[Enthalpy]

Hello dear friends!

Wind turbines rotate at low angular velocity because their blades are huge, around R=100m these days. Though, generators prefer a high azimuthal speed that increases the voltage hence power at identical current and ohmic losses. Two solutions have competed for decades:

• A gear increases the angular velocity for a fast, small, cheap and efficient generator. But the gear for huge torque is costly and needs maintenance.
• A wide short thin generator, like D=10m, rotates at the turbine's speed. Less maintenance, but the generator is costly and it can waste 10% of the harvested power.

Vestas makes some fuss presently by announcing a superconducting generator. Logic: accept small voltages because high currents have no drawback.

This option differs so much, with a small but costly generator of unknown reliability and maintenance, that comparisons are difficult.

==========

I should like to remind the electrostatic machines I described over a decade ago. One may try the wayback machine:
  http://www.physforum.com/index.php?showtopic=26432
  https://lofi.physforum.com/Electrostatic-Alternator_26432.html
  https://saposjoint.net/Forum/viewtopic.php?f=66&t=1684

Their are powerful according to my figures, nothing intuitive nor banal for electrostatic devices. They have essentially no ohmic losses, so they accept a slow rotation with excellent efficiency.

==========

Now, I suggest induction machines with chilled metal conductors as one more alternative. Copper and aluminium resistivity drops /10 around 80K and /100 around 45K, so do the ohmic losses.
  https://nvlpubs.nist.gov/nistpubs/Legacy/TN/nbstechnicalnote365.pdf [5MB] pages 40 and 20

That is, instead of dissipating at 300K 10% of the harvested power, the conductors would dissipate 1% at 80K or 0.1% at 45K. Carnot limits the cooler's efficiency to 1/3.8 and 1/6.7. If the cooler is no worse than 1/2.6 or 1/15 of Carnot's limit, chilled metal brings a net power advantage. The power advantage can then be traded for size reduction.

The machine is more complicated than at room temperature, but far less than at superconductor temperature. Big superconducting devices, for colliders or NMR, still operate at 4K presently: the brittleness of 77K superconductors must stop the designers.

Other applications need slow machines and considered superconductors, for instance orientable propulsion pods for boats. Chilled metal seems an alternative for them.

Marc Schaefer, aka Enthalpy

[I will need your help for liquids with high permittivity that insulate very well, more to come.]

[Enthalpy]

Looking for information, explanations, enlightenment...

My electrostatic machines need insulators. Vacuum or gas for high speed, liquid insulators when low speed allows. The produced power is proportional to the permittivity, high is hence better. But the liquid must insulate perfectly - I should determine the need, but I remember that 10⁴Ω×m is clearly too little.

These needs resemble aprotic polar solvents in chemistry: polar for permittivity, aprotic to insulate. Among the liquids with high permittivity are carbonates of ethylene or propylene glycols, DMSO, acetonitrile... Opinions please? Are solvents the good choice?

I fear that all such liquids are hygroscopic. What do you mean?

Do you expect such a liquid to conduct by itself, or only after moisture absorption? For instance, is the resistivity a known method to check how dry such a solvent is? Would you know data about resistivity versus moisture?

Many thanks!

[Enthalpy]

So, does someone know whether the moisture traces make the residual conductivity in polar aprotic solvents like propylene carbonate? As usual, I spent hours on the Web with no access to science papers and found nothing.

If propylene carbonate and similar liquids with high permittivity produce no ions by themselves, just applying current should remove the conducting impurities. Water would be electrolysed and escape as gas, nonvolatile ions would reach the electrodes that could catch them, maybe as resin, maybe behind membranes. It needs electrodes that don't inject ions. This resembles known processes
https://en.wikipedia.org/wiki/Electrodialysis
https://en.wikipedia.org/wiki/Electrodeionization
An electrostatic machine in operation would even do it spontaneously: if some moisture arrives in the dielectric liquid, the strong electric field suppresses it until the current leak is over.

But if propylene carbonate and siblings conduct so much by themselves, they are just excluded from the choice. I checked that molecules like phenanthrene, pyrene... with less stiff Homo electrons improve the permittivity to 3 instead of 2 for alkanes. No big gain, and the liquid ranges let prefer branched alkanes. Ramblings already done in the other thread
http://www.chemicalforums.com/index.php?topic=56069.msg246917#msg246917

Unless someone knows a better example, I admit that high permittivity in liquids demands a strong dipolar moment prior to any external field. Whether such liquids can make good insulators is still unclear to me. Then the design of electrostatic machines shall assume a low permittivity.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Having seen no excellent insulating liquid with a high permittivity, I'll consider only nonpolar compounds as a liquid dielectric for slow electrostatic machines (quick ones would rather use a gas or vacuum).

Paraffins are the standard, much refined to avoid bubbles and ageing multiple bonds that lead to electric arcs. Better purity results from synthetic insulating liquids like 1,1-diphenylethane.

My present choice is farnesane because it has become abundent
  https://www.chemicalforums.com/index.php?topic=56069.msg392278#msg392278
and combines excellent melting point, flash point, viscosity.

But if phytane gets cheap, it will improve the flash point further, and could form a eutectic with farnesane.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Alkylsilanes are expected to widen the liquid range over alkane homologues. Silane isn't cheap. Production seems easy but dangerous
https://www.chemicalforums.com/index.php?topic=56069.msg327456#msg327456
https://www.chemicalforums.com/index.php?topic=103039.msg362089#msg362089
with many tuning possibilities.

I suppose the hydrophobicity and the electric properties resemble the alkanes homologues, so the design of the machines remains the same.

[Enthalpy]

I put few quick figures on a wind turbine alternator with chilled Al coils and I didn't get a complete proof-of-concept, for lack of technical documentation and time.

I started from the Vensys 136 turbine: 3.5MW @0.178Hz rotation, magnetic gap estimated 1.0T D≈5.4m L≈1.4m. Possibly 288 slots, where 65% Cu lose ≈4% (not accurate).

The currents induce only 0.2T=2kOe in the slots. This multiplies the resistivity of 20K pure Al by only (1.0+2.2) while Cu isn't usable
  lss.fnal.gov
At 33% of Carnot's limit, and without iron losses, the cryocooler would dissipate 0.4% of the 3.5MW. Gained 3.6% of mean 3.5MW/3 would pay a turbine half a year earlier.

I didn't check the eddy current losses in cold Al at low induction and frequency. Solutions exist qualitatively.

Heat insulation demands vacuum. The magnets too must be cold unless we insulate the magnetic gap. Small shaft diameters ease seal rings, a significant change for wind turbines with slow alternators. Then, multilayer insulation leaks very little heat.

Transmitting the huge torque leaks very little heat if done near the gap diameter with trusses of tubes. Titanium, stainless steel or glass fibers suffice.

Extracting the current does leak heat. Figures are bearable if 3kV reduce the section of the 3 conductors of pure Al. 500V seem little. A resonant power supply might transmit magnetic power over the temperature drop, but I dislike that.

The magnetic materials I checked dissipate too much. Hysteresis losses need improvement because this loss would be extracted from cold. Fe-Si worsens at cold, Fe-Co is too expensive, Fe-Ni carry less flux. Existing amorphous Fe-Si-B provide no big dimensions.

Maybe an adequate magnetic material exists - somewhere. Or small Fe-Si-B parts can make a big magnetic path if overlapping as in the past or if pressed together. Coil insertion would get easier too.

Then the design should be reoptimized, rather at 40K, with more poles, possibly less induction and sleeker deeper slots. I stopped before.

Marc Schaefer, aka Enthalpy