A recent US study comes to similar conclusions.
Let me explain the chart as I understand it. The left and right axes refer to both charts. The left axis shows mean (average) renewables generation as a ratio to total generation, so 1=100%. To satisfy electricity demand at all times, you will need on average more than 100% output from renewables because there are times when output from renewables is too low to satisfy demand, even if on average 100% (or more) of total demand could be produced from renewables. The right axis shows output relative to renewables capacity. Because capacity factors are not 100% the scale runs from 0 to 20, not 0 to 4. The dotted lines show where 100% of demand is satisfied. The horizontal scale measures the percentage from renewables. The charts in column a show this from 0% penetration to 100%, with a linear scale. The charts in column b show the same data but plotted on a log scale from 90% to 99.99%, because it's the last 10% that's hardest. Moving down the charts, each one shows different percentages of wind and solar, with each of the different lines (black, pale green, green and dark green) showing what happens with different percentages of storage.
Now let's start with 100% solar. Note that the black line (zero storage) goes exponential at about 50%, which is entirely logical: the sun doesn't shine at night. So after the sun goes down, you would need infinite quantities of solar panels because output is zero.
100% wind (the lowest chart) goes exponential at about 90%. This is because there are periods when the wind just doesn't blow, even if on average it blows a lot. It's not as bad as 100% solar without storage, because the wind does blow at night, but it would still mean (without storage) that 10% of the time there would be power shortages.
The best combination (without storage, black line) is the 75% wind plus 25% solar, closely followed by 50% wind plus 50% solar, as one might expect. The best combination with 12 hours storage (pale green line) is the 50/50 option. In fact the 50/50 option is the best with all levels of storage above zero--12 hours, 4 days and 32 days.
Now note that we could, with wind, use additional capacity to cover those days when the wind doesn't blow. In fact reading off the chart (right axis) if we installed 8 times the capacity, with 12 hours of storage, we wouldn't run short of power. That would obviously be horribly expensive (storage would be much cheaper). And wasteful: most of the time we would be curtailing the output of wind farms and solar panels because we would be generating far too much electricity, just to ensure we have enough power 10% or 1% of the time.
Some points to remember:
- For a mixed wind/solar system, the problem only arises as we go above 60 or 70% penetration. Below that, just some storage is required. We won't get to 70% average renewables penetration for at least another decade. "Controversies about how to handle the end game should not overly influence our opening moves. (1)"
- The model assumes no hydro power. Obviously, with hydro power, we can open the turbines when needed and keep then closed when not needed.
- Though there will be a need for grid-level storage, lots of storage will be behind the meter with households and businesses. A single Tesla Powerwall will store enough power to get you power you through the evening peak, the night and the early morning before the sun comes up.
- Battery costs are falling very rapidly. As Tony Seba points out here, batteries are falling 20% in cost/KWh per annum, and far from slowing, the rate has accelerated since 2010. At this rate, battery storage costs will halve (continue to halve) every three years. 24 hours of utility-scale lithium-ion battery storage costs about US$107/MWh now. In 6 years' time that will have fallen to $28/MWh, In 15 years, to $4/MWh. 4 days of utility-scale storage will add just $16/MWh to wholesale electricity costs.
- We will need seasonal storage, because to reduce the risk of blackouts for 9 hours a year (0.1% of the year) we'll need 32 days of storage, i.e., far more storage than we'd mostly use. Currently these unusual periods of strong demand/ low supply are met by gas peaking plants. We can go on using gas peaking plants as well as reduce emissions via the Sabatier process, which produces synthetic natural gas using surplus green energy (power-to-gas). In most countries, the gas grid stores at least 6 to 8 weeks of gas.
- We will also probably have 50% overcapacity of wind turbines and solar panels in the grid, as we do now with fossil fuel generation, and curtail output when supply is excessive. Note that in the chart of 50% wind/50% solar, 50% additional capacity, with 12 hours of storage, will take us to 99% reliability.
- Demand management will also play a role. Big consumers will be offered cheaper rates (say 10% off normal) if they agree to have power cut for X hours a year. In fact, it may be cheaper for them to accept this offer and install their own batteries than to insist on 100% supply relaiability. And EV owners can be persuaded to recharge their cars when wholesale prices are low, i.e., when supply is higher than demand.
[Read more here.]
The picture is becoming clearer. Most countries in the world produce 10% or less of total electricity from wind and solar, and those which produce more have substantial hydro to provide the storage backup needed. So for now, the expansion of renewables can continue without the risk of blackouts. However, as the percentage increases, the amount of storage will have to rise to reach 12 hours by the time we hit 60% or 70% after which it will have to rise exponentially. Fortunately, storage by then will be dirt cheap, and if it isn't we can use power-to-gas to ensure 100% reliability.