What if we had no li-ion batteries?

Well, we'd have no laptops, mobile phones, tablets, no cordless drills or saws.  But would we be unable use renewables in our electricity grid?  No, we could still use them, and it's instructive to see just how.

Before renewables the grid was powered by baseload power stations, mostly coal, but with some nuclear.  But demand fluctuates, and coal power stations can't be rapidly ramped up or down to meet demand.  The output from a coal power station can be slowly increased, on a 24-hour cycle, so that output peaks roughly when demand peaks, but more frequent and rapid fluctuations in demand had to be met using peaking gas plants, which took only 5 minutes to fire up and could be powered down again within another 5 minutes if that was necessary.  To maintain grid stability (the frequency and voltage on the grid) they used things called 'synchronous condensers' which spun up using power from the grid, and added power back to the grid using the momentum of the spinners whenever power momentarily dropped. The grid 20 years ago had little storage, and yet managed to keep supply going.

Now let's add solar to the mix.  In 2009 in the USA, new industrial scale solar cost 3.1 times the cost of new coal.  In 2020, new solar costs 1/3rd of new coal, and in fact, a brand-new solar farm delivers power to the grid as cheaply as or more cheaply than the operating costs of old coal power stations, which are fully depreciated and paid off.  So, it makes sense to use solar, if we can.  But the sun doesn't shine at night.  So, without storage, what we'd do is reduce the output of coal power stations during the day, and step them up at night.  Let's suppose coal power stations can cut output to 40% of capacity during the day and expand it to 80% at night (the numbers are a bit different, but they'll do).  Let's assume nighttime demand is 2/3rds of daytime demand (again, the numbers vary from place to place, but we're discussing this as an example of what could be done).  That means (ignoring peaking gas) that coal will go from 100% of supply to 60% (1/3rd during the day and all at night).  Without batteries, we will have cut emissions from electricity generation by 40%.

But, you'll say, what about rainy days, or clouds passing over the solar panels?  Well, in a continent-wide grid, such as Europe, or most of North America, or China, or eastern Australia, the sun is always shining somewhere.  Because of this, geographically diversified solar farms will on average deliver much less variable output than a single farm.  And to cater for residual variability, just as we have excess capacity in coal-powered generation, we'd do that in solar too.  50% extra capacity would cost 50% more, but that's still hugely cheaper than new coal and not a lot more than existing coal.  And when we don't need surplus solar power produced by all that overcapacity, we can either disconnect solar farms from the grid (i.e., curtail output) or use that surplus power to produce green hydrogen and green methane.

Adding wind to the mix changes the numbers.  First, wind is at worst uncorrelated with solar (e.g., the wind blows at night when there is no sunshine) and at best negatively correlated with it.  In other words, when sunshine is reduced, in many places, wind is increased.  Think of the winds around a thunderstorm, compensating for reduced solar as the storm clouds pass overhead, or the storm winds in winter in higher latitudes, compensating for the relative absence of sunlight.    Putting both solar and wind into a grid reduces overall variability, because of this diversification of supply.  Once again, a continent-wide network of wind farms will provide much more stable output than a single wind turbine.  If there's no wind in eastern Victoria, you can be sure that there's plenty in western Victoria or South Australia.   But to allow for residual variability, you'd have at least 50% overcapacity, just as we did with solar and as we do with baseload coal.

Remember, we're assuming no storage capability.  So you'd probably still need some coal.  Let's suppose coal would provide half the power overnight, and one third during the day.  Wind and solar would each provide one third during the day, and wind would provide half at night, with coal providing the other half (night demand assumed to be 2/3rds of daytime demand).  This would reduce emissions by 60%, and would require no daily ramp up and down of coal power station output.  If we ramp up coal from 40% capacity during the day to 80% at night, coal could go from supplying 1/6th of daytime demand to half nighttime demand, with solar increasing from 1/3rd of daytime to 1/3rd plus 1/6th, or half of daytime supply.  So out of total day plus night demand, solar would provide 30% of the total, wind would provide 40% and coal 20%.  An 80% reduction in emissions.  Without batteries.

