MTF Illustration

Point out news stories, on the net or in mainstream media, related to polywell fusion.

Moderators: tonybarry, MSimon

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Post by MSimon »

rnebel wrote:This is basically LINUS revisited. LINUS was an experiment done at NRL in the 70s where a plasma was imploded with a liquid wall. I believe the plasma was a Z-Pinch. The idea behind MTF plasmas is that the magnetic fields provide thermal insulation, but not confinement forces which come from the liner, or in this case an imploding wall. I believe that they are using FRCs as the target plasma. I don't know where they are with the program, but in my opinion this is an approach which should be taken seriously.
The mechanical problems are daunting. Getting 200 steam pistons operating together to within 100 nS of each other (mechanical tolerances not given nor length of motion) to provide 1 uS of confinement is no easy task.

Mechanical resonances alone may make it impossible. Friction and wear variables complicate things.

Here is some resource material from 1999.

http://fusionenergy.lanl.gov/Documents/ ... May_99.pdf

A 30 year time line and a billion dollar cost.

Another paper from 1994:

http://www.fas.org/sgp/othergov/doe/lan ... 405980.pdf

One of the things to consider is the speed of sound in steam (it determines the rate at which forces can be applied). At room temp and pressure you are looking at ~340 m/S for air. The upper range for a detonation wave in high explosives is 10,000 m/S. Speed of sound in Steel is 6,000 m/S. Diamond 12,000 m/S. Speed of sound in water is ~1,400 m/S. In mercury about the same. I would expect a lead/lithium mixture around that range i.e 500 m/S to 3,000 m/S. So that gives some idea of the speed of a pressure wave through the material.

And then you have to maintain an steam inlet pressure constant with all the valves pulsing. Then conduct all the exhaust properly away from the machine to prevent differential heating.

I believe electromagnetic field approaches show more promise. At least from an engineering standpoint.
Engineering is the art of making what you want from what you can get at a profit.

bwang
Posts: 9
Joined: Mon Nov 10, 2008 6:11 am

How General Fusion is planning to make the timing work

Post by bwang »

Because of the high accuracy of the impact timing of the numerous pistons (~1 μs), an electric means of controlling the exact piston trajectory is required. But this system can control only a few % of the piston energy. In particular, it can be a braking only system not requiring any high electrical power components. The pistons are sent a few percent above the required velocity and a servo loop applies just the required breaking to adjust the impact time and velocity. The US patent application #11/072,963 describes such a system in more details

MirariNefas
Posts: 354
Joined: Thu Oct 09, 2008 3:57 am

Post by MirariNefas »

So, how long until they build a full speed, full scale one?

djolds1
Posts: 1296
Joined: Fri Jul 13, 2007 8:03 am

Post by djolds1 »

Skipjack wrote:I think I saw someone else asking this before:
why so many pistons? One should assume that you can shape the shockwaves in a different fashion, especially in a liquid?
Maybe I am missing something...
Good point. This design is reminiscent of the Gadget, which had quite a few external detonators. Within 12 years the nuc labs got that down to 2 point implosion. Shouldn't the same be possible here?
MSimon wrote:One of the things to consider is the speed of sound in steam (it determines the rate at which forces can be applied). At room temp and pressure you are looking at ~340 m/S for air. The upper range for a detonation wave in high explosives is 10,000 m/S. Speed of sound in Steel is 6,000 m/S. Diamond 12,000 m/S. Speed of sound in water is ~1,400 m/S. In mercury about the same. I would expect a lead/lithium mixture around that range i.e 500 m/S to 3,000 m/S. So that gives some idea of the speed of a pressure wave through the material.

And then you have to maintain an steam inlet pressure constant with all the valves pulsing. Then conduct all the exhaust properly away from the machine to prevent differential heating.

I believe electromagnetic field approaches show more promise. At least from an engineering standpoint.
Do the rams need to respond in the usec range, or is it that the waves need to converge on the cavity with that timing? Any chance of creating a mechanical/acoustic analogue of the "low density and high density explosive lenses" used in the Gadget and Fat Man? Rams fire at slightly different times but the acoustic waves arrive at the cavity within the microsecond tolerance, perhaps using acoustic interference to create the "final" incoming wave pattern?

