Hacker News Reader

Comment by arc-in-space | Hacker News Reader

0 comments after 20 minutes on an energy post? How am I supposed to know why this won't work, won't be useful, won't be cost effective, won't scale, and that it's just a fad?

Jokes aside, this seems impressive, I have no idea what the best applications would be but wikipedia claims that current similar devices have fairly bad efficiency(https://en.wikipedia.org/wiki/Thermoelectric_generator).

View on HN

This is not similar to anything on that page, as it operates on temperatures of thousands of degrees. The comparison with steam engines is also quite bad, as the Carnot efficiency on that kind of temperature difference is way above 90%.

So, it's just an overrating research PR piece, like the ones people like to complain. This thing probably scales just fine, and may be quite useful. The entire problem is that science gets divulged on those insane PR pieces where it's compared to completely different things, or promise completely impossible results.

> the Carnot efficiency on that kind of temperature difference is way above 90%.

This is definitely not my area, but is Carnot efficiency directly comparable to the efficiency numbers cited in the article? Or is the "work" in Carnot efficiency the mechanical work, prior to being converted to electricity?

Yes, it's directly comparable. You can interconvert mechanical work and electricity freely; electricity isn't like heat. Everyday machinery does it with 95% efficiency, but there's no fundamental limit.

At those temperatures, Carnot efficiency is between 80 and 90 percent.

The comparison to steam engines is misleading, but there’s an important distinction. Steam or gas turbines would reach very high efficiencies at these temperatures too, but won’t because of material properties and limitations thereof.

These limitations don’t seem to exist for this new technology. Hence, reaching very high efficiencies becomes possible. In theory… In practice, I don’t see how heat sources with temperatures that high are feasible or could stem from renewable sources. (something with the thermal battery? Wasn’t explained much in the article)

In any case, in comparison to steam turbines, the technology presented here does absolutely nothing in terms of decarbonising the grid, as claimed. It’s just potentially more efficient. But what’s the source for the primary energy?

I think the (unstated) idea is to use an arc furnace, during peak solar/wind output to heat graphite and recover the energy later using their fancy new TPV cell. That's going to require some really good insulation, since your heat source is intermittent and your temperature difference is huge.

My first thought was let's use it in fission (and later fusion) reactors.

It's too hot for a fission reactor. Probably for fusion too, but that's not clear.

I got the impression they would heat the graphite with concentrated solar power.

The problem is it only generates electricity when the input temperature is 1900-2400 C whereas uranium melts at 1132.2 C.

Passively slurping the heat off a critical puddle of molten uranium is the disruptive startup pitch I'd make to secretly record audience reactions.

With unmoderated fast neutrons, and critical geometry your startup will be always exceptionally close to booming and taking over a huge flank of the market.

You can use this as an inspirational soundtrack for the presentation: https://sentientruin.bandcamp.com/album/wormboiler

Most civilian reactors use uranium dioxide, which has a melting point of 2865°C. Or uranium carbide, which melts at 2350°C

I think they're talking about something like this:

https://www.theverge.com/2022/2/22/22945975/rondo-energy-dec...

Basically, in their words, "a large insulated shoebox full of brick". And I could be wrong, but I think you should be able to scale the amount of "brick" up to whatever size and keep the insulation the same thickness, so the storage capacity would increase by the cube of the scale and the amount of insulation would only increase by the square of the scale.

That would allow you to minimize the fluctuation in temperature - i.e. if it takes 10 days to get up to temperature, because it's big, you don't have to cool it all the way back to room temperature when you take an afternoon worth of energy back out.

Heliogen for solar, and Quaise for geothermal. Both have high temperatures to deal with.

"in comparison to steam turbines, the technology presented here does absolutely nothing in terms of decarbonising the grid, as claimed"

I understood the claim to be that this would have close to no moving parts. That sounds it's cheaper to mass manufacure and operate?

