Flywheels store electricity - cheap enough

[Enthalpy]

Storing electricity is difficult at the scale and prices of the power grid, that's a good reason not to have succeeded up to now. The existing method pumps water from a lower to a higher lake and turbines it when needed; it needs mountains, lakes, and isn't perfectly efficient.

 

Speaking only from my attempt, I have no corporate interest, and after this significant effort (longer than an employer would have paid me on such a project) flywheels is the best I've found. I'm happy that batteries can be more efficient than I thought.

 

Electricity storage is a demand of renewable energy. As JC pointed out, other primary energies can be converted according to the demand. So if any corporate interest speaks against storage, then the one of fossil fuels, including uranium. But again, the endeavour is difficult enough without other reasons.

 

Losses in the motor-generator-electronics are a small worry. In the MW range, they're under 1%. A battery as well needs a conversion to AC and from a different DC voltage if fed by solar cells, from AC if a wind turbine feeds - Tesla Motors don't provide it, arguably for flexibility. Adding the losses up, a big flywheel loses less than the battery's 8% - but a home-scale flywheel is probably worse.

[CasualKilla]

Storing electricity is difficult at the scale and prices of the power grid, that's a good reason not to have succeeded up to now. The existing method pumps water from a lower to a higher lake and turbines it when needed; it needs mountains, lakes, and isn't perfectly efficient.

 

Speaking only from my attempt, I have no corporate interest, and after this significant effort (longer than an employer would have paid me on such a project) flywheels is the best I've found. I'm happy that batteries can be more efficient than I thought.

 

Electricity storage is a demand of renewable energy. As JC pointed out, other primary energies can be converted according to the demand. So if any corporate interest speaks against storage, then the one of fossil fuels, including uranium. But again, the endeavour is difficult enough without other reasons.

 

Losses in the motor-generator-electronics are a small worry. In the MW range, they're under 1%. A battery as well needs a conversion to AC and from a different DC voltage if fed by solar cells, from AC if a wind turbine feeds - Tesla Motors don't provide it, arguably for flexibility. Adding the losses up, a big flywheel loses less than the battery's 8% - but a home-scale flywheel is probably worse.

Im curious, if this technology is used to store energy for 2 or 3 days without usage, what % of the total stored energy would be lost from friction, air resistance etc. 1% efficiency does not really makes sense in this scenario.

Edited May 4, 2015 by CasualKilla

[Enthalpy]

I give estimates in the message #2, already one page ago. For the "small" (transportable) flywheel there,

• The hydrostatic bearing loses 0.9% in 10h (I have confidence in that one estimate)
• Existing roller bearings would lose 3.3% in 10h, but
• The modified bearing with big intermediate rolls loses 1% in 10h (this is extrapolation)
• Magnetic levitation gives similar figures
• Air flow loses 0.2% over 10h thanks to my flow calmer. If not, use vacuum.

0.9%+0.2% extended to 3 days would lose 7.7%, as much as the battery does independently of the (reasonable) duration.

- But at the small Powerwall size, the flywheel would be very bad

- Conversely, the flywheel improves further with the size, and I describe in message #8 how to build bigger ones.

 

To my opinion, the flywheels I describe are ready to develop for large-scale use. A dozen D=6m h=6m wheels smoothen out the consumption for a big power plant.

 

It's not only a matter of building fewer plants or enabling renewables. With a smoother production, the power plant operates at its most efficient pace, and if electricity is stored near the consumers, the power lines get more efficient.

 

And you know what, Japan lacks peak capacity to produce electricity since the tsunami, so storage is an obvious market there.

[Enthalpy]

A recent French Company called Energiestro proposes to store energy in flywheels
http://www.energiestro.fr/?page_id=71
with other design choices: home size, concrete.

Their wheel is concrete "because strong steel is expensive", and since I explained here on March 22, 2015 that the energy capacity results from the tensile strength times volume, they claim to have an original concrete reinforcement, better and undisclosed. Many people have already researched concrete reinforcement because this has many applications, so let's wait and see.

Adding concrete to the useful material reduces the speed at identical energy, which may or not linder the losses at the bearings and the flow if moving in air. It also makes the flywheel bigger and heavier, which at some point means cost.

Their 5kWh flywheel weighs 1700kg with D=0.8m h=1.5m while carbon steel band would need 270kg. I didn't check if my flow calmer performs well at this size, but a vacuum vessel does cost a bit, even if helped by concrete.

Their vertical axis has radial ball bearings and a "passive magnetic" axial bearing. An active one consumes indeed too much power at that size, as does repulsive induction. Their must rely on the repulsion by permanent magnets. NdFeB achieves it, but pairs of 5mm*10mm*20mm N45 repel 36N at 1mm, so it takes some 2*460 elements to lift the wheel at just 1G. Bought by 100, each magnet costs 1.09€, so material for this bearing costs 900€ - strong steel would improve that, I may give figures in the future.

