Four ways to store sunlight

The post opens by taking the word “intermittent” away from renewables. Demand is intermittent by the second and by the season, and fossil supply is intermittent too, just generationally: we know which wells flow tomorrow, not which ones our grandchildren will find. The system has always answered that with inventory, and the fossil system holds an enormous amount of it in coal piles, oil tanks and gas reservoirs. Electricity has almost none, because until recently pumped hydro was the only way to store it at scale and little was built. The conclusion Lubershane draws is deliberately undramatic. He expects no tipping point at which renewables make grid operators throw up their hands, and says there is no magical percentage. Instead the cost of dealing with intermittency inches up alongside renewable penetration: gas turbines run more erratically and lose efficiency doing it, curtailment becomes near-daily in spring and autumn, and multi-day windless, cloudy stretches start to bite. None of that makes storage necessary. It makes storage a steadily better business, which is why the rest of the post is about prices rather than physics.

What follows is organized by duration, because duration decides which technology can compete for which job. The first grid batteries did sub-hour work, smoothing the system in real time, which proved a good business and a tiny one. The daily net-load peak is the market that matters now, and Lubershane thinks lithium-ion is very hard to displace there: it is proven, it rides on the electric vehicle supply chain, and challengers keep underestimating how much further it can go, as with the shift to iron-phosphate cathodes and the arrival of sodium-ion. So he states an explicit bar a rival would have to clear. The turn comes at longer durations. Once renewable penetration reaches the 40 to 60 percent range he is confident many regions will reach, the peak left over after four-hour batteries have done their work is a 12 to 16 hour problem, and a much larger one. Beyond that the ladder has a gap: past about 20 hours he says extra duration is worth practically nothing until a new realm opens at around 100 hours. For lithium-ion, adding hours means adding batteries, which leaves it several times too expensive to compete with a gas turbine for the 12 to 16 hour job. Halving its installed cost is the specification the rest of the post is written against.

Heat is his answer, because the material holding the energy can be rock, brick or carbon, cheap enough to leave room for a whole system under the target. The previous attempt, molten salt, stalled for operational rather than material reasons: the salts are toxic and corrosive, they need complex pipes, pumps and tanks, and the whole system breaks if the salt is ever allowed to cool. The new generation mostly uses solid media and charges from cheap electricity rather than concentrated sunlight, which he reads as a testament to how far solar costs have fallen. Heat storage also has two modes with completely different economics. Turning electricity into heat and delivering heat is almost lossless, because the loss is itself heat, and it displaces fuel burned in industrial furnaces, so one purchase buys both electrification and storage. Turning stored heat back into electricity through a steam turbine throws most of it away, and his defense of that is the move the whole post keeps returning to: efficiency matters much less when the input is power that would otherwise have been curtailed.

Past a day he names a company rather than a technology class, and the design point is roughly four days, about what it takes to ride through an unusually poor stretch of wind and sun almost anywhere. Then he argues against his own case. He is skeptical that we are heading for a nearly all-renewable grid, because siting and permitting for renewables and for transmission will bind first, so he has to justify multi-day storage without that premise. He gives three reasons: it lets capital-heavy plants that want to run flat out, nuclear, geothermal or gas with carbon capture, do exactly that; it lets wind-rich states go very high renewable locally even if transmission caps what they export; and it can anchor a microgrid at a data center, hospital or campus, giving backup while earning grid revenue the rest of the time. He then concedes more than that last case needs: in most places in North America he expects gas generation to remain cheaper than storage as a multi-day balancing resource, and rates microgrids one of the highest-value uses for a small amount of gas. Hydrogen covers the strategic scale, and he is blunt that its round trip is bad at every step. His case for it rests almost entirely outside that chain: storing it underground is extraordinarily cheap, the world will build hydrogen production, pipelines and caverns anyway because the molecule is an irreplaceable chemical feedstock, and the gas turbine fleet is being built ready to burn it. It can also be upgraded into ammonia or methanol, denser molecules with existing global markets, which is why he ends up treating hydrogen as a common currency between sectors rather than a grid storage winner. Bulk power storage, on his view, is a spillover application for infrastructure justified by other things.