The actual generator has no moving parts. The "tanks" for storing the heat can be made from graphite, but the thermal battery made by combining the storage tank with the generator that they propose has to pump heat around using liquid tin (or perhaps liquid silicon) as the working fluid, at temperatures up to 2400C. That requires not just moving parts, but some pretty far out engineering. All of the metals we commonly build things like pumps out of are liquid at those temperatures, after all. And of course, you want pumps that run reliably for years in that hostile environment.

Of all things keeping the energy transition back, steam turbine manufacturing is probably very low on the list. I’m not aware it’s an issue. It’s an old and proven technology.

Not to beat a dead horse, but nuclear?

For non-mobile storage, it seem that the waste heat (from cooling the TPV) would still be at such a high temperature that it can be used for co-generation to improve total system efficiency. Do existing technologies exists to make optimal use of this "temperature bandgap"? Would direct to steam work efficiently?

This device doesn't really change the energy landscape. Let's rephrase the title: "A new heat engine is as efficient as a steam engine but needs a thermal source 1,800 degrees celsius hotter to work". The device described in the article is interesting in that it has no moving parts and might have an application one something like a nuclear powered spacecraft. Actually trying to harvest energy from TEGs is exceptionally difficult, since renewable energy sources aren't nearly as energy dense as thermal fuels like hydrocarbons or fission. The thermal gradients produced through renewable sources like solar are tiny [1]. It could be used for something like geothermal power, but again it needs temperatures way hotter than conventional steam engines which already work fine for geothermal energy production.

1. https://www.youtube.com/watch?v=Tdge8vEODeY

Re. thermal gradients, have you considered plain old mirrors? https://en.wikipedia.org/wiki/Odeillo_solar_furnace

> 54 metres (177 ft) high and 48 metres (157 ft) wide > more than 2,500 h/year [sunlight] > peak power of 3200 kW > Temperatures above 2,500 °C (4,530 °F)

Sounds like it could be useful as a "default load" inside an otherwise inactive solar furnace at least.

You're describing solar thermal energy [1]. Use solar collectors to turn light into heat, then use a heat engine to turn that heat into electricity. This TEG could be used as a heat engine for this task. But again, our heat engines are already capable of this task and don't need such high temperatures. A solar collector array even getting to this TEG's operating temperature might not be feasible.

Photovoltaics just turn solar energy into electricity, and don't need the heat engine. This has made them way cheaper to deploy than solar thermal energy. So unless there's something very important about this new TEG, the solar thermal vs photovoltaic calculus doesn't really change.

1. https://en.wikipedia.org/wiki/Solar_thermal_energy

The point is to store the thermal energy for later use to smoothen out photovoltaic's intermittency issues.

Right. But we already have that technology with conventional heat engines which have the advantage of much, much, lower operating temperature requirements. If you have a 3,000 degree vat of thermal storage material this new engine stops working after draining 1000 degrees. Existing heat engines can usually work down to several hundred Celsius - though superheated steam engines need around 700 Celsius. But that's still an extra 1000 degrees you can bring it down, even in the conservative case.

This is the first of its kind to reach this efficiency, correct?

I assume that means there could be room to significantly improve its efficiency or operating requirements with more investment and research.

Having one example, even if 'useless' or elementary, is key to developing new technology.

Compared to other TEGs. Not compared to steam turbines. The article is actually being very generous in saying it's "as efficient" as steam turbines. Steam turbines are more efficient with scale, and industrial ones for power generation are over 90% efficient [1]. This new TEG's efficiency is "around 40 percent". Higher than the previous TEGs in the 25-35% range. But not compared to steam engines, that also benefit from much lower operating temperatures.

1. Multistage (moderate to high pressure ratio) steam turbines have thermodynamic efficiencies that vary from 65 percent for very small (under 1,000 kW) units to over 90 percent for large industrial and utility sized units.

https://www.epa.gov/sites/default/files/2015-07/documents/ca...

This is an absolute game changer for planetary exploration, where RTGs are common.

RTGs do not get anywhere close to 1800 degrees Celsius. Even if they did, it wouldn’t be a game changer, because you can make up for loss of efficiency with a bigger RTG.

OK, perhaps I stand corrected (or just tempered).

> "A new heat engine is as efficient as a steam engine but needs a thermal source 1,800 degrees celsius hotter to work"

But this title would not be correct.

Methane combusts at 1,957 °C, right in the band for this thermal cell, and is a normal heat source for steam power.

The steam doesn't reach this temperature, of course, but nor does it need to.

>why this won't work, won't be useful, won't be cost effective, won't scale

Not an expert, but reading this a few negatives popped out. Basically they are heating a black body to 2400C and then making electricity from gathering the emitted light in a cell. They get to pick a temperature to match the bandgap of the cell.

The key problem is getting something that hot without using another (lossy) form of power. The Sun's surface is ~5600C so that's enough headroom to get there from solar. That's cool. But are there any fission reactors that get (or could get) that ridiculously hot?

Chernobyl did.

I think most try to keep temperatures under 1000C. I think many FAST reactor designs are looking at 600C operating temps with peak temp reaching maybe 1200C during emergency testing. But my memory might be wrong.

Biggest key problem imho is how they expect to store this heat energy. It looks like this cell will, like a PV cell, constantly be absorbing photons. If those photons aren't creating electricity/voltage across a gap then they are being converted into heat. So to keep the medium at temperature you would need to insulate it, to wrap it in a mirror, only exposing the flux to the energy-collecting cell as needed. That means moving parts.

As for available temperatures from fission reactors: https://en.wikipedia.org/wiki/NERVA

"When the reactor was operating at full power, about 1,140 MW, the chamber temperature was 2,272 K (2,000 °C)"

This is a bummer, but probably a reality. Perhaps some kind of LCD type tech could make for it to be more digital/less big metal moving around.

"The team’s design can generate electricity from a heat source of between 1,900 to 2,400 degrees Celsius"

That's way up there. That's well above the melting point of steel. That's above the highest temperature jet engines made for experimental aircraft.[1] Most jet engines try for exhaust gas temperatures around 600C or so, for a long useful life. Typical nuclear reactors, around 300C.

It's not impossible to operate up at those temperatures. Every steel plant does it. There are ceramic and brick materials that can deal with such temperatures.[2] The storage medium would probably be some molten metal.

This seems way too much trouble just to store energy.

Now if this thing worked at 600C or so, there would be more uses.

[1] https://www.nasa.gov/centers/dryden/pdf/88068main_H-1375.pdf

[2] https://www.ceramicsrefractories.saint-gobain.com/refractory...

The article mentions the planned storage medium is insulated banks of graphite.

I'm a little unclear on the concept. You say storage medium, as if the graphite is expected to remain at very high temperature for an extended period. Preventing a chunk of super hot graphite from cooling is, if anything, an even larger engineering challenge than getting it that hot in the first place!

I think they’re envisioning a no-moving-parts kind of system, which makes dealing with difficult environments easier.

No, they claim no moving parts for the generator, but refer to engineering designs that use pumped liquid tin to move the heat within the system. Clearly the challenge there is building a pump that can handle liquid tin at 2400C.

Liquid metals can be pumped with a linear motor type magnetic field, with the liquid metal being the moving part.[1]

But all this just to replace a battery?

[1] https://www.comsol.com/model/inductive-liquid-metal-pump-590...

I think the sheer size would make it interesting. The heat energy potential of acres full of graphite is enormous and presumably much cheaper to construct than an energy equivalent battery. Now I wonder how it holds up to other methods of storing energy.

I could be wrong, but I don't think you'll find EM pumps in use for any liquid metals at temperatures above about 1000C. The 2400C temperatures required for this thermal battery concept present a significant materials engineering challenge for anything that touches the heat storage medium. There just aren't many mechanically sound, or electromagnetically capable solid phase engineering materials to work with at that temperature.

I'm not saying what they describe can't be done, only that getting the photovoltaic part to work isn't the biggest engineering challenge the concept faces.

