Sunday, December 10, 2017

Notes On The VLAB Fusion Event

Introduction:


On Tuesday November 7th 2017 a fusion panel discussion was held on Stanford’s campus.  The panel was organized by Virtual Labs, a non-for-profit organization in Silicon Valley. The goal of VLAB is connect entrepreneurs to technologists.  Former members of the VLAB group have included managers at some of the valleys’ premier technology companies.  The event was organized by Michelle Tsing, a lawyer and an entrepreneur.  The panel was moderated by Dr. Rachel Slaybaugh a member of nuclear engineering department at Berkeley.  The panel also included Dr. Michele Lebarge from General Fusion, Dr. Thomas McGuire from Lockheed Martin and Ray Rothrock, a former member of Venrock capital.  It also included Dr. Matt Thompson from TA Technologies (Tri alpha rebranded, more on this later).  I reached out to Michelle and she was gracious enough to share a 90 minute recording of the panel.  You can watch the event through facebook - here.  After watching this talk, I wrote this quick summary of the event.  Enjoy. 




Dr. Rachel Slaybaugh (Moderator):

Slaybaugh opened the discussion with a quick take on fusion. Her style was very positive - despite a somewhat tough crowd. One issue was the triple product Dr. Slaybaugh mentions at the start of the talk. Though it is a common metric used to measure fusion but I think it is a bad metric for a commercial plant. A better metric is the energy balance for a plasma-based approach. This equation was laid out by John Lawson in 1956. That eventually became the Lawson criteria, but I argue it is much more instructive to look at his original energy balance.


Each one of these terms points towards key design principal. Fusion rate has been the focus of the last several decades. We drove up fusion rates, with little consideration to other parameters like cost, efficiency and scale. Conduction is the loss of mass from the plasma (and the energy that leaves with it). Here, metal surfaces are the enemy -- and containment is the goal. I have seen chambers built out of glass to stave off conduction losses. Radiation is all the energy that leaves as light (X-ray, IR, UV, visible, ect). Not a lot of opportunity to innovate here – tuning plasma and light reflection are two ideas that have been kicking around for years. Finally, efficiency is probably the most unexplored topic in fusion research. I would put advances in superconductors, direct conversion and capacitors as all improvements in machine efficiency. Below is a chart on purposed fusion approaches. Note that this chart is (1) incomplete and (2) contains ideas that may not work.




Dr. Matthew Thompson (TA Technologies): 

Dr. Thompson was a very polished speaker. Of the group, his view point feels the most detached from science. He is not there to talk experimental physics; he wants to talk about electricity on the grid. It is as if the fusion problem has already been solved. That is some serious confidence from an organization that just rebranded. Why did Tri Alpha rebrand? Weirdly, it was to move out of the energy space. Tri Alpha has created some of the strongest particle beam sources in the world. The beams are very small and can create 1,000,000 electron-volt particles. Those specs alone to get the market interested – those beams are probably superior to the ones used on ITER. Each beam has two parts: a particle source and an accelerator. Tri Alpha also plans to use those particle sources and make a cancer product. Ions from those sources can be fused. This makes neutrons. Those neutrons can be used in cancer treatments.

This past week the CEO of Tri Alpha laid out a vision for this new product – one of which has already been sold. The product, a machine about the size and cost of an MRI machine, will be used in boron neutron capture therapy. Basically, boron is sent into cancer cells inside the human body. Neutrons are then passed through. The boron absorbs the neutrons and atomically splits. When it does, it acts like a tiny bomb, killing the cancer cell. This company can make these neutrons from a fusion reaction. By selling a commercial machine to do so TAE has likely become one of the first companies to connect fusion into the medical space. How much will this side effort affect Tri Alpha core mission? One cannot criticize the company for trying to turn a profit. Making money, makes the company permanent. Thompson finished his talk with a video on the history of the company. The images presented were a rare insight into company history. For the first 10+ years of this firm, everything was so secretive. They were notorious for this. The company did not even have a website for the first 10 years.


Dr. Thomas McGuire (Lockheed-Martin): 

Of all the groups, we know the very least about Lockheeds’ effort. Unlike everyone else, they have not published. Everything Tom McGuire said added some new knowledge to this approach. For example:

Size: The Lockheed team has now expanded to 25 people. This is a great sign. The results have likely been very promising.

Speed: McGuire estimated that his team iterates every year. Great. Only a small team could move that fast.

Fuel: Lockheed is planning to put tritium in their feed stock. This will add NRC and EPA regulations to their effort. Where I got my PhD it took a team of a couple dozen professionals, several million dollars and several years to create a tritium handling system.

Design: McGuire laid out a neutron blanket around the device. There are issues associated with a lead-lithium blanket and ironically General Fusion is the world leader in that technology. The blanket is baked into General Fusions’ approach.

