Helion Energy to demonstrate net electricity production by 2024

[charliem]

Regarding pulse length, if I recall it correctly, in some older Helion's machine FRC formation, acceleration, collision, and compression took a few tens of microseconds. Intuitively, this phase should take more or less the same time in all their machines, or at least stay in the same order of magnitud.

We have little hard data, but know that Venti, their 5th generation device (Trenta was 6th, Polaris will be 7th) achieved an energy confinement time of 40 microseconds (JASON review, Sep 2019).

Skipjack argued in this thread some time back that, given the published scaling law, if Venti did 0.04 ms, Trenta may have been able to reach 0.5 ms. Applying the same to Polaris I obtain 1+ ms, maybe 2 ms.

Anyhow, having long pulses would present some disadvantages that could make prolonging the cycle over 1 ms inadvisable. For starters, the speed of Tritium creation grows linearly once fusion conditions are reached (in a He3-D mix). Add to this that those T ions thermalize in a few milliseconds, making D-T fusions much more probable. This will imply upping the creation of 14 MeV neutrons, and those should be kept at a minimum because of their tendency to wreak havok with the hardware.

[Skipjack]

From what I understand, pulse length in Polaris is somewhere around 1 ms, give or take a few hundred microseconds. They can not make it much longer for the reasons outlined by Charliem. Again, as mentioned, Helion would rather terminate the pulse early and recover the energy than keep the plasma alive longer while leaking particles and dealing with Tritium, etc.

[mvanwink5]

pulse length in Polaris is somewhere around 1 ms

Plasma duration at point of compression in the device center is one time, which I suspect SJ is saying is ~1ms, then there is the magnetic field sequential firing to accelerate the plasma from each end to collide & to be further compressed, and that is extremely short comparitively (less than a microsecond?). Dave was trying to calculate the energy expended to do this plasma formation, acceleration, & compression. I cannot see how Dave would be able to calculate this.

[TallDave]

thanks mvan and charlie, I am a bit more interested in the compressed (fusing) plasma lifetime (since that's when power is produced), but the total pulse length might help us estimate utilization factor (the proportion of time that a reactor can spend producing power) which could then give us an idea what the fusion power has to be for a given continuous power plant output

e.g. if utilization is 10% you'd need 100MW fusion pulse output to produce 10MW of continuous power

it will be interesting to see how Helion balances pulse length against required power

as for trying to calculate the energy expended to do this plasma formation, acceleration, & compression, I am just assuming they all end up as losses to brem or transport (though as noted, this might be pessimistic)

from the output power, we could then infer from potential Q what the inputs must be to provide the necessary losses

I think this approach is okay as a rough estimate because since the fusion doesn't self-power the reaction (i.e. no ignition) there's nowhere else for lost energy to have originated from

[charliem]

thanks mvan and charlie, I am a bit more interested in the compressed (fusing) plasma lifetime (since that's when power is produced)

Helion and its predecessors have published an empirical formula for FRC tau(n) (particle confinement time) for their machines. For example here (jump to minute 16:00).

Making some reasonable assumptions about Polaris dimensions, an accepting particle densities similar to those achieved with previous generations, I obtain that Polaris' tau(n) ~ 2.5 ms

Of course that's only acceptable, at best, as an order of magnitud, but this value is congruent with other known data-points. The computed thermalization time of 1 MeV Tritons (**edit_1) under the expected conditions is also about 2+ ms (if I applied correctly the formula from my old plasma physics book). And also, there's the issue of minimizing Tritium generation.

Helion will not keep an FRC longer than tau(n), the particle and energy losses would make it useless. In fact, as long as they can recover most of the energy injected to create that FRC (and they think they can do it), less than that would be better for said Tritium reasons.

From what I've watched and read, Polaris aims for one pulse per second; the generation, acceleration, collision, and final compression of the two FRCs may take a few hundred microseconds, and as SJ said, the final FRC will be kept for about 1 ms.

So, 1 ms of fusion per second for Polaris. Although we don't know the average power they are aiming for, this means an instantaneous power one thousand times higher.

as for trying to calculate the energy expended to do this plasma formation, acceleration, & compression, I am just assuming they all end up as losses to brem or transport (though as noted, this might be pessimistic)

Well, yes, according to Kirtley this is too pessimistic.