But, wait.  All these options will still require peaking gas, which is a fossil fuel, though far cleaner than coal.  As I've discussed before, we could manufacture synthetic natural gas, using CO2 and hydrogen created by splitting water using electricity generated by renewable power (The Sabatier process).  Remember all that surplus wind and solar capacity we've installed to minimise variability?  The variability of wind and solar works both ways.  Just as sometimes there is too little, at other times there's too much.  Because of overcapacity there would often be too much.  And instead of the output of wind and solar farms being curtailed, we could produce green hydrogen with the surplus, to fuel our peaking gas plants.  (That's cheating, a little—used this way, gas is a kind of battery, storing energy for later use.)   Most of the high cost of peaking gas is not the gas, it's the fact that we use the peaking gas plants for only a small part of the day, but they have to be maintained, depreciated and paid off just as if we were using then continuously.  So even though gas produced via the Sabatier process would prolly cost twice what natural gas costs, the impact on the cost of peaking gas is small.

If we look at the grid this way, it becomes clear that even without batteries, we could cut coal use by 80%.  And of course, this is exactly what's happening.  Coal power stations are being shuttered almost everywhere, coal power station capacity utilisation is falling everywhere, and nearly half coal power stations globally are running at a loss.

Now let's add batteries.  Yay!  Battery costs are falling by 20% a year.  This means they'll fall in cost by 2/3rds over 5 years.

The first thing batteries are going to be used for is grid stabilisation, replacing synchronous condensers.  Batteries respond in milliseconds to fluctuations in grid frequency, giving the grid operator time to fire up the peaking gas plants, which take 5 minutes to get up to speed.  Tesla's "big battery" in South Australia is doing this already, and is not just very profitable, but also stopped a system blackout in South Australia when gale force winds recently blew over some of the pylons in the interconnector which links the SA to the Victorian grid.   The previous time that happened, before the construction of the "big battery", power was out is South Australia for 24 hours.  The Right, of course, blamed the blackout on renewables, not the loss of  power lines.  The "big battery" has proved them wrong.

The second use will be as peaking plants.  Instead of using horribly expensive gas peaking plants, we'll use much cheaper batteries.  The batteries will charge up when power is plentiful (wholesale prices are low) and discharge when demand peaks.  Typically, every afternoon and early evening, demand surges as ppl get home from work, put on the kettle, the heating/cooling, cook supper, watch TV.  Right now, this surge in demand is catered for by peaking gas plants.  Increasingly, this will be done using batteries.

The third use will be to facilitate the elimination of coal completely.  If wind provided half of daytime demand and so (with unchanged output) 75% of nighttime demand, we'd need storage for 25% of nighttime demand, or 2 hours of average daily demand.  Add the 2-4 hours needed for peak afternoon/evening demand and we're talking about 4-6 hours of storage.  Remember, we've bumped up capacity of wind and solar by 50%, so we have plenty of (potentially) surplus electricity.  And we could continue to produce gas via the Sabatier process using surplus wind power.  But the key point is that even adding 4 hours of storage to the grid to cater for a high renewables percentage and the afternoon peak still doesn't make renewables more expensive than new coal.  True, they're twice as expensive as existing coal, but remember, that includes replacing peaking gas, which is 3 times as expensive as renewables with 6 hours of storage, as well as being even more expensive than coal.

This is all hypothetical.  Every grid and every region within a grid will have different supply profiles.  In high latitudes, wind will dominate.  In equatorial latitudes, solar will dominate.  And it also assumes a continent-wide grid.  So that will require more HVDC lines connecting regional grids to their neighbours and their neighbour's neighbours.  But my back-of-the-envelope calculations show that even without batteries, we'll be able to cut emissions from electricity generation by perhaps 80%.  As battery costs plunge, we'll add batteries for grid stability, then to replace peaking gas, and then to provide power at night.  And burning coal to make electricity will stop, not just because policy setting will change but because renewables plus storage will be cheaper than coal.

[Chart updated 7/7/21 for 2020 data]