Dr. Bussard Passes Away, Polywell Fusion Continues 79
Vinz writes "Dr Bussard, the man behind the Bussard Collector and inventor of the Polywell fusion device, passed away last Sunday in the morning. He leaves behind him a legacy of EM fusion devices, and a team determined to continue his efforts. The news of funding extension for the construction of his WB-7 fusion devices made it to slashdot months ago (as well as his talk at google). They may be a serious candidate in the run to bring commercial fusion, and may work at lower scales than other projects. Let's hope the project continues in good shape despite his departure."
Electron losses (Score:5, Interesting)
Personally I think stellarators are more promising. For those who don't know stellarators are a bit like Tokamaks, except rather than relying on an electric current in the plasma to create the necessary twist to the magnetic field for confinement, they twist the confinement vessel itself ( a bit like a moebius strip ), making them a lot more stable than Tokamaks, and allowing them to operate continuously (You can't induce a DC current in the plasma so Tokamaks necessarily operate in pulses ). Main problem seems to be that since stellerators have a lot less symmetry than Tokamaks the calculations become more difficult, but if computing power continues to rise this will probably be solveable.
As a bonus stellarators look damn cool ; )
http://www.efda.org/pictures_html/stellarator_schema_and_live.jpg [efda.org]
http://www.psl.wisc.edu/hsx.jpg [wisc.edu]
Warp Factor 11 (Score:3, Interesting)
In principle, the Bussard ramjet avoids this problem by not carrying fuel with it. An ideal ramjet design could in principle accelerate indefinitely until its mechanism failed. Ignoring drag, a ship driven by such an engine could theoretically accelerate arbitrarily close to the velocity of light, and would be a very effective interstellar spacecraft.
So what would happen to people or computers travelling inside the ship?
Would they move forward through time at accelerated speed? or end up in deep-space oblivion?
Re:Electron losses (Score:4, Interesting)
Re:A remark captured my attention (Score:5, Interesting)
It's one of the things that had the alarm bells ringing about the Polywell because it's something you'd expect from someone who wants to sell you the Brooklyn Bridge.
But I think in his case he just saw the writing on the wall. He knew he wouldn't see a full-scale reactor if it was done step-by-step, he was just too old for that.
I really hope someone with the required expertise will take an honest look at the Polywell. The concept sounds good and the central question seems to be whether the plasma will move into thermal equilibrium or not. And the paper every critic cites is one master thesis written by the student of one of Bussard's rivals for Navy funding. Hmmmm...
Now, the fact that your opponent's not trustworthy doesn't mean that you are, but I think that considering all the money that goes into ITER a few million for looking at different approaches (mostly this and lasers/inertial confinement =) are a good investment.
Wee, this is very bad. (Score:1, Interesting)
Re:Electron losses (Score:5, Interesting)
From Maxwell's equations div B = 0, so magnetic field lines cannot suddenly stop, and thus magnetic fields alone cannot confine charged particles in a plasma which has the same topology as a sphere ( a charged particle that travels along a magnetic field line will escape the confinement ). Consequentially you WILL have electrons leaking out of the magnetic mirrors, and this effect will increase as the potential well height increases.
Tokamaks and Stellarators don't have this problem because they are topologically equivalent to a torus, and thus their magnetic field lines can completely enclose the plasma, while simultaneously not penetrating the plasma facing components.
There are further problems with the polywell design. As an example, even at optimal energy levels the reactants will fail to fusion in many of the collisions, and thus the ions will thermalise much quicker than they fuse. Bussard claimed he could avoid thermalisation of the ions, but this is simply not possible in the polywell design since it would require a spontaneous process to transfer energy between the ions in such a way that their overall entropy decreases. While the polywell is not a closed system, and thus not subject to the second law of thermodynamics, there is no meaningful energy input other than the initial potential energy of the ions, and thus for thermalisation to be avoided there would have to be a large entropy flow out of the plasma, and thus it would quickly cool to levels bellow that required for meaningful fusion. In short, you will rapidly get thermalisation of the ions, which in turn leads to X-ray losses from the electrons. If you did heat the plasma, by say injecting microwaves or neutral particle beams, it would still not avoid the problem of thermalisation unless you managed to selectively accelerate the low energy ions, while simultaneously slowing the fast ones ( and of course, if this energy exceeds the fusion power, as it will have to do in order to overcome the speed of thermalisation, then you won't get net energy out of the device ).
