More storage? Or more capacity?

When wind and solar first started to be used for generation on the grid, the aim was to use as much of the power as they created because their cost was so high.  Curtailing output would just make an expensive thing even more expensive.  But the costs of wind and solar have declined rapidly, and continue to decline.  "Wasting" some of the renewables output by curtailment is more acceptable, and indeed, as renewables penetration has increased, necessary, when good wind and/or solar conditions mean that otherwise there would be too much output, leading to the grid burning out.

Also, studies have repeatedly shown that although little storage is needed at low levels of renewable penetration as penetration rises, more storage is needed, and the need rises exponentially after penetration reaches 70%.  I wrote a long piece a year ago summarising some interesting research, here.  In a grid without baseload generation, such as hydro or nuclear, even with a mixed 50%/50% wind/solar generation base, we would need 32 days storage to take the grid from 90% to 99.99% renewables.  At that time, I pointed out that we wouldn't be getting to 70% penetration for 20 years, by which time storage costs would have fallen by 99% if current trends continue.  Which they prob'ly won't—but the current 20% per annum decline will very likely continue for another 10 years at least, which will mean by 2030, battery costs will have fallen 90%.

But what if we just added extra capacity, instead of or in addition to storage?  First, that won't work with a grid with 100% solar.  No matter how much capacity you have, the sun doesn't shine at night.  Outside the tropics, some night-time demand will be satisfied by wind.  Without wind, i.e., just solar, you would need at least 12 hours of storage.  I've assumed 8 hours of storage (1/3rd hourly daylight demand x 12 hours) will be needed, plus wind, to be conservative.  This will cover the day-to-day fluctuations.  But what about seasonal deficits?  I mentioned 32 days of storage above, which would be essential to give the grid 99.99% guaranteed supply even in prolonged cloudy, wintry calm, when demand is high and supply low.  Or during an Arctic vortex event, such as hit the N.E. USA earlier this year.

Let's have a look at a potential example.  I went to the ever reliable PVWatts (run by by NREL) and got them to tell me how much power is generated each month on average in Minnesota.  Solar panel output is lowest in December (229 kWh from 4 kW of panels) and highest in July (629 kWh).  December output is 36% of July's.  So if you were going to run all of Minnesota on solar electricity, you would need triple capacity to provide for that one month in winter, and would dump/curtail lots of power in summer.  Of course, at such a high latitude, no one would run a grid just on solar.  So let's assume 50% wind, and allow for some variability.  In that case double capacity of wind and solar  (plus 8 hours storage for nights) should be enough.  Double capacity, though, would more than double the cost, because some output would be curtailed.  By my calcs, about 1/3rd of the excess capacity would be curtailed in Minnesota if solar capacity were doubled, so doubling capacity would be 2.3 times as expensive.  I'll come back to that in a minute, below the chart.

In the chart below (I've shown variations before) I've taken the average LCOE estimated by Lazard from their latest report for each year since 2009 for wind, solar, coal and gas.  Wind+solar is the average of wind and solar individually.  I've assumed that the rate of cost declines for the last 5 years continues for the next 3.  I've estimated battery costs using battery pack prices, and in a change from my previous published estimates, assumed a 30% premium for the concrete base, connection etc.

The different green lines in effect show the different cost structures as renewables penetrate the grid.  The solid green line is a 50/50 average of wind and solar without storage, and would be appropriate for low penetrations of wind and solar in the grid.  That crossed the coal line in 2012.  The dotted green line adds the cost of 8 hours of energy, and would be appropriate for 40 to 60% penetration in the grid.  That combo became cheaper than new coal in 2015.  The dashed green line shows the costs of doubling capacity, still with 8 hours of storage, which is still though not for much longer, more expensive than coal.   However, that's a conservative estimate, as I'll explain below.

What would happen if large chunks of potential electricity generated were to be wasted via curtailment?  It would be "free" electricity.

First off, wind and solar farms would add more on-site storage so that when they were told to curtail output by the grid operators, they would divert production to their own storage for delivery later.  For example, wind blows all the time, but demand is mostly in the day.  In South Australia, with 50% renewables penetration, most of it wind, wholesale prices can go negative in the wee hours.  So why not store their surplus production then for delivery into the afternoon peak, when wholesale prices soar?  Of course, that's just what they would do.

And for prolonged periods of surplus output, say during summer, it would make sense to store that surplus energy as hydrogen or as synthetic natural gas.  Just to remind you, you take green electricity, use it to split water into hydrogen and oxygen, pass the hydrogen with CO₂ over a catalyst at high pressure and temperature, and that gives you methane.  This is called the Sabatier process, and I have talked about it often.  There is a 65%+ energy loss in this process, plus capturing the CO₂ from a gas power station flue adds $6-$34/MWh.  But if the energy is "free", the energy conversion loss is irrelevant.  And thus power-to-gas becomes much cheaper.  So instead of curtailing output from wind and solar when supply is excessive, that surplus supply would be used to produce hydrogen and methane, which would be stored to cover high electricity demand in winter.  Curtailed output would then have some value, reducing the cost of the extra capacity.

Note that the conservative costings for doubling capacity, i.e., ignoring any revenue from selling otherwise curtailed output for power-to-gas, cuts across the coal cost line in 2020, and the gas line in 2023 (not shown on chart).   And that's US gas, which is a lot cheaper than in the rest of the world.

The future grid will have:

• a mixture of wind and solar, varying by latitude
• plant-level and prob'ly grid-level battery/pumped hydro storage equivalent to a minimum of 8 hours storage and likely more
• long distance HVDC lines to bring power from other regions where the weather and the climate is different
• much more generation capacity to ensure supply at times when wind is low and the sun isn't shining
• seasonal storage using the Sabatier process to create synthetic natural gas, i.e., methane, to provide reserves for winter.

And. it occurs to me, hydrogen-cell cars may actually be a thing.  I've always dismissed them before because of the costly energy conversion losses caused by producing hydrogen from electrolysis.  But if we need extra capacity in the grid, and the result is cheap hydrogen, hydrogen-cell vehicles might yet work.   Ditto, methanol fuel cells.



[Hat Tip to CleanTechnica which started me thinking, with this article: No Joke: We Should Build More Solar & Wind Than Needed — It’s Cheaper]