Runtime: McGuire describes a machine that is self-sustaining: “…Like a fire, it will keep burning, but you need to continually add fuel...” Today, there are fusors that can do fusion like this. I am skeptical of this claim; it feels premature. In this article, I have written extensively on Lockheeds’ approach. It would be interesting to extend this analysis to look at runtime.

Future: When he looks forward, McGuire is envisioning an effort that would probably be too big for Lockheeds’ management. “…This is going to get bigger with each experiment, turning into an organization and technology similar to Tri Alpha Energy…” How would Lockheed manage such a large organization? Would they need to spin out this out into a new company?

McGuire has gotten much better at pitching. In May of last year, he spoke on a panel at the Milken Global Institute. I wrote a review of that panel discussion here. Having watched both, I can say his style has improved greatly. His answers are clearer and more succinct. I also liked his comparison between his cusp confined plasma and a mosh pit. Excellent comparison.


That said, it is painful to hear him talk about putting fusion an airplane. His reactor weight is estimated to be between 200 to 1000 metric tons, with its’ blanket. This means this reactor may in fact be heavier than a 747 airplane. How could any airplane lift such an object?

Ray Rothrock:

Ray Rothrock worked as nuclear engineer before heading to Silicon Valley to become a very famous venture capitalist. He was critical in helping Tri Alpha Energy secure funding from Venrock capital. Ray retired from Venrock in 2013. Through his use of resources and conn
ections, Mr. Rothrock has done more than many people to push nuclear. . He also funded Pandora’s Promise, an ideological movie supporting nuclear power. Mr. Rothrock also pushed the Department of Energy to pay attention to new nuclear. “… [the DOE] did not believe that there were 50 nuclear startups out there; they asked me to put them on the phone with some…” These actions created the Gateways for Accelerated Innovations in Nuclear (GAIN) program. GAIN has initiated several efforts into gas and molten salt reactors. Could Mr. Rothrock do the same for fusion power?

Dr. Michel Laberge (General Fusion): 

Overall, Dr. Laberge is probably one of the strongest pitchmen in fusion right now. He is relatable. A passionate, technical guy with a touch of goofiness in him. His style is very good and remained unchanged for many years. In his talk, Dr. Laberge offered some insight into pitching fusion that should be mirrored by others. General Fusions’ pitching strategy relies on de-risking capital. Their goal is to get the technology to a point where it can be acquired by a bigger company or secure more funding. It is at the moment of acquisition, that their investors see returns on their investment. The company does this on a 6 to 7 year cycle, fixed by milestones. Dr. Laberge says: “…this is how we get away from the idea that fusion is a little long…” Currently, Dr. Laberge is chasing a 300 million dollar investment for his next big machine. This sounds ominous. Scaling up is always a bad sign, unless the current machine has yielded some excellent results.

The questions and answers 

How do you keep your organization focused?

Everyone agreed this was a hard challenge. Ray Rothrock argued that the VC actually have the hardest role in keeping fusion folks on task. The money men tell the researchers they cannot follow any side projects. In contrast, Dr. Laberge says he has to fight his board to keep the focus on fusion and not spin-offs. Dr. McGuire says his team drives the focus “we really do not want to do the same thing for 20 years.”

How much of your effort is dedicated to simulations?

Any good fusion effort has a theory/simulation team as well as an experimental team. Of the three of these, the prototype is king. However, simulations will go over very well in Silicon Valley. That community knows high power computing very well. Simulations are supposed to be scouts for any fusion organization, mapping out an unknown operating space. General Fusion stated they had a quarter of their scientists are doing simulations. But these simulations were really only looking at only small parts of the machine.

How do use outside review teams?

Ray Rothrock had the best answer for this. He describes an evolving processing, where outside review panels are brought in as needed. The first panels are always all scientists. But as the company learns, its’ needs change. The second panel includes engineers and material scientists and from there business experts are brought in. “…if you don’t know, the first approximation for the cost of a power plant is the amount of copper, steel and concrete you use…” Dr. McGuire said that his review team actually helped him. They found the minimum reactor size at which they could understand the plasma.

How do you pitch? How do you hire? Everyone agreed that there was no shortage of motivated employees. Dr. McGuire described an entirely internal funding model which was totally milestone based. Tom has to pitch to the Lockheed leadership. “… What really seems to resonate is the risk versus our milestones, so we don’t have a set time… but we have an idea of the steps we have to take and what it means when we get to each step…”



Ray said a VC partnership lasts 10 years at most – but Rothrock sold this investment on a 20 year timespan. Ray also pushed the idea of adapting the biopharmaceutical investing.

How does this effort compare with China? 