While we are at it, no, you are not going to produce a Boron plasma with any significant number density without getting electrons in it, just calculate the electrostatic force you would get on an electron outside the device from 1 mole of boron nuclei and you quickly see that this is absolutely impossible. Even if the proton/electron ratio is just 5/4, Q = N_a, so you are talking roughly 6*10^23 times the proton charge ( or 60 million Coulomb ).
You then have to take into consideration other problems, like sputtering of plasma facing compounds, giving impurities that cool the plasma ( and since all potential plasma facing compounds have Z numbers of 6 or above, this will further increase X-ray losses ). There is no proposed way to design a divertor, so the device could most likely not operate for extended periods of time.
Basically I don't see this getting a confinement time even close to that of a Tokamak or Stellarator. The number density will be dramatically less ( since it is limited by the height of the potential well ), and it just doesn't seem likely you will get even close to the lawson criterion. Granted, you don't need to achieve ignition in order to extract a lot of energy, but you won't get a high value of the nTtau triple product without raising T to very high energies, which impacts the amount of energy you can gain.
Makes perfect sense to me. (Score:5, Interesting)
In this case he believed he had the scaling laws down. With power proportional to the seventh power of the radius and energy gain proportional to the fifth power, you were only talking about building a device maybe 10 times the radius of the lab device. That's TINY as fusion experiments go, and also compared to fission plants. And the thing is basically a slightly gassy vacuum tube with some magnets in it, i.e. mostly empty space, very little material.
If there are any gotchas you'd have to scale it up about that much to find them. So why go halfway and then build a full-size one when, if it turns out there AREN'T any gotchas you've got an operating power plant on the next step?
His plan was to do two more small prototypes, to get some more solid data than his three-neutron final run and compare two geometries for the final deaign, then go for the gotchas-or-gold. If it works, it gets you to production right away and you didn't spend a dime on yet another intermediate prototype. If it doesn't, you're not out all that much more than if you built some intermediate size that was maybe big enough to find the gotchas.
Suppose there AREN'T any gotchas. Then we get to working fusion power years sooner. Ditto if there are gotchas that only show up at the scale between the intermediate prototype and the full-size design. In either case the time spent on the middle-size below-break-even prototype was wasted.
Baby steps are for people who get their money from researching and will be looking for a new job once things are actually working. Big steps are for people who want to get to the finish line.
Re:Electron losses (Score:3, Interesting)
Leslie Woods (modeling problems) (Score:1, Interesting)
I know it is a bit off topic, but seeing as we are discusing confinement, I would love to hear someone comment on Leslie Woods' (a late mathematician from Oxford) equations. Basically, from what I understand from the Nerenberg lecture he gave [apmaths.uwo.ca], he claimed that the majority of the gap between what the equations predict and what is observed in reality is not due predominantly to missing turbulence, but rather to a missing non-turbulence term.
Unfortunately, from what I understand of what he said, despite his corrected equations predicting very closely what has actually been observed in practise, and turbulence approaches not predicting anything very well beyond their degrees of freedom, the current community (or at least a key subset of it with a lot invested in the turbulence approaches) is uninterested in hearing anything more about it or allowing the work to be published. Apparently they forstall under such issues as demanding it be derived from Boltzmann's equations, despite his objections that this cannot be done due to the fundamental assumptions underlying Boltzmann's equations.
Personally, although I do not have the physics background to comment on the derivation, I must admit that, if the equations are indeed doing a much better job of matching up with what is being observed, I am quite bothered to hear this. I am reminded of such issues as those between Laplace and Fourier regarding the latter's (seminal) work on the Fourier series and the former's repeated objections to it or anything to do with it. Really, shouldn't the final test always be how well it does in the lab? I hate to think of all that great plasma engineering that is being help up over lack of ability to really model these phenomenons well in the lab.
Re:Read the Wiki Article (Score:3, Interesting)
Re:Electron losses (Score:5, Interesting)
Instead of producing lengthy expositions about the flaws of technologies that you don't understand, why don't you try learning about them instead? From your post, it is clear that you neither understand, nor have you read any of Dr. Bussard's papers on the subject. Given the topic of this story, you could at least have enough respect to do so, before spreading FUD about ideas.
First of all, no one is claiming that the divergence of a magnetic field is non-zero. The fact is, the "wiffle ball" trapping of electrons in a Polywell is more than adequate for the task. Electrons escaping through the cusps do not equate to losses, as they usually follow the field lines right back into the machine.
