Steps towards FF-1 Feasibility

[Joeviocoe]

asymmetric_implosion wrote: I have hope for the anode because you can use the physics of a highly directional beam to help you. The switches are the problem. I guess it is good for creating jobs and manufacturing to replace switches every few days if not every day. The switches are going to require a heavy investment to realize a 5 MW power plant operating at 200 Hz for more than a few days.

Can’t reliable switching (for greater than 2 months) be achieved by a greater number of capacitors and switch… at the sacrifice of the size of the whole system? Instead of 12, 48 capacitors for instance… or even 120, or 240 capacitors?

[asymmetric_implosion]

Dividing up the parts is possible, but you still have finite part lifetimes. Two months of operation is ~1E9 shots.

A common trick in rep-rate pulse power is to use 100 kV caps for a 40 kV application. The reversal (opposite polarity ring in PF) is a key factor in determining the capacitor lifetime. The reversal is usually quoted as a percentage of the maximum voltage. Typical high voltage, high current cap lifetimes with <10% reversal are 1E7-1E8 for GA Series S caps (check out the General atomics high voltage website). Larger capacitors are more like 1E5-1E6 pulse lifetime (GA Series C). This limits you to ~ 1 day operation at 200 Hz with ~1E7 shots per day. Ceramic capacitors can last longer (~1E8) but they are limited to sizes of 10 nF so you will need hundreds or thousands of units. They are also limited to 50 kV. I forgot that high current cap lifetimes were so miserable.

Switches are a bit more complicated. Switches are typically limited by erosion of the electrodes. You can operate a switch below the theoretical current (really a Coulomb limit as the eroded mass is determined by the total coulombs that pass through it) but typically you can only derate by 2-3X which buys a 2-3X increase in lifetime. A typical spark gap switch is good for 1E6-5E6 pulses. Spark gaps are the basis for most high current switches. Thyratron switches are good to 5E7 shots but they can only carry about 10 kA while the spark gap can carry 100 kA with ease. Railgap switches are the spark gaps big brother with multiple conducting channels instead of a single channel and they are limited to ~1E6 shots (Check out Perkin Elmer high voltage and L-3 communication electronic devices). New switches are underdevelopment but the incentive to go beyond 1E6 or 1E7 shots for a high current, high voltage system is very low. Cutting edge high current machines (>1MA) fire less than 1000 shots per year with very few exceptions. Fusion concepts like MAGLIF embrace the pulse power limitations by asking how they can maximize the fusion gain per shot to limit the number of shots. If it works as conceived, MAGLIF would only fire one shot every 10 s and operate at 1000 MW electric. With only ~9000 shots per day you can operate for a year before you have to work on the switches. Once a year shutdown for maintenance is comfortable to utilities that will likely supply the power. Even at 5E7 shots for a thyratron, you are limited to less than 3 days. With a spark gap, you are operating for at most 7 hrs. Pretty miserable having to replace switches 3-4 times a day.

Another problem with these switches is the realistic rep rate. Spark gap type switches are limited to ~10 Hz operation. Active cooling, flowing gas and other needs are likely required. Thyratrons operate up to 1 kHz but they need external heating at 300 W or more per switch to work. Take a 2 MA PF operating at 70 kV (reasonable thyratron limit). You need 200 switches to accommodate the current with no room to grow. That is an additional 60 kW of electrical power to operate the switch. Then you need to trigger 200 switches at low jitter (<10 ns). Each thyratron has a trigger board which needs to be timed. Once timed the switches are pretty good. It might be difficult to find a bad switch/cap module and one is enough to drag you down.

One might argue that solid state technology could be force fit into this problem with the advantage of >1E9 shot lifetime. I believe that a 2MA, 70 kV bank can be built but the actual lifetime and the cost will likely be a problem. Solid state costs something like $1-10/Joule stored at this level. Gas switch technology is more like $0.01-0.1/Joule stored without any effort. A company called SRL built an 80 Hz, 260 kA PF device. It took a true pulse power genius (Rod Petr) to design the bank to operate a 8 kV. The next immediate question is why not operate at low voltage if it is easier? In a plasma focus there is a 10-20 mOhm loss due to the moving plasma that cannot be avoided. To overcome this impedance means a minimum of 20 kV to drive 2 MA. Account for the other losses in the switches, capacitors and bus bars and you are at 40 kV to drive 2 MA. If I am correct, FoFu-1 is at ~1 MA and 40 kV. Increasing the cap bank size does not help as you still cannot get around the 10-20 mOhm impedance in the moving plasma; it simply reduces the impedance of the bank. It will take a great deal of resources to have a pulse power system that operates for 2 months at the 2 MA level without maintenance. It might not be possible with available materials.