Not mentioned in the article is power density. How quickly can the energy be released? Consider solar panels, you need a table sized cell to get 100W. That can make for a big battery to get grid scale power output if these cells are only as power dense as solar panels. The energy density of a heat based solution can be very high- metals can get very hot and they are dense enough to store a lot of energy. But if you can’t get the energy out of the battery fast enough that limits the applications. By comparison lithium ion batteries can dump power out extremely quickly, which is what makes them great for cars. Hydro is even better.

Surface area is relevant for solar because the sun is so far away. A local heat source allows you to surround it with 3D shapes not just a flat plain.

As to temperature this thing is for very high temperatures: can generate electricity from a heat source of between 1,900 to 2,400 degrees Celsius. At 40% efficient you need a wide temperature difference which would suggest a high energy density.

For long term energy banking and if we can get them working, flow batteries seem vastly superior to all alternatives, by scaling storage with regards to tank volume. Instead of some difficult-to-manufacture structure.

This design is photovoltaics, just like solar, but optimized for infrared photons. There is no avoiding the reality that energy storage density will scale as the cube of the facility size, but power density only as the square. And at 2.38 w/cm^2, the scale coefficient is not all that great.

Picture a stack of flat panels with each layer consisting of: (Cold)(Panel)(Hot)(Panel)(Cold) held vertically. So: (Cold)(Panel)(Hot)(Panel)(Cold)(Panel)(Hot)(Panel)(Cold)(Panel)(Hot)

Now you add hot gas at the bottom and have say 4 layers per m. So a 3mx3mx3m cube would be 4 layers * 2 panels per layer * 3m * 3m * 3m = 216m2 of panels taking up a 3mx3m section of floor. At 2.38w/cm2 * 10000cm2/m2 * 216m2 = 5.14 MW of power.

I think their application is grid scale and you can scale across hundreds of batteries to provide the throughput you need. I don't know how I feel about having a small molten ball of metal inside the hood of my car. Turns my car into the most dangerous gusher in the case of an accident (for those who aren't familiar, gushers are a gummy like candy with juice inside).

The article in Nature quotes an energy density of 2.38 w/cm^2. Which means a Gw battery would require 10e5 m^2 of absorber surface, exposed directly and at close range to the radiation from molten metal (which is the heat transfer fluid they propose). It has to be direct, and at close range, because the efficiency they quote relies on the absorber reflecting non-absorbed photons directly back into the emitter, where they are re-absorbed as heat and potentially re-emitted.

That's about 25 acres of absorber, and an implied 25 acre surface area of the liquid metal emitter pool.

There is a basic challenge here to the design - the energy storage density for the thermal battery they envision scales as the cube of the characteristic dimension of the plant, but the power density that can be delivered scales only as the square of dimension. Not saying that can't be dealt with in engineering, but it ain't going to make this easier or cheaper.

> won't be useful, won't be cost effective

> The team’s design can generate electricity from a heat source of between 1,900 to 2,400 degrees Celsius

Pretty high temperature for me, copper melts at 1800C.

High temperature is the point. The efficiency of heat engines depend on the temperature difference (relative to ambient). The hotter you can go, the better. (granted this thing isn’t really an “engine”, but the trend still applies)

But for practical applications, the availability and cost of materials to survive the operating temperature is also relevant.

A Carnot heat engine operating between, say, 2600K and 400K can reach almost 85% efficiency.

The higher the temperature, the higher the share of exergy in the heat flux. At high enough temperatures, it’s no longer a feat to convert to electricity at high efficiencies.

It would be pretty hard to create a carnot heat engine that can withstand 2600k. I'm not even sure if for example Tungsten has structurally integrity at that point.

That might be when copper liquifies. It will lose its structural integrity and deform easily at much lower temperatures.

Pfff! Amateur! You should try pulling the plug out of the tub when its filled with my wife's bathwater.

Thanks for the laugh! So true.

Exactly why I clicked in, but man this technology sounds like a serious gamechanger.

No news is good news, this one probably will work I guess.

Reply on news.ycombinator.com