of course also useful for process heat as well as solar or thermally driven electricity generation).
I wonder why silicon is chosen. It is highly reactive and at 2000 c it will simply eat away graphite. Tin could be a better material.
Wouldn’t it be more efficient to use the heat to directly drive a turbine generator?
Carnot efficiency for the heat engine operating between temperature limits 2400°C and 1900°C shown here: https://pubs.rsc.org/en/Image/Get?imageInfo.ImageType=GA&imageInfo.ImageIdentifier.ManuscriptID=C8EE02341G is ((2400+273)-(1900+273))/(2400+273)*100%=18.7% by my reckoning. ‘Real world’ engineering always achieves less than the Carnot efficiency
Not entirely sure of the efficiency of using solar cells; I suspect that it depends what temperature they run at – and any re-radiation from their surface.
If the heat source was being used to drive a gas turbine then it would depend upon the exhaust temperature of it (or they, if cascaded with, for example, steam turbines) to define the Carnot efficiency (I believe that is what is planned for high temperature reactors).
Such an off-the-shelf gas turbine could be nitrogen driven – and thence limit reactivity with graphite; though I am not too sure if graphite is too hot as CN might result ).
Both liquid tin and lead have been proposed for high temperature (nuclear) reactors (sodium boils at < 1000 degC) – but corrosion and toxicity are issues (so getting materials right is ultra-important)
If a gas turbine was used (with appropriate heat-exchanger) then its (gas) inlet temperature (and efficiency) would vary somewhere between 2400 degC and 1900 degC (If exhaust was a at 273 degK then the Carnot efficiency would be (1-273/2673)x100% ~= 90% )
I do not understand why today’s engineers keep thinking in terms of traditional expansion turbines whether Rankine cycle, or Open Brayton cycle, when it has been demonstrated clearly that the closed Brayton cycle is superior in every way. SCC (supercritical carbon dioxide) has the energy density required, and compression energy input at the low temperature is far less than that for air or nitrogen compression, as the compressor operates almost as a liquid pump.
The material science here is key, as is the selection of support of the rotating equipment. Oil-less bearings are necessary, as SCC is a great solvent. Use “air-stone” bearings, or magnetic, as either will work in this situation. I think one will find the efficiency of PV is far higher than initially calculated by some on here, as it is not a “mechanical” process, although the basic idea is that there is a theoretical limit, the calculation here was done incorrectly.
If the inventors can improve on Carnot efficiency they will be in line for the Nobel prize in physics, for disproving the second law of thermodynamics https://en.wikipedia.org/wiki/Second_law_of_thermodynamics#Corollaries
Totally agree. The efficiency of power in to non-heat power out (electrical/mechanical/chemical/gravitational/whatever) cannot exceed 500/2673, or about 19%, with given temperatures. Probably far less in practice. Which is not necessarily bad, depending on use for the wasted heat.
Though the use of supercritical carbon dioxide seems interesting I think caution would be required in using it at elevated temperatures with graphite; because “producer gas” , essentially carbon monoxide is the result of endothermic reaction – usually at 500 degC or above (and I have no idea if there is a supercritical carbon monoxide…)
So it sounds good – as part of a combined cycle (and alternative to steam) – with a gas being used for the high temperature part of the thermodynamic cycle (argon or nitrogen seem to be cheapest inert gas options for such).
Exactly. My guess that startup costs for the 1414 C version would be way lower then this proposal. So its a question of whether it can come so much more efficient that it would cover the higher startup costs.
A sensible heated storage device has been mentioned that stores at 1414C – but no details given. It would be interesting to know how this compares with the silicon one mentioned in the article.
Also.
I have done a few calculations and note that the power density could be much higher ( more than a thousand fold ) than from insolation; can solar cells handle such amounts of power ?
The radiant emission (Stefan-Boltzman equation) is at temperatures between 2400 and 1900 and so has a radiant emittance between 2.9 and 1.3 MW/m^2.
So a rough estimate is that in order to emit the energy, to get the source temperature to drop by 500 degrees, something of the order of 5×10^5 seconds for a metre square of emission area – or 139 hours; if the emitter area was 10 sq m then it would take about 14 hours (Also the peak wavelength is about 1 micron).
The 1414 C device that has been mentioned would have a black body emittance of about 0.5 MW/m^2 (and a peak at 1.7 microns)
So why would there be any lose of energy in this closed system?
As far as I can tell, the only energy entering or leaving this closed system is electricity.