At the 2016 and 2017 FPA conference, Dr. Yican Wu (head of the INEST office) gave a nice overview of China’s fusion/fission research. INEST is an office at the Chinese Academy of Sciences and would be equivalent to a very small national lab. He said that INEST has a staff of 500 PhDs, and that their average age was 31. In another talk, Dr. Baonian Wan (director institute of plasma physics, China Academy of Sciences) gave an estimate on Chinas’ investment in fusion. He estimated their budget at 2.5 billion yen (about 377 million dollars) per year, for the next 5 years. This fusion budget was worked into the latest 5 year plan in China.


INEST had wide interests. In fission, they were looking at molten salt reactors. In fusion – the Chinese were building a mirror machine. The US has no mirror program; that was abandoned years ago. China is likely doing this to mimic the Gas Dynamic Trap mirror machine in Russia. China also has a very high powered neutron source for materials testing. China is also working directly on the thermal blanket technology – this is another area where the US is not funding the research. The world leader in blanket technology is still General Fusion, a Canadian company. Still, much of Chinas’ work was focused on ITER.

What about Fusion in Nanotubes?

I once did a survey of a fusion web forum and found many people speculating about fusion in combination with some other cutting edge idea. This seems very common.

Which Approaches are pulsed? Which are continuous?

Dr. LeBarge describes General Fusion’s final machine as a 200 MW pulsed machine. Both Lockheed and TA Technologies are going continuous. A pulsed system at 1 hertz that would generate 100 gigajoules with each shot. This energy would be recycled, lost and emerge from the power plant as 200 MW of electricity. A plant of that size would be akin to a small combined cycle natural gas plant. CC Natural Gas has been very popular in the past 10 years. Those plants are relatively cheap, run on cheap feedstock and are just the right size to meet power needs in Asia. GF plant would be much smaller than a conventional fission core.

Can Fusion Ramp?

Probably. In most cases, you could store any excess energy (from pulsing) and then meter it out in a ramp. But, it depends on which approach ends up hitting net power. For example, in laser fusion (inertial confinement fusion) the "ramping" depends on how often they can shoot the target. Here is a summary of the ramps for different fusion approaches:
  • Fast ignition or direct drive, right now the best is ~45 minutes on the Omega laser system. 
  • Indirect drive: you are talking NIF where shots likely take several hours between each other. 
  • Heavy ion beam ICF: the shot time likely shrinks to seconds (ion beams are a more established technology). But ion beam ICF may not work, they are not as mature a technology. 
  • Magnetized Target Fusion: General Fusions' approach is a cross between an ICF-like compression with pistons and compact torus (think self-sustaining, spinning plasma apple). General Fusion would likely ramp by shot rate.
  • Plasma Liner Experiment at Los Alamos would also ramp by rep rate. 
If you are talking tokamaks, spherical tokamaks, stellarators, and the like you see mixed performance in run-time and temperature. You could "ramp" by raising or lowering the ion concentration inside the machine - but your somewhat limited because you have operating windows. For example, tokamaks run best in "H-Mode" and you need target temperatures, pressures and densities to hit that. You could also ramp, by pulsing these machines; here are some run times:
  • Spherical Tokamaks: The longest run time I have seen was ~26 hours by Tokamak Energy. But, their temperature and density was low. 
  • Dynomak: This approach looks to be finicky - so very limited in operating space. You would likely ramp by pulsing. 
If you are talking pinches, they would also ramp by controlling their pulsing. There are several types of pinches, pinch-hybrids and exotic pinches. These include:
  • Focus fusion: I don't know anything about this. 
  • Screw pinch: theta pinch and Z-Pinch all are short pulsed machines (pico and nano second run times). These were some of the first fusion machines and in (my opinion) these have a low chance of going commercial. 
  • MAGLIF is a combination of pinch and laser compression. This is also pulsed, but you need to wait tens of minutes between shots, because a laser is used and the lens need to cool down. 
  • Levitating Dipole Experiment I call this an exotic "gentle pinch" for plasma. They fused for like 10 - 20 seconds. Even though MIT showed they could run the machine for 7-8 hours, they did not get the funding to really try. 
If we did magnetic mirrors, like the gas dynamic trap or machines like The Mirror Experiment (TMX) we are taking pulses in the milliseconds, up to seconds. It is unclear if mirrors could ramp. Ken Fowler, Ralph Moir and Thomas Simeon recently published a paper arguing we should re-start mirrors.

So in conclusion, ramping in fusion could be done by:
  • Controlling rep time (most common method) 
  • Energy storage for re-metering later (less common)
  • Changing the ion concentration (rare) 
  • Changing the operating window (rarest) 
Whichever method we use, will depend on which technology wins.


9 comments: