Renewables could be the world's primary source of energy if only someone could solve the storage problem—how to store lots of electricity cheaply on a wide scale? Batteries are too expensive and don't last long enough. Pumped hydro is cheap but not feasible for most locations. Thermal storage is promising but still too expensive or hard to scale. Compressed air is cheap and scalable but not yet efficient enough (although LightSail, a new company backed by Peter Thiel, Vinold Khosla and Bill Gates, hopes to change that). And what about flywheels? The biggest player, Beacon Power, went bankrupt in 2011.

Flywheels may be getting a second life, however. Silicon Valley inventor Bill Gray has a new flywheel design that would deliver distributed and highly scalable storage for around $1,333 a kilowatt, making it price competitive with pumped hydro and compressed air. With an efficiency of more than 80 percent, it would rival the best storage alternatives, and come with a 10-year guarantee. And it would make a perfect complement to an off-grid house with a solar photovoltaic (PV) system, able to charge fully in five hours—within the charging time of most solar PV systems—and store 15 kilowatt-hours of power, enough to run a modest house from sunset to sunrise.

Gray calls his invention the Velkess (for VEry Large Kinetic Energy Storage System). He is currently raising money for the prototype in a Kickstarter campaign.

The Velkess improves on traditional flywheels by better managing the natural "wobble" of a spinning mass. Traditional flywheels have been very expensive because engineers align the natural axis of the wheel's rotation with the desired rotation of the generator. Thus, they are always struggling to minimize the natural wobble of the wheel using very expensive magnets and bearings, high-precision engineering and materials like high-grade carbon fiber or rigid steel. Beacon's flywheel for grid storage cost a whopping $3 million per megawatt-hour.

Instead of trying to fight the wobble, Gray redirected it by suspending the wheel within a gimbal—the same concept that makes a gyroscope work.

The gimbal in the Velkess is asymmetrical, so the two axes of rotation—the flywheel axis as well as that of the rotor, which drives the brushless, inducting DC motor—are not on the same plane, and have different periods of frequency. This dampens the resonance effects that make traditional flywheels hard to control (a resonant disturbance in one of the planes can intensify until the device shatters). With the gimbal, resonance in one plane is translated into the other, which is nonresonant at the same frequency. Accordingly, only very loose engineering tolerances—about one sixteenth of an inch—are required to build the device.

Gray relied on pioneering work by mechanical engineer John Vance, a retired professor at Texas A&M University, who has conducted extensive research on flywheels, machine vibration and rotor dynamics. Gray also reduced the cost of materials. Instead of making the wheel out of expensive steel or carbon fiber, Gray made his out of cheap "E-glass"-grade fiberglass—the same stuff that's used in things like shower doors and fishing rods. Because it's a far more flexible material, the fiberglass wheel tends to warp and wobble much more than steel or carbon fiber when its spin velocity changes.

Gray's focus on bringing down capital costs of storage when other flywheel manufacturers have focused on increasing density and generating capacity seems to have paid off. The Velkess could store electricity for $300,000 per megawatt-hour, or about one tenth the cost of the Beacon unit, Gray says. "I am glad to see this concept receiving publicity, as I believe it holds some promise for success," Vance says of the Velkess.

Gray's design also makes it easier to control the possible failure of its most critical parts: the wheel itself and the bearings. Because the flywheel is a bundle of thousands of fibers, if one fiber breaks, it will just wiggle out of the bundle rather than directly stressing the rest of the wheel and causing it to break. The flywheel would simply "shed" lightweight material.

For the bearings, the Velkess uses "angular contact ceramic hybrid" (silicon nitride) ball bearings running in a stainless steel track, where both the bearings and the thrust load are floating on magnets. If the bearings began to fail, the heat they generate would be detected early on by a simple temperature sensor.

Either type of failure could be easily detected long before a catastrophic event, allowing the device to throw a warning and trigger a shutdown. In an off-grid shutdown the device would discharge hot air until it spun to a stop—roughly equivalent to a 1,500-watt hair dryer running for 10 hours. In an on-grid application, it could simply dump the power to the grid.

The whole device is contained in a vacuum-sealed steel box with about the same footprint as a household refrigerator, only a bit shorter. The flywheel itself is about 66 by 66 centimeters in height and diameter, and weighs about 340 kilograms.

It will be optimized to provide up to three kilovolt-amps of continuous power output at 27 amps, but can handle higher "burst" power loads that occur when heavy-draw appliances like water pumps and circular saws start up. The unit could discharge at any rate up to three kilowatts until the flywheel spins down to its "fully discharged" speed of 9,000 revolutions per minute.

Gray intends to target the 48-volt off-grid residential market initially, where the Velkess would be a drop-in replacement for typical 48-volt battery systems. After that would come the 240-volt residential and small commercial markets, where the Velkess could provide the backup for grid-connected solar PV systems when the grid goes down. Eventually he hopes to get into the 600-volt utility-scale solar market.

Velkess could succeed where Beacon failed on several counts. The latter device, like most of its competitors, could only discharge large amounts of power for very short durations, whereas Gray's would do the opposite: It could discharge as slowly as needed for hours. And where Beacon's system was so expensive that it only really made sense for industrial applications, Gray's would be cheap enough to make good economic sense in the residential and small commercial markets.

Further, the Velkess satisfies the final crucial factor for electrical storage—scaling. Multiple units can be linked together in parallel.

According to an analyst at Boston-based Lux Research, energy storage services could be a $31.5-billion market globally by 2017. If the Velkess prototype can be built at the price and performance advertised, it could take a big chunk of that market, and solve the intermittency problem of renewables once and for all.