[asymmetric_implosion]

Ran out of space….

The next solution is hot swapping the pulse power system. That is a great deal of waste and sunk cost per unit. Spark gaps at the 20 kV, 25 kA level cost $1000 per unit in the 1-10 per purchase range. If the machine requires something like 20 switches at higher voltage, say 100 kV switch to be safe and 200 kA per switch, you are talking about something like $5000 per switch at 1-10 units. You need something like 100 switches per day so you could probably get them for 25% of the 1-10 unit cost or $1250 per switch. So each day, you need to buy $125,000 in switches. Caps cost about 25% of the switches so you spend $160K per day on the pulse power alone. Thyratrons are $3500 per unit and you need two hundred units every three days. Again, bulk discount to 25% of single unit you only need $60K per day in switches and another $30K in caps or ~$100K per day on the pulse parts alone. A tech to replace these components costs ~$100K per year including overhead like benefits, pension, etc or $272 per day. You will probably need 4-5 techs so they person cost is minimal. The change over time is a minimum of 1 hr per change at 1 change every three days so you get 71 on hours for 1 off hour. The off hour costs you $100K in parts and people. The cost of electricity in CA is $0.15 kW-hr as a sort of average. The reactor generates 5000 kW at steady state for 71 hrs out of 72 hrs possible. That is 360,000 kW-hrs or $54K. You need to more than double the cost of electricity to make this a cost effective venture as I neglected fuel gas, regulatory compliance, operator pay, facility overhead costs, etc with existing technology. This also neglects any supply chain problems, contracts to purchase parts and people to keep the parts flowing. One might argue that clean energy has other benefits so the gov’t will step in but in the end you are still paying for it. I will admit that one could do better than a 75% discount over single unit price but 10% or so is the best I can imagine. That buys you some space ($27 K per day) but you are barely breaking even all things considered.

[Joeviocoe]

asymmetric_implosion wrote:
One might argue that solid state technology could be force fit into this problem with the advantage of >1E9 shot lifetime. I believe that a 2MA, 70 kV bank can be built but the actual lifetime and the cost will likely be a problem. Solid state costs something like $1-10/Joule stored at this level. Gas switch technology is more like $0.01-0.1/Joule stored without any effort. A company called SRL built an 80 Hz, 260 kA PF device. It took a true pulse power genius (Rod Petr) to design the bank to operate a 8 kV. The next immediate question is why not operate at low voltage if it is easier? In a plasma focus there is a 10-20 mOhm loss due to the moving plasma that cannot be avoided. To overcome this impedance means a minimum of 20 kV to drive 2 MA. Account for the other losses in the switches, capacitors and bus bars and you are at 40 kV to drive 2 MA. If I am correct, FoFu-1 is at ~1 MA and 40 kV. Increasing the cap bank size does not help as you still cannot get around the 10-20 mOhm impedance in the moving plasma; it simply reduces the impedance of the bank. It will a great deal of resources to have a pulse power system that operates for 2 months at the 2 MA level without maintenance.

Okay, lets talk about solid state for a moment. Are you talking about solid-state switches or capacitors?
[em] I calculated 1 MJ stored in the capacitor bank for each shot … (50KV x 2MA = 1E11 watts for a 10 microsecond shot / 10^5 = 1 million watt-seconds / 3600 = 278 watt-hours = about 1 MegaJoule (MJ)[/em]
So that would be anywhere from $1mil to $10mil if that cost scaled linearly. *I would hope that at such high demand, manufacturers would build a lot of production capacity and drive down costs.
So that would ruin the price estimate that LPP currently has for the Focus Fusion device.

But even a $5 million dollar 1st generation 5MW fusion power generator ($1/watt) would still be hailed as a great achievement and still MUCH cheaper than the $30 billion dollar 1GW ($30/watt)…. IF THE CAPACITORS AND SWITCHES COULD LAST FOR YEARS.

Do solid state capacitors (Is there such a thing) have that potential, to last a REALLY long time?
Are there any non-capacitor methods to store electrical power that can absorb and release 100 MegaWatts of power in 10 microseconds with low jitter???

—————————————–

Okay, reading your second post. The economics certainly don’t favor any 5MW device that requires constant maintenance. Essentially, for Focus Fusion to be a viable economic venture… they will have to build a device that meets LPP’s claim of “lasting for 1 month” without any maintenance needs. And the replacement parts must be reasonable in price. Million dollar solid state switches and capacitors are not gonna cut it. There needs to be some serious development in the engineering.

If LPP can get away with only replacing electrodes every month, I think they could be much cheaper $/kilowatt-hour than even Coal.

Bottomline, this is a massive engineering problem that I look forward to seeing solutions for.. As an electrical engineering student, I might want to help tackle this myself 🙂
Hopefully, FoFu-1 will prove the feasibility of the physics and unleash some serious government funding. As LPP’s plan goes, a few hundred million dollars and maybe a thousand extra personnel, a few dozen extra engineers… let’s get it solved.

—————————————–

P.S. I asked about your background at the bottom of page 1… and would just like some context for all this knowledge you bring to these forums. Thanks.

[asymmetric_implosion]

I am referring to solid state switching in my comments about cost per J stored. It seems a bit odd to use statements like $1/Joule stored when talking about a switch but the pricing goes that way. The caps I mentioned are state of the art for high current applications. Smaller caps used in everyday electronics have ~1E11 cycles. It is a complicated matter why the lifetime is so short for high current but the numbers on the GA website are realistic. I’ve found out the hard way. (see page 2 of this thread about 1/3 the way down to see my interest in this.)

FYI: Caps are mainly solid state devices but some liquid is used in almost any high voltage cap.

There are other power storage methods but none are rep rate compatible. Flux compression generators (explosively driven devices) are one shot and motor type solutions take too long to brake to allow 200 Hz operation. Out there ideas are floated about superconducting magnetic energy storage but I don’t have much faith in them.

————–

The problem is materials with caps and switches. As an example, gas switches are basically arcs internally. The cathode of an arc erodes material to support the plasma. Gas inside the switch reduces the cathode erosion but some finite amount of metal is removed every shot. This metal is vaporized and redeposits around the switch. Eventually the walls get coated. How do you stop erosion and redeposition? Stopping erosion is impossible so it must be minimized. You pick a material like tungsten or Rhenium (very expensive) but it only buys you a factor of 2-3 in erosion rate. So if you can’t stop erosion, stop it from redepositing in “bad” places. You can try baffles and other tricks but these materials get coated. Once coated, they are partial conductors. The gap between the baffles and the electrodes is reduced thus the switch hold off voltage is substantially reduced. One might ask about operating a switch that does not arc. Ahhh, welcome to the 1930’s and the thyratron. It operates in a diffuse arc mode with little erosion compared to a spark gap but it only last 50X longer than a spark gap at best. I don’t know of a gas switch option that does not operate in some kind of arc or diffuse arc mode. Caps are another story. I believe the failure is driven by magnetic forces at discharge and holding static potential for “long” periods. The switches are going to be the problem moving forward. A number of smart people are working on better switches but lifetime is not in the cards right now. The effort is put into reducing inductance, increasing voltage hold off, transferring more Coulombs and jitter. These are the most important parameters to high current folks. For those that might not know, the Z-machine at Sandia had a problem with jitter of ~3 ns with a 90 ns current rise time. To solve the problem, 6 ft by 6 ft panels of acrylic were inserted into the switches (6 ft by 6ft). To punch through the panels requires a certain amount of energy. This process reduced the jitter to less than 1ns. The PF does not have this strict a jitter (~10 ns will probably be fine), but it gives you an idea of how large current switching is handled. For reference, Z fires something like 200 shots per year. Switches with 1 shot lifetime are more than enough for the cutting edge science done at Z.

Ahhh, the cheaper than coal argument. A little history…when fission was the new thing, business models were build around the electricity being free. This was after they demonstrated an actual power generating system. I hope the day comes when fusion is cost competitive with coal. That would be a huge victory. Don’t get me wrong, I like the idea of clean fusion energy. The problem is I’ve been disappointed so many times that I try to approach fusion and other problems as a skeptic and ask questions. I think the LPP approach is as viable as any other fusion concept at this point. I hope they make the breakthrough but it will be a long road after that to a working reactor.

[Joeviocoe]

asymmetric_implosion wrote: How do you stop erosion and redeposition? Stopping erosion is impossible so it must be minimized.

I think the welding industry has a solution that may work. Feed wire. If you have a materials problem, sometimes mechanical engineering can solve it. Let it erode away, provide a flow of gases to carry away deposits before they solidify, and shape the electrodes so that they can be fed in at a constant rate. I am not sure of current switch design though.

There may be many design opportunities that present themselves. Solutions that were never thought of before since they have lacked any real incentives.

[Joeviocoe]

Right now, I think LPP is focusing on getting reliable power output from the switches and capacitors for about 1 shot per every few hours.. and getting the jitter down. I don’t think they will be focusing too hard on getting cycle rates up to 200 Hz and the cycle life up to trillions of shots before replacement… until the feasibility of the physics is established. Are the other PF devices trying to redesign switches and caps for this purpose yet?

I know Eric mentioned getting new switches from Raytheon… but I don’t know what improvements he hopes for.
Eric? and elaboration on that?

[Joeviocoe]

asymmetric_implosion wrote:
Ahhh, the cheaper than coal argument. A little history…when fission was the new thing, business models were build around the electricity being free. This was after they demonstrated an actual power generating system. I hope the day comes when fusion is cost competitive with coal. That would be a huge victory. Don’t get me wrong, I like the idea of clean fusion energy. The problem is I’ve been disappointed so many times that I try to approach fusion and other problems as a skeptic and ask questions. I think the LPP approach is as viable as any other fusion concept at this point. I hope they make the breakthrough but it will be a long road after that to a working reactor.

For that, I love this axiom:

“Chindia” price: the price at which India and China would adapt a technology for
economic reasons. “Everything’s a toy until it reaches that point” –Vinod Khosla

The over-promise of fission was fuel in, waste out. Neither of which could truly be mitigated by technology. They were bounded by the physics. Fissile material must be mined and refined, could be proliferated, and is radioactive to handle. That is costly, and is inherent in the physics of the fuel. Same with the waste. Not much can be done about it.

Fusion, aneutronic especially, may be costly to build, but technology can advance enough to cheapen the process… since fuel input is inherently cheap, clean, abundant and safe.. and waste output is just heat and helium. Delicious!

[asymmetric_implosion]

The switch problem is not that the electrodes erode away. That is a problem for the anode in the PF. The switch problem is the materials erode leading to a plasma/vapor of that metal. When the current turns off, the plasma/vapor expand and coat all surfaces. After some number of shots, a thin metal film builds up on the insulating ceramics or plastics that normally hold off the high voltage. The finite conductivity can lead to problems with voltage hold off, excessive leakage current in the switch. These conditions can lead to pre-fire and large jitter. Pre-fire and large jitter were the reason LPP is changing switches. The switches by R.E. Beverly had a poor materials choice internally. Scroll back through the photos on the switches.

I think LPP is on the right track with a physics demo but to do a complicated physics demo you need the correct equipment. FoFu-1 has been hampered by switch problems since its inception. The Raytheon switches, if they are the one I know of, will improve reliability dramatically. They are used on a 1 MA pulse power system that fires at 1 Hz and holds off 200 kV. Look up the linear transformer driver if you want to know more. I’m guessing these are the little brothers but they are good switches. They are talked about at many conferences. A few folks I know use them and they frequently comment on the high quality of the switches.

To my knowledge only a few groups operate their PF device at more than 1 Hz. The SRL group operated a PF using solid state up to 80 Hz at 260 kA. The NTU/NIE group in Singapore can operate two devices at up to 10 Hz. A few others operate at 1 Hz and above (~80 kA and ~300 kA) but they are rare. I’ve heard rumors of an Italian PF at 1 MA and 1 Hz but I haven’t come across any data. They were interested in radioisotope production so they would publish in journals I don’t frequent. I haven’t run across any talks at meetings I usually attend. The group I work in is using thyratron switches for 10 Hz, 60 kA and a type of spark gap for 0.25 MA, 1 Hz. Very few people have interests above ~10 Hz. The beauty of the PF and Z-pinches for most folks is the strong non-linear scaling in yield with current.

Many problems in fission can be mitigated. After fuel is burned it can be reprocessed extracting the useful components and burning the other products in an accelerator. Proliferation is a concern but thorium cycles can by-pass many of those concerns. Ask the Indians about whether they are investing in fusion or thorium fission. The Chinese have a more diverse picture but they are investing in tokameks for their fusion program.

[Joeviocoe]

asymmetric_implosion wrote: The switch problem is the materials erode leading to a plasma/vapor of that metal. When the current turns off, the plasma/vapor expand and coat all surfaces. After some number of shots, a thin metal film builds up on the insulating ceramics or plastics that normally hold off the high voltage. The finite conductivity can lead to problems with voltage hold off, excessive leakage current in the switch. These conditions can lead to pre-fire and large jitter.

That is why I mentioned a gas flow system that keeps the material vapor from coating the switch internals. The flow gas must be inert and kept hot to prevent solidification. But that should be doable.

asymmetric_implosion wrote:
Many problems in fission can be mitigated. After fuel is burned it can be reprocessed extracting the useful components and burning the other products in an accelerator. Proliferation is a concern but thorium cycles can by-pass many of those concerns. Ask the Indians about whether they are investing in fusion or thorium fission. The Chinese have a more diverse picture but they are investing in tokameks for their fusion program.

I certainly like the diverse approach to energy that the world has in general. I wish the U.S. was as bold sometimes.
LFTRs and even Travelling Wave reactors are important to fund and develop regardless of any progress on the fusion front.
I was just remarking how improper it was to regard fission as an energy silver bullet when it was the ‘new thing’ on the block.
Focus Fusion may be the holy grail, but that just means it tackles many issues at once. But it won’t save the planet without help. Different reactors for larger marine propulsion or large centralized power needs. Stacking 100 DPFs seems impractical. A 100MW polywell might serve better.

[delt0r]

If you could get a fast opening switch with larger than 40kV standoff, we would all be using magnetic energy storage. You don’t even need superconductors. At 1T you can store 400kJ per m^3 which is *easy* (this is without magnetic materials). With materials you get much higher densities. It would be like a switched mode power supply on steroids. You can get high voltages easily and high currents with some effort.

Catch is opening switches like this just don’t exist. If we move to solid state, say diamond or SiC switches that don’t just np junctions (avoids the voltage drop problem), then perhaps something could be developed.

[Joeviocoe]

delt0r wrote: If you could get a fast opening switch with larger than 40kV standoff, we would all be using magnetic energy storage. You don’t even need superconductors. At 1T you can store 400kJ per m^3 which is *easy* (this is without magnetic materials). With materials you get much higher densities. It would be like a switched mode power supply on steroids. You can get high voltages easily and high currents with some effort.

Catch is opening switches like this just don’t exist. If we move to solid state, say diamond or SiC switches that don’t just np junctions (avoids the voltage drop problem), then perhaps something could be developed.

Could you give an example of “magnetic energy storage” in use today? Is 1 Tesla common for this?

[delt0r]

Well no i can’t *because* there are no fast opening switches. In small scale well yea, every switch mode power supply uses inductors and the main energy store which makes them far more compact than a transformer/ripple cap combination. But the physics is straight forward and power density is pretty good compared to caps. Current at the MA level is a little more challenging without superconductors. But then MA is not trivial with caps either.

But if you can get a solid state switch with high current and high standoff. There is a strong case to use magnetic storage rather than caps.

[asymmetric_implosion]

Joeviocoe wrote:

The switch problem is the materials erode leading to a plasma/vapor of that metal. When the current turns off, the plasma/vapor expand and coat all surfaces. After some number of shots, a thin metal film builds up on the insulating ceramics or plastics that normally hold off the high voltage. The finite conductivity can lead to problems with voltage hold off, excessive leakage current in the switch. These conditions can lead to pre-fire and large jitter.

That is why I mentioned a gas flow system that keeps the material vapor from coating the switch internals. The flow gas must be inert and kept hot to prevent solidification. But that should be doable.

asymmetric_implosion wrote:
Many problems in fission can be mitigated. After fuel is burned it can be reprocessed extracting the useful components and burning the other products in an accelerator. Proliferation is a concern but thorium cycles can by-pass many of those concerns. Ask the Indians about whether they are investing in fusion or thorium fission. The Chinese have a more diverse picture but they are investing in tokameks for their fusion program.

