Pages

December 18, 2008

Update on General Fusion : Steam Punk Approach to Nuclear Fusion


General Fusion is a venture capital funded company that is taking an acoustic wave approach to magnetized target fusion (MTF). The approach is to have a sphere surrounded by steam pistons that drive a pressure wave inward to generate fusion compressions twice every second. They have performed some actual proof of concept experiments at 25 times slower piston speed and performed full scale computer simulations.

This site has covered General Fusion before when their venture capital funding was announced.

There was also an external review of the work and plans

The goal is to build small fusion reactors that can produce around 100 megawatts of power. The company claims plants would cost around US$50 million, allowing them to generate electricity at about four cents per kilowatt hour.

Assuming a repetition rate of 0.5 Hz, 100 MJ acoustic pulses, a fusion gain of 6 (relative to impact energy) and a thermal to mechanical efficiency of 33%; this machine would produce ~50 MW electric.

Computer simulations show only about 12% of the incoming energy makes it into the plasma. This 12% seems a robust result; many changes of the parameters aimed at improving this coupling did not produce significant changes. We therefore require an intrinsic fusion gain of 50 to get a total energy gain of 6. This forces us to use a bigger machine delivering more energy to the plasma to achieve these higher gains.

So we can see that a machine of 2 m diameter with an input energy of 120 MJ can give gains of interest.

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).






Proposal Details

General Fusion proposes a new compression system that offers many advantages. A near spherical vessel ~2 m in diameter is filled with liquid lithium-lead alloy (Li-Pb). This liquid is under consideration for fusion reactor blankets; it has low a melting point, low vapor pressure, re-breeds the tritium, and good nuclear characteristics. The liquid is spun in the vessel by pumps that inject the liquid tangentially near the equator and pumps it out near the poles. This creates a vertical vortex tube in the liquid metal. The vessel is surrounded by many steam actuated pistons. High pressure steam accelerates the pistons to ~100 m/s. [recent tests were at 4m/s] The pistons impact the spherical vessel and send a strong acoustic wave in the liquid metal. The pressure developed at the impact is: P=ρvcs/2 where ρ is the density, v the speed of impact and cs is the sound speed in the impacting material. For steel ρ=8000 kg/m and cs=5000 m/s so the pressure developed is 2 GPa. Good steel can handle up to 450 kpsi (3 GPa) of compression. The efficiency of the driver can be quite good. About 33% of the thermal energy goes into piston kinetic energy (the usual thermal to mechanical efficiency of a steam cycle at a realistic temperature). For steel and liquid lead (density 10.8, cs= 2 km/s), the acoustic impedance (density*speed of sound) match is good with 91% of the energy going into the liquid lead. The wave then focuses in the center, getting stronger. Just prior to the wave collapsing the center vortex, two spheromaks (a toroidal magnetized plasma configuration) of reverse helicity are injected from the top and bottom of the system. They move rapidly to the center where they merge to produce a stationary FRC (Field Reverse Configuration). The advantages of this plasma target are that it can be rapidly sent in the center just prior to collapse and then stay there with low velocity while the vortex collapses and compresses it. The toroidal magnetic field is canceled and its energy goes in thermal energy heating up the plasma just prior to compression. Also, it has been observed that when merging, the resulting plasma has higher ion temperature than electron temperature. As radiation losses increase with electron temperature but fusion goes with ion temperature, this may somewhat improve the operation. After compression, the fusion energy is released in neutrons that heat the liquid metal. The cycle is repeated at ~1 Hz. The liquid metal goes in a heat exchanger to make steam. The steam is directly used to push on the pistons. Therefore the re-circulated power does not have to be converted in electricity, reducing the cost of the turbo-machinery and generator. The steam is directly used to push on the pistons. Therefore the re-circulated power does not have to be converted in electricity, reducing the cost of the turbo-machinery and generator. Typical MTF systems use pulse power technology worth around 3$/J. For typical fusion systems of order 100 MJ this is $300 million just for the pulse power system. 100 MJ of steam at 1500 psi in a 10 m**3 tank plus associated fast acting valves will cost of the order of $500 000; a considerable saving. Because of the high accuracy of the impact timing of the numerous pistons (~1 μs), an electric means of controlling the exact piston trajectory is required. But this system can control only a few % of the piston energy. In particular, it can be a braking only system not requiring any high electrical power components. The pistons are sent a few percent above the required velocity and a servo loop applies just the required breaking to adjust the impact time and velocity. The US patent application #11/072,963 describes such a system in more details. The spheromak generator will use a pulse power electrical system. But as only ~1% of the compression energy is required for the initial plasma, this should be only a 1 MJ 3 system worth ~3 M$. Most neutrons and all other radiations are stopped in the ~1 m radius of Li-Pb so the neutron flux at the wall is much reduced. This is extremely advantageous over many other fusion systems where neutron and radiation wall loading is a difficult and mostly unresolved technical issue. Radio isotopes produced by neutron activation of the structure is also a problem, especially for maintenance, in most proposed fusion machines. Expensive robotic maintenance is the usual answer to this problem. It is much less of an issue for our proposed machine. Many MTF systems under consideration also require the destruction and replacement of substantial amounts of hardware for each pulse; a costly and complex proposition. Our proposal does not require hardware replacement for each pulse.


Recent Experiment
In order to demonstrate our concept, a small experimental machine driving a spherical
shock wave to compress a pre-formed plasma was built. Because of the expected
development time of the precision piston system, we used a simpler electrically driven shock generator for this experiment. In D-D shoots, an average of 2x10**3 fusion neutrons was detected. The intent of this document is to seek peer review as a step toward raising funds for Phase 2, which aims to build and test a large 100 MJ piston driven spherical shock generator capable of achieving break-even.

We compressed a pre-formed plasma of about 3x10**16 cm-3 and 7 eV with a lithium tube collapsing at speeds up to 4 km/s. A 22 kJ shock wave in water focused on the tube to drive the collapse. We observed an average D-D fusion yield of 2x10**3 neutrons. A simple computer simulation of the experiment predicts a similar yield if an impurity concentration of 10**13 cm-3 and a symmetry limited compression ratio of 7 is assumed. Observation of the collapse with electrical pins indicates an asymmetry of 1 part in 7, consistent with the fusion yield.

It takes five days to prepare, fire and clean the set-up for each shot. About 30% of the shots fail because some parts of the set-up misbehave. The set-up is often damaged, requiring lengthy repairs, especially when new parts need to be ordered. This unfortunate reality conspires to reduce the amount of data obtained. A total of 7 successful shots were performed.






blog comments powered by Disqus