If you know something about renewable energies, you have surely heard about its intermittent drawback: the sun does not always shine and the wind does not always blow, so producing energy from renewable energies does not give the type of incessant energy production from fossil fuels. . oxidizers or nuclear energy.
There is also the downside of overproduction of renewables, which could threaten power internet surges, and the excess power must essentially be drained from the grid, ruining its potential usefulness.
Although battery technology has advanced in recent times, we are not at the point where large-scale energy storage is possible. Until then, the intermittency drawbacks of renewables are hampering their widespread adoption.
This is the challenge that researchers at the University of Southern California are trying to solve with a novel solution: information batteries.
Information what?
The idea of an information battery is not so strange if you think about it. The problem we are trying to solve is to make renewable energy more consistently productive. The only reason we generate power is to turn it into some kind of practical work, whether it's running a car engine, running your home's air conditioning system, or powering a Google data center.
While the long-term solution to bringing renewable energy to all is traditional battery technology, there are steps we can take to speed things up. This is where information batteries come into play.
"The way things are going, in 5 years the amount of renewable energy wasted in California from year to year is going to be equivalent to the amount of energy used from year to year in the city of Los Angeles," said Barath Raghavan, a teacher computer science assistant at Viterbi School of USC. . Engineering.
Finding productive use of excess energy could go a long way toward offsetting the energy demands placed on renewables during periods of low power generation. The idea is to effectively shift this energy consumption from periods of underproduction to periods of overproduction and save the result of this work for later use.
You can't really do this with a vehicle or even an air conditioner, but you can do it with data processing, where the term data battery comes into play.
Storing power from renewables is hard enough, but storing data is impressively simple, so instead of saving power from a solar panel that a data center might use at night, ask the data center to do predictable calculations on moments of excess supply. The data center can then save those results until they are needed later, which is a much less power-intensive operation.
"We have found that if we can forecast possible calculations that may happen in the future, we can perform those calculations now, as long as there is free energy, and save the results, which now have embodied energy," Raghavan said.
How could you plan IT work in advance?
One of the key peculiarities of the information battery idea, which Raghavan and Jennifer Switzer, a doctoral student at the University of California, San Diego, describe in a recent article published in the ACM Energy Informatics Review, is that a lot of computational work know in advance.
"Imagine a huge computational task that looks like a huge puzzle, where each piece is a computational piece," Raghavan told TheComparison this week. “You could do it all at once, a puzzle in one piece, if you know in advance what each and every calculation is going to be. But often you don't know XNUMX% what a future task is going to be. multiple smaller puzzle pieces.
"Although you can't do everything in advance, you can do a lot. Consequently, only a small amount needs to be done in real time (the few small parts that weren't precomputable), the rest leveraging speculatively executed computed coins."
Also, many of these calculations are likely to be reusable for different applications or calculations, so the energy savings of repetitive real-time computing can really start to add up.
Doesn't it cost energy to store all this data?
Yes, but not on exactly the same scale.
To read, write or otherwise interact with the battery of information, you will obviously need to expend energy. But with long-term storage, once the data is written to the battery and indexed so that it is easily accessible, the energy cost of using it is minuscule compared to the energy required to recalculate that data in time. real.
"It depends on the storage medium and the type of computing, but we're talking much, much more efficient in general," Raghavan told us.
"As a very rough estimate (far from the envelope), a high-end server hard drive has an embodied energy of about two GJ, which is quite a bit more than a smartphone and would equate to about 1 kJ/MB in steady state , and the MB here is the output data (which is going to have the intrinsic energy of the computation).
"Its operating power consumption is low, around four W."
Would this energy saving be enough to solve the inconvenience of the intermittence of renewable energies? It may not be enough on its own, given that the amount of energy required to run the planet's computers is increasing at a rapid rate. But before we can really talk about generating more energy, making sure that we fully use the energy we do generate is an essential step in the right direction.