In any case, it is highly disingenuous to claim that a Tokamak has no difficulty confining a plasma. While the topology of a Tokamak (or a dipole as in the LDX) may be a better configuration for containing charged particles, this ignores the fact that the ions have a much greater mass. After a number of collisions, it is inevitable that they will smash into a wall. The only solution to this problem is to make the machine bigger, but it is still far from ideal.
Your calculations concerning a Boron plasma are complete nonsense; as described in his recent paper, only a slight deviation (1E-6) from neutrality is necessary to make a well nearly as deep as the drive energy.
Overall, there are at least as many, if not more challenges, in producing a commercially viable Tokamak. I won't discount either approach yet, but the Polywell certainly looks a lot more promising. A quasi-spherical potential well simply seems like a much better place for a sustainable fusion reaction than a divergence-less B field. Wether or not it works out, it certainly deserves more attention and less unfounded condemnation.
Re:Electron losses (Score:3, Interesting)
Then you've plainly not looked at Bussard's design on even a classical physics level, as he agreed with you. The magnets are held at a high positive potential and electrons are allowed to exit through the cusps and be drawn back in by the positive charge. It's an open re-circulating design and reduces electron losses farther than any mirrored design ever could.
Bussard claimed he could avoid thermalisation of the ions, but this is simply not possible in the polywell design since it would require a spontaneous process to transfer energy between the ions in such a way that their overall entropy decreases.
You mean a spontaneous process to transfer energy from high energy ions to low energy ions? Thermalization will do that for you. The key problem was Nevins suggested losses to a high energy tail that would form. Bussard modelled the thermalization at the edges(where ions spent the vast majority of their time) and found it slowed the maxwellianization long enough. Long enough just means longer than the average ion time to fusion.
Read Tau Zero for an extreme answer (spoilers) (Score:3, Interesting)
Re:Electron losses (Score:3, Interesting)
Well I have seen Bussard's alleged explanation for why he can maintain a non-maxwellian velocity distribution, and quite frankly, it can't work. He seems to argue that there is a spontaneous process in the system which restores the non-maxwellian distribution because the ions thermalise at low energies at the perimeter of the device. However:
a)Restoring a particle distribution to a non-maxwellian energy distribution requires work. It doesn't matter how you do it, if the energy distribution changes towards a more mono-energetic distribution, no matter what the mechanism is, work is required to account for the change in entropy.
b)Because of this a spontaneous process which restores the non-maxwellian velocity distribution must either drain the plasma of energy ( i.e you have large power losses ) or it will require a similarly large input of work.
c)In a p-B plasma you will have a lot of B-B collisions as well as p-p collisions, both of which have neglectable fusion rates and scatter the ions in all directions, and greatly spread their energy distribution. Furthermore, even at optimal energies the p-B collisions have a low fusion probability as compared to the probability for simple scattering.
These three assumptions is enough to calculate the minimum amount of power required to maintain a monoenergetic velocity distribution at a given collision rate and energy. Since the collision rate also determines the rate of fusion at a given ion energy, you thus have a given amount of work required to maintain a monoenergetic energy distribution at a given rate of fusion. To estimate if this is more or less than what is required to make it practical to maintain a mono-energetic velocity distribution, you only need to know the fusion cross section as compared to the scattering cross section, at the given energy. This is exactly what Rider did. The problem is simply that the fusion cross section, even at optimal energies, is very much lower than the cross section for scattering.
Re:Electron losses (Score:3, Interesting)
a)Restoring a particle distribution to a non-maxwellian energy distribution requires work. It doesn't matter how you do it...
And your missing the same thing Rider did. Thermalizing a distribution is just nature. Because of the inherent design of all IEC devices, ions spend >90% of their lifetime in the outer edge of the plasma. That means thermalization is working for maintaining a mono-energetic distribution 90% of the time. Energy isn't being lost, no work is being performed. Energy is just naturally being exchanged between low and high energy ions. Basic high school physics still underly plasma behaviours.
work is required to account for the change in entropy.
Only if your already assuming conditions...
Ions staying in the system naturally thermalize, on average higher energy ions slowing down and lower energy ions speeding up. NO WORK! Ions gaining enough energy from a collision in one pass(or close consecutive passes) to leave the system will be an energy loss. Thermalization works FOR the mono-energetic distribution because of the basic design. Ions spend most of their lives at low kinetic energy thermalizing. A thermalized, maxwellian distribution of ions in the outer edges gives a natural mono-energetic distribution in the core.
Dr. Bussard was so dismissive of Rider's thesis for the same reasons you are putting forward. IEC devices have not previously had a problem maintaining non-maxwellian distributions.