For starters, Bremsstrahlung is expected to be relatively low due to a high temperature differential between electrons and ions (1 to 10?).

Also, Helion aims to recover most of the energy used to create that FRC, and say they think they know how to do it. In fact, this is key for the viability of these machines. Kirtley's mentioned a goal of recovering up to 95%, but even if they only get back 50%, that would mean halving the fusion power necessary to reach engineering break-even.

Add that their plan does not include generating electricity via a thermal cycle and (if they succeed) they should be able to get away with Q-sci values much lower than most of the competition (theoretically Q-eng>1 should be possible even with Q-sci<1).

Of course, not a given, but I'm hopeful.

**edit_1: Originally the text was: "The computed thermalization time of 14 MeV Tritons ...". The Tritons produced from D-D reactions carry 1 MeV only. I used 1 to compute their 2+ ms thermalization time, not 14.

[TallDave]

Making some reasonable assumptions about Polaris dimensions, an accepting particle densities similar to those achieved with previous generations, I obtain that Polaris' tau(n) ~ 2.5 ms... So, 1 ms of fusion per second for Polaris

thanks charlie, that sounds plausible to me

For starters, Bremsstrahlung is expected to be relatively low due to a high temperature differential between electrons and ions (1 to 10?).

the brem/transport losses expected at a given KeV are from Kirtley's graph (a few posts back) where Te/Ti is .1, so that is baked into my BOE analysis

sorry for lack of clarity, I meant pessimistic in the sense Helion might recover some fraction of the energy used to create the pulse... but now I'm realizing that's also baked in because even if you recover (say) 90% of the energy, the transport/brem losses would still have to come from the other 10% you didn't recapture (no free lunches!)

but there's also the fact "that a 14.7 MeV proton in a D–He-3 plasma environment will actually impart more energy through direct nuclear elastic scattering with the fuel ions, than the traditionally modelled Coulomb collisions" as the paper says, so we'll see how Q actually turns out

Kirtley's mentioned a goal of recovering up to 95%

yes this is a bit unclear... Kirtley has said they demonstrated 95% recapture of pulse energy in smaller machines, but I got the impression he meant they're doing that to prove they can do what's really important: capture 95% of the fusion output power (which is up to 5 times greater)

it doesn't hurt to recapture as much of the leftover pulse energy as they can, but it's only marginally beneficial relative to capturing the fusion power

[TallDave]

(theoretically Q-eng>1 should be possible even with Q-sci<1)

at one point I also thought this might be true, but then I decided it can't be because then there's nowhere for the losses to come from (as per above)

i.e., if per Kirtley's graph you have to spend 10MW in losses to get 50MW in output, then you cannot also recover 9MW of that 10MW back through the magnets

you can maybe recover some fraction of the brem (there are concepts for doing this), but presumably not the transport (and note brem falls with KeV while transport rises)

[TallDave]

(sorry, had to break these replies up, forum not allowing as one)

From what I've watched and read, Polaris aims for one pulse per second

interesting, that makes sense as the website says Polaris will do 1+ pulse per ten seconds (way up from Trenta's 625 seconds)... of course with 1 ms at 1 Hz we're still not getting near a practical utilization factor at 1/1,000th, but then Polaris is only a prototype... one net-power pulse per second would still be world-changing, let's hope they get there

main goal for Polaris seems to be getting into that >20KeV region

[RERT]

So 1Hz, circa 50 MJ in, say 55 MJ out, is 5MW.

Is that right?

Avenues to higher power out are increasing the pulse rate - maybe with additional capacitor banks - or increasing Q, or increasing pulse size (power in at constant Q).

As Talldave alludes to, increasing pulse rate does seem a very attractive option, though the issues they might encounter are a complete closed book to me.

[Skipjack]

pulse length in Polaris is somewhere around 1 ms

Plasma duration at point of compression in the device center is one time, which I suspect SJ is saying is ~1ms, then there is the magnetic field sequential firing to accelerate the plasma from each end to collide & to be further compressed, and that is extremely short comparitively (less than a microsecond?). Dave was trying to calculate the energy expended to do this plasma formation, acceleration, & compression. I cannot see how Dave would be able to calculate this.