I certainly like the diverse approach to energy that the world has in general. I wish the U.S. was as bold sometimes.
LFTRs and even Travelling Wave reactors are important to fund and develop regardless of any progress on the fusion front.
I was just remarking how improper it was to regard fission as an energy silver bullet when it was the ‘new thing’ on the block.
Focus Fusion may be the holy grail, but that just means it tackles many issues at once. But it won’t save the planet without help. Different reactors for larger marine propulsion or large centralized power needs. Stacking 100 DPFs seems impractical. A 100MW polywell might serve better.

Gas flow is used in these switches and they still have only 5E6 lifetime in the case of spark gaps. Thyratrons cannot flow gas and work correctly. There are other components in the switch I didn’t mention that are required for operation and stagnant gas is a must.

I think the US has a pretty diverse energy investment. Coal is king but we use natural gas, fission, solar, hydro and wind. DOE is supporting all these efforts via numerous funding vehicles and incentive programs. Nuclear is a tough sell in the US for a number of reasons. The main reason is legal. It only takes a few folks to file legal action which can shutdown a plant start up. The Nuclear Regulatory Commission is hearing license requests and new design approvals at a high rate these days and every one is heavily challenged. It takes years to get approval for a design let alone a site license. On the nuclear front, we are operating Gen II reactors in the US. Gen III reactors are going on-line in China while Gen IV reactors like traveling wave and thorium reactors could be licensed before Gen III reactors are built.

In my opinion the Polywell has a longer road to fusion than the PF. There are two sides in the Polywell argument going on in literature and from a pure physics standpoint the Polywell seems difficult. You rely on beam generation which is easily demonstrated at low density. As you increase density, which is required to achieve gain, Debye lengths and plasma fluid effects kick in limiting your ability to produce beams. The PF produces beams in a dense plasma because of it’s instabilities. The Polywell is a steady state device that is unlikely to see the same instabilities.

[asymmetric_implosion]

delt0r wrote: Well no i can’t *because* there are no fast opening switches. In small scale well yea, every switch mode power supply uses inductors and the main energy store which makes them far more compact than a transformer/ripple cap combination. But the physics is straight forward and power density is pretty good compared to caps. Current at the MA level is a little more challenging without superconductors. But then MA is not trivial with caps either.

But if you can get a solid state switch with high current and high standoff. There is a strong case to use magnetic storage rather than caps.

A large slow inductor will not be able to efficiently feed a fast pulse system like a plasma focus. It is better suited for tokameks. Over a decade of effort was put into the so-called plasma opening switch. The idea was to charge an inductive energy store and switch it to the load. The problem with inductors and fast pulses is the coupling. Plasma opening switches have an efficiency limit of 25%. The switch limits the efficiency to 25% when the switch is perfect. Marx technology has an efficiency limit of 33%. Regular RLC banks can do better depending on the match to the load.

Diamond is unobtainable as a practical switch at the moment. The material is mature enough to be used as a switch but the trigger is a problem. Diamond cannot be doped n I believe so you need to trigger it with UV photons, e-beams ,etc. The switch is only on for as long as the photons are present when operating at high voltage stand off. You need a high power UV laser below 200 nm, a flash lamp or particle beam system. The military invested heavily in diamond and it fell on its sword. For someone that’s saying I’m being negative, again, I’m speaking from experience. My employer worked on diamond switching for a decade or more. The trigger is the show stopper right now. A laser could be built to do the necessary triggering but it would be high power and high rep rate. Not a good combination.

SiC holds a great deal of promise but I don’t think it will reach the necessary voltage without stacking the switches in series which is the problem with Si switches. When you run switches in series you run the risk of jitter and the voltage showing up across one switch. A 20 kV switch does not like having 40 kV across it for long. This is the problem of solid state. Gas switches self break above their rated voltage and recover. Gas has not organization to it so who cares if it arcs. In fact, the switches use arcs to carry the current. Solid state switches arc under self break and solid materials don’t recover. There are tricks to avoid this problem but it takes someone steeped in solid state pulse power technology to make it work. Even the masters stay away from high voltage, high current because of the risk and cost.

[Author]

Posts