Duane
Vae Victis

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Re: How General Fusion is planning to make the timing work

Post by MSimon »

bwang wrote:Because of the high accuracy of the impact timing of the numerous pistons (~1 μs), an electric means of controlling the exact piston trajectory is required. But this system can control only a few % of the piston energy. In particular, it can be a braking only system not requiring any high electrical power components. The pistons are sent a few percent above the required velocity and a servo loop applies just the required breaking to adjust the impact time and velocity. The US patent application #11/072,963 describes such a system in more details
Yeah. No doubt it can be done - maybe for 5 servo loops. Possibly 10. But 200? What about drift due to temperature variations? Friction changes? Control loop jitter? Steam valve jitter? etc.

And let me mention just one fly in the ointment. You can have a servo that controls timing. Or you can have a servo that controls velocity. Choose one. Because they are mutually exclusive. OK maybe you do a dual loop. One controls inlet pressure to get the velocity right. One controls braking to get the timing right. To make the system stable the timing loop is going to have to respond a lot faster than the velocity loop. Usually 10X is the stability criteria. So you want overall control to 1 uS. That means a 100 nSec velocity loop and a 10 nS timing loop. the 10 nS timing loop dictates that the servo be within 1 nS distance of the feedback signals. Rule of thumb: 1 nS = 1 foot.

If this was easy it would already have been done.

Let me just say that I have had a little experience with servo loops. I have designed auto-tuning servo loops. Phase locked loops (another kind of servo). Co-ordinated servo loops.

Take a constant tension paper winder. As the paper roll gets bigger you have to vary the rotational speed of the winder. To keep the tension constant you actually have a loop in the paper (called a dancer) held at constant tension by the tensioner loop and the loop in the paper is allowed to move up and down (within limits) to allow for small variations in speed in the paper web. Which is why the loop is called a dancer.

Some nice pictures. Scroll down:

http://www.drivesourceusa.com/pdfs/Tens ... Winder.pdf

The short version: to control linear speed AND tension you need buffers. OK. Where are the buffers in the MTF system?
Engineering is the art of making what you want from what you can get at a profit.

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Post by MSimon »

Let us look at the Polywell from a control stand point.

Let us start with D-D because it is easy and add POPS because it may help.

What are the servo loops?

1. Gas pressure - with 1 second of gas inventory in the reactor the loop needs a response time of 100 mS and the actual gas valve should come in at 10mS. Not too tough.

2. High voltage 1 KHz (3db) response for small signal variations. 0 to full voltage in 100 mS or better. Voltage held to .1% or better. Noise 80 db down or better. Not too tough. Might be able to increase those specs by a factor of 10 each with very careful design.

3. POPS waveform (100KHz in a full scale machine) to 1% accuracy from desired (distortion 40 db down). Timing jitter to 10 nS or better non cumulative. Timing done by feedback from the actual beam. So the actual process controls its own timing. Not too tough. Might be able to increase those those specs by 3X to 10X with very careful design and tricks (like two A to Ds each operated with a slight timing offset and summed - driving the power amplifier for 2X improvement in waveform or four A to Ds to get a 4X improvement in waveform)

==

Now things get trickier if it becomes necessary to hold the pB11 reaction resonance with the DC drive voltage. But I haven't used all my tricks yet (like a separate pilot frequency to control DC drive and phase detectors to demodulate it)

All in all I'd rather control a BFR than an MTF.
Engineering is the art of making what you want from what you can get at a profit.

bwang
Posts: 9
Joined: Mon Nov 10, 2008 6:11 am

Jan 2009 Popular Science has several pages on General Fusion

Post by bwang »

The latest issue of Popular Science has several pages on General Fusion.

They indicated that they have raised $7 million of their $10 million second round which would fund a full scale version.

A prototype device with a tank diameter of ~2 m and an input energy of ~120 MJ could provide interesting fusion gain. Such a system could be built at a reasonable cost of ~10 $M in about 3 years. (For a prototype with low repetition rate, no tritium re-breeding, no heat exchanger and no turbo-generator).

So 2010-2011, a series of devices leading to that device gets made.

The third phase for them is $50 million for a net energy gain device with a target date of 2013 if the second/third phase are roughly on schedule.

Then they try to raise $300-500 million for commercialization.

the patent has 17 pages of text description of what they are doing and 14 pages of diagrams, flowcharts etc...

Definitely their plan has challenges but it could work and modified designs for more reliability are possible.

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Re: Jan 2009 Popular Science has several pages on General Fu

Post by MSimon »

bwang wrote:The latest issue of Popular Science has several pages on General Fusion.

They indicated that they have raised $7 million of their $10 million second round which would fund a full scale version.

A prototype device with a tank diameter of ~2 m and an input energy of ~120 MJ could provide interesting fusion gain. Such a system could be built at a reasonable cost of ~10 $M in about 3 years. (For a prototype with low repetition rate, no tritium re-breeding, no heat exchanger and no turbo-generator).

So 2010-2011, a series of devices leading to that device gets made.

The third phase for them is $50 million for a net energy gain device with a target date of 2013 if the second/third phase are roughly on schedule.

Then they try to raise $300-500 million for commercialization.

the patent has 17 pages of text description of what they are doing and 14 pages of diagrams, flowcharts etc...

Definitely their plan has challenges but it could work and modified designs for more reliability are possible.
Do you have a link to the patent?

OK here are some pdfs:

Computer Simulations:
http://www.generalfusion.com/files/acoustic_wave.pdf

Reactor Patent:
http://www.generalfusion.com/files/mtf_pat.pdf

Acoustic Generator Patent:
http://www.generalfusion.com/files/acou ... or_pat.pdf

Evidence of Fusion - includes some piston specs:
http://www.generalfusion.com/files/evidence.pdf

Note that this machine is a D-T only machine. D-D is highly unlikely and pBj is out of the question.

From:

http://www.generalfusion.com/documents.php

I'm studying the material. I will report back.
Engineering is the art of making what you want from what you can get at a profit.

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Post by MSimon »

Experimental set up:

A capacitor bank set up vaporizes aluminum spirals to set up shock wave the capacitor bank spark gap switches are triggered to within 5nS and the main variation in the pressure wave is the variation in construction of the aluminum spirals.

So no pistons.

===

When they go to pistons they will pressurize the pistons and use triggered latches to let them all go at the same time. The final velocity of the piston will be 100 m/S and the shock wave it induces will be traveling at 2.5 km/S. If we assume 100nS timing to assure reasonable addition of all the induced waves you get a distance variation at that speed of 10 um. Which is 2X what his measurement system can do. Good. And the measurement system will require laser diodes for the light source and gratings to do a linear encoder.

He plans to control the timing and velocity by repeated shots to adjust the parameters. Clumsy but OK.

Can it be made to work? There is an outside chance. Repeatedly and reliably? I doubt it. Wear is going to be a killer on the latches. He proposes to fix the piston/cylinder wear with an air bearing. The problem is that the space in front of the cylinder must be evacuated before every shot.

I'd want to see a dummy steam powered machine with 2 or 3 cylinders operating reliably for a day or two meeting all specifications for timing and velocity before I set up the full machine. Given that he expects to run the whole 200 cylinders on 100 Mj or so a 2 MW steam boiler ought to suffice for experiments.

I will say that he seems like a careful experimentalist so he might pull it off. He might even get the machine to function for a few shots. Continuous operation for a year?

Tthat would require a 2,000 year MTBF or better for each cylinder and its associated electronics for a reasonably good probability of 1 year of operation. that is roughly better than 20 million hour MTBF. That is space rated electronics parts and space rated assembly or better because the mechanical parts have to have a failure rate on that order as well.

Let us think in terms of an auto engine. 4,000 hour MTBF at 5,000 rpm. 8 cylinder. Say 10 billion piston strokes.

vs 200 pistons at 1 stroke per second at 3,600 strokes an hour at 24 hrs a day at 365 days a year. About 6.5 billion strokes. So the ball park is right. Except for one minor detail. The piston in the auto is not slamming into the cylinder head every time it reaches TDC. Its timing is constrained by the crank shaft.

Let us look at timing in the auto at 5,000 RPM it completes a revolution in 200 uS. To get the timing of the spark right it must be controlled to within 1/2 uS. (about 1 deg) . But in the auto you have rotational inertia helping keep things constant and feedback from gas sensors to tell you when the timing is right.

So I would not put making all this work outside the realm of possibility. It will not be easy. Not easy at all. And the MTBFs of every part in the cylinder assembly must be beyond space rated.
Engineering is the art of making what you want from what you can get at a profit.

djolds1
Posts: 1296
Joined: Fri Jul 13, 2007 8:03 am

Post by djolds1 »

MSimon wrote:So I would not put making all this work outside the realm of possibility. It will not be easy. Not easy at all. And the MTBFs of every part in the cylinder assembly must be beyond space rated.
Any chance of switching the design to a 2 point stroke?

Duane
Vae Victis

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Post by MSimon »

djolds1 wrote:
MSimon wrote:So I would not put making all this work outside the realm of possibility. It will not be easy. Not easy at all. And the MTBFs of every part in the cylinder assembly must be beyond space rated.
Any chance of switching the design to a 2 point stroke?

Duane
What is a two point stroke?
Engineering is the art of making what you want from what you can get at a profit.

djolds1
Posts: 1296
Joined: Fri Jul 13, 2007 8:03 am

Post by djolds1 »

Vae Victis

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Post by MSimon »

djolds1 wrote:
MSimon wrote:What is a two point stroke?
The General Fusion design is reminiscent of this:
http://upload.wikimedia.org/wikipedia/c ... imated.gif

I'm thinking of something like these:
http://en.wikipedia.org/wiki/File:Linea ... ematic.png
http://en.wikipedia.org/wiki/File:Swedi ... c_Bomb.png
http://en.wikipedia.org/wiki/File:Swan_ ... ission.png
The pistons are loaded to 2/3rds ultimate strength. So it may be a materials question for repetitive operation.
Engineering is the art of making what you want from what you can get at a profit.

jmc
Posts: 427
Joined: Fri Aug 31, 2007 9:16 am
Location: Ireland

Post by jmc »

When I first saw this approach my initial was: too many moving parts.

And yes I'm pretty sure our ability to diagnose what went on in the plasma in final machine will be precisely zero. I suppose there my be some possibility or developing analogous plasmas compressed in a less economic and repeatable manner which could be diagnosed though.

My instinct is the chances of success are slim.

But from a materials point of view I love it. Having the neutrons slowing down in a few metres of liquid lead-lithium is a brilliant solution to neutron damage which I am not aware is a possibility in any other fusion schemes so I think its worth giving it a go.

I still haven't been given a satisfactory answer wrt tokmak fusion to alay my fears of a coolant pipe burst in the blanket of a fusion reactor as a result of neutron damage.

bwang
Posts: 9
Joined: Mon Nov 10, 2008 6:11 am

Rapid maintenance and sufficient reliability

Post by bwang »

Thanks M Simon for the reliability analysis. I knew that was key but did not take the time to quantify it.

http://en.wikipedia.org/wiki/Mean_time_between_failure

MTBF is Mean Time Between Failures, which is described at the wikipedia link. The reliability of the pistons is a key aspect.

Although having a system for monitoring the pistons and system for wear and having an easy method to rapidly swap out pistons could maintain a sufficiently high operating reliability. Keeping 90% uptime throughout a year with an average of one piston or some other component replaced every day with a 15 minute hot swap would be doable. Quality and cost of components would need to be balanced against overall costs.

http://ntrs.nasa.gov/archive/nasa/casi. ... 015833.pdf

Nasa has stirling free piston cryocoolers with mean time between failure of over 500,000 hours. Aircraft engines wiith 40,000 hours of mean time between failure are common and microturbines have 14,000 hours of mean time time between failure. Twice weekly and even daily maintenance can be doable if the maintenance can be fast (using a robotic arm to pop a piston assembly out and replace it). Faulty units can be refurbished and put back into service if that keeps costs down without sacrificing overall reliability.

Post Reply