The hurting point is that lead batteries for trucks store 12V*75Ah and cost 165$ if bought in units at the retailer. Six of them store the same 5kWh, need no development, take little room, and cost 1000$. While I agree that batteries have a limited life, they can be replaced and are recycled, so my opinion is that flywheels must be better to have a chance.

Edited July 11, 2015 by Enthalpy

[Enthalpy]

The force between magnets of Nd-Fe-B or ferrite isn't that difficult to compute. Their relative permeability (1.06) can be neglected. The coercivity H is represented as a current per thickness unit around any surface element, and as the currents compensate at the common edges of surface elements, it's equivalent to a current sheath flowing only at the magnet's facets parallel to the direction of magnetization.



A flywheel would have complete tracks of magnets at the stator and rotor, so the butt facets compensate an other, leaving only azimutal current sheaths. Iron can increase the force a bit but I neglect it here. The big track radius makes it nearly straight, so the force can be summed from the ampere's definition: 2*10^-7N/m for 1A*1A at 1m. For thickness elements du and dv at the stator and rotor, the (implicit in the following) force element at 1m and the summed force for one facet pair are



This sum is clumsy to compute, but I do it differently - understand it as a Lebesgue integral if you like, or as a substitution x=u-v and y=u+v.

The uniform current sheaths make uniform force elements spaced by some x and cumulate for the distance x a sheath thickness lambda(x) distributed as a triangle, making an easy sum that is even finite in contact.



How accurate? Two 40mm*10mm*10mm N45 magnets create observed 86N at 2mm spacing.

• 1080kA/m and 40mm bring the implicit term 9331N*m;
• e=1mm E=11mm make F=91N for each facet pair;
• But the crossed facets attract an other. Mean distance=15.6mm and cos=0.768 reduce by 46N per facet pair;
• Two pairs result in computed 90N.

That's more accurate than usual in electromagnetism. The finite long facets make less force per length unit than estimated but the short facets add force.

Marc Schaefer, aka Enthalpy

 

[Enthalpy]

Some permanent magnets layouts improve thrust bearings and adapt to other bearing forms.

The current sheaths make it intuitive:

• Broadening a pole lane too much doesn't increase the force, since the current sheaths flow only at the inner and outer radii.
• But splitting the area radially in North and South lanes does increase the number of sheaths and the force.
• Two current sheaths add at the north-south transitions, both at the stator and the rotor, to quadruple the force.
• More lanes increase the force up to the point where N-S transitions radially too close interact to reduce the force. The optimum could be magnets 2-3x wider than thick.
• Some tiny radial spacing between the lanes improves the force slightly.

 

This holds for a small gap. Alternate lanes reduce the force quickly at bigger distance, useful. And if an iron path improves the force already, lanes must gain less. Also, I didn't estimate how unstable the radial position gets with such lanes. It may load the radial bearings, which have to be stiff: rollers, needles.

Like for electric motors, cogging will improve by:

• Skewing the transitions between successive magnets.
• But alternate the skew direction to minimize axial loads.
• Taking different numbers of magnets at the rotor and stator, with a small GCD like 1, or for symmetry 2 or 3.
• Offsetting between the lanes the azimutal magnet transitions.
• The lanes have different lengths at a thrust bearing. Different numbers of magnets per lane let the cogging torques beat; if needed and when possible, have magnets of varied length, or introduce tiny gaps.

Marc Schaefer, aka Enthalpy

 

[Enthalpy]

Here's an example of steel toroid lifted by magnets. D=20m d=14m to store significant energy, and 1500MPa resist 364m/s +20%.

The gap is 2*e=6mm, the N45 magnets are E-e=15mm W=40mm. 2*75 lanes of mean 53.4m need dangerous 2*2.40m³ costing 3.6M€ at 2kU price, possibly 2.4M€ in that amount.

The lifting force is 30MN. It permits 2.0m thick steel at 1.2G, and plain bearings shall act at stronger quakes. 2516t steel store 167GJ = 3h*15MW. 20 tori store 3h*300MW, enough to smoothen the production of a 1,300MW plant, and their magnets cost 50M€.

That's not cheap, because the flywheels have other costs, but could be affordable to save 1/4 of a 2G€ plant and improve its efficiency. 1M * 900Wh truck batteries may cost 90M$ but are too inefficient.

The thin magnetic gap precludes my flow calmer and imposes costly vacuum. The load hence losses at the radial roller bearings aren't obvious. Magnets must be sorted and laid out to remove short-period fluctuations of the induction. The bare edges or the magnets would induce unbearable eddy currents in the facing magnets, but capping the magnetic lanes with thin, very permeable material solves it.

I prefer hydraulic or roller bearings and the flow calmer, but magnets are credible.

Marc Schaefer, aka Enthalpy

[Enthalpy]

British MPs want to develop energy storage because renewables are intermittent:
http://www.bbc.com/news/uk-37664880
which is obvious logic and will become an ever stronger need as the fraction of renewables increases.

[Externet]

Flywheel storage is already in use in the Falkland islands, paired to wind turbines. Tried to find the manufacturer, only saw this smaller? unit :

 

----> http://www.bomara.com/powerware/pf2-flywheel.htm