The acceleration and merging times are insignificant, on the order of a hand full of microseconds.

[Skipjack]

(theoretically Q-eng>1 should be possible even with Q-sci<1)

at one point I also thought this might be true, but then I decided it can't be because then there's nowhere for the losses to come from (as per above)

i.e., if per Kirtley's graph you have to spend 10MW in losses to get 50MW in output, then you cannot also recover 9MW of that 10MW back through the magnets

you can maybe recover some fraction of the brem (there are concepts for doing this), but presumably not the transport (and note brem falls with KeV while transport rises)

It is a theoretical possibility, not necessarily a practical one. They are very sure that they can recapture 95% of the input energy from the magnetic field.
Fusion energy is obviously a mixed bag because some of the energy is lost to neutrons and X-rays, though there are ways to potentially recapture the latter directly. All that of course also depends on the chosen fuel. D-He3 releases all of the products as charged particles. D-D has a larger percentage in neutrons and with D-T most of the energy is in neutrons. That said, they can even extract energy from the Alphas with D-T and it might even be easier to get net electricity from D-T due to the lower requirements.
Recapturing energy from gas is less efficient than the magnetic field. How much less is anybody's guess.
There are other things to consider. In one of his presentations (I believe the Princeton one), David mentioned that with a pulsed system assuming that all of the losses in the graph are lost to the system, is a bad assumption to make.

[Skipjack]

https://www.spglobal.com/commodityinsig ... -by-summer

This does confirm that the final assembly of Polaris has shifted to the right by a few months and is now scheduled for summer.

[mvanwink5]

The article conflates Polaris starting operation with operation to produce net electric power by mid 2024. I am happy to hear Helion's progress towards operation of Polaris. So is this schedule improvement or delay, I cannot tell; if it is a delay, any idea the cause?

[charliem]

Plasma duration at point of compression in the device center is one time, which I suspect SJ is saying is ~1ms, then there is the magnetic field sequential firing to accelerate the plasma from each end to collide & to be further compressed, and that is extremely short comparitively (less than a microsecond?).

I don't know about merging time, but acceleration time doesn't look too hard to estimate (roughly).

Kirtley's said that, in Trenta, both FRC's speed, just before collision, was 300 km/s (far from relativistic), and we know that this happens in a distance of a few meters from each formation chamber. If we assume constant acceleration, then t= 2e/V (where e=traject length and V=final speed). Let's say e=7.5 meters for simplicity, and t = 2*7.5/3*10^5 = 5*10^-5. Fifty microseconds.

Dave was trying to calculate the energy expended to do this plasma formation, acceleration, & compression. I cannot see how Dave would be able to calculate this.

Regarding energy, I think there are a few things we could try:

We know the temperature of the plasma at various stages in the process (from Helion's own words, and from previous papers), and we can make guesses as to the amount of gas per pulse (I heard that in Trenta that was a fraction of a milligram, so for Polaris, maybe, one to a few milligrams? ). Put those together and we'd have a first assessment.

Or we can take on the problem from the other side: Add the energy density of a magnetic field (part of Dave's signature) and a volume estimate, and derive total energy. Given that these machines' beta is almost one, that'd approximate total plasma energy.

Also, for a Deuteron, average speed of 300 km/s translates to 1.9 keV of energy. For a He3 it is 2.8 keV, and for an electron it's negligible.

And, the kinetic energy of 1 mg of mass travelling at 300 km/s is 45 kJ.

I still don't know how to put all this together to do what Dave is trying to do, but I sense a path.

[TallDave]

HERT:

So 1Hz, circa 50 MJ in, say 55 MJ out, is 5MW.

yes but unfortunately it's only 5MW for 1ms, which is not very useful, practically speaking, as it would only provide 5KW over the 1 second, or enough to power an average home (although again I have to emphasize this is still 50 years ahead of major programs like ITER and NIF)

since actual usable power is always in watts over time (e.g. kilowatt hours) the recovered fusion power required to maintain a given power plant output is the reciprocal of the reactor utilization factor

my guess would be Helion hopes the next-generation pulse rate will be as fast as 10Hz and the fusing plasma time as long as 10ms, which gives them a utilization of 10% meaning a 100MW power plant would require 1GW recovered power (seems plausible), although both of these obviously represent some challenges