A Little Arithmetic: The Costs Of A Solar-Powered Grid Without Fossil Fuel Back-up

Yesterday’s post made the point that states or countries seeking to march toward 100% “renewable” electricity don’t seem to be able to get past about the 50% mark, no matter how many wind turbines and solar panels they build. The reason is that, in practical operation, due to what is called “intermittency,” no output is available from the solar and wind sources at many times of high demand; therefore, during those times, other sources must supply the juice. This practical problem is presented most starkly in California, where the “renewable” strategy is based almost entirely on solar panels, with only a very small wind component. Daily graphs published by the California Independent System Operator (CAISO) show a clear and obvious pattern, where the solar generation drops right to zero every evening just as the peak demand period kicks in from about 6 to 9 PM.

Commenter Sean thinks he has the answer: “Given the predictable daily power generation cycle of solar in sunny places like California and the predictable daily demand which peaks in the evening perhaps solar generators should be required to have electricity storage equivalent to the daily generation of their PV system.”

I thought it might be instructive to play out Sean’s idea to see just how much solar generation capacity and storage it would take to make a system out of just those two elements that would be sufficient to fulfill California’s current electricity requirements. Note: this is an exercise in arithmetic. It is not complicated arithmetic. There is nothing here that goes beyond what you learned in elementary school. On the other hand, few seem to be willing to undertake the effort to do these calculations, or to recognize the consequences.

We start with the current usage that must be supplied. Currently, the usage ranges between a low of around 30 GW and a high of around 40 GW over the course of a day. For purposes of this exercise, let’s assume an average usage of 35 GW. Multiply by 24, and we find as a rough estimate that the system must supply 840 GWH of electricity per day.

How much capacity of solar panels will we need to provide the 840 GWH? We’ll start with the very sunniest day of the year, June 21. California currently has about 14 GW of solar capacity. Go to those CAISO charts, and we find that on June 21, 2021, which apparently was a very sunny day, those 14 GW of solar panels produced at the rate of about 12 GW maximum from about 8 AM to 6 PM, about half that rate from 7-8 AM and 6-7 PM, and basically nothing the rest of the time. Optimistically, they produced about 140 GWH for the day (10 hrs x 12 GW plus 2 hrs x 6 GW plus a little more for the dawn and dusk hours). That means that to produce your 840 GWH of electricity on a sunny June 21, you will need 6 times the capacity of solar panels that you currently have, or 84 GW. When 7 PM comes, you’ll need enough energy in storage to get you through to the next morning at around 8 AM, when generation will again exceed usage. This is about 13-14 hrs at an average of 35 GW, or around 475 GWH of storage.

That’s June 21, your best day of the year. Now let’s look at a bad day. For the past year, a good example would be December 24, 2020, which besides being one of the shortest days of the year, must also have been rather cloudy. Production from the existing 14 GW of solar capacity averaged only about 3 GW, and only from 9 AM to 3 PM. That’s 18 GWH in that window (3 GW x 6 hrs). Then there was another about 1 GWH produced from 8 to 9 AM, and another 1 GWH from 3 to 4 PM. About 20 GWH for the whole day. You need 840 GWH. If 14 GW of solar panels only produced 20 GWH for the day, you would have needed 588 GW of panels to produce your 840 GWH. (14/20 x 840) That 588 GW of solar panels is some 42 times your existing 14 GW of solar panels. And when those 588 GW of capacity stop producing anything at all around 4 PM, you are also going to need at least 16 hours worth of average usage in storage to get yourself to 8 AM the next morning. That would be around 560 GWH of storage.

So you can easily see that Sean’s idea of providing storage “equivalent to the daily generation of the PV system” doesn’t really get to the heart of the problem. Your main problem is that you will need capacity of close to 15 times peak usage (nearly 600 GW capacity to supply peak usage of around 40 GW) in order to deal with your lowest-production days of the year.

Cost? If you assume (charitably) that the “levelized cost” of energy from the solar panels is the same as the “levelized cost” of energy from a natural gas plant, then this system with 15 times the capacity is going to cost 15 times as much. Plus the cost of storage. In this scenario, that is relatively modest. At current prices of around $200/KWH the 560 GWH of storage will run around $112 billion, or around half of the annual budget of the state government of California.

But you may say, no one would build the system this way, with gigantic over-capacity in place just to cover the handful of days in the year with the very lowest solar output. Instead, why not build much less solar capacity, and save up power from the summer to cover the winter. Since the average output of the solar facilities in California is about 20% of capacity averaged over the year, then you ought to be able to generate enough power for the year with capacity of about 5 times peak usage, rather than the 15 times in the scenario above. You just will need to save up power all the way from the summer to the winter. Oh, and you will need a huge multiple more storage than for the one-day-at-a-time scenario. If 180 days per year have less production than usage, and the average shortfall of production on each of those days is 300 GWH, then you will need 54,000 GWH worth of batteries (180 x 300). At $200 per KWH, that will run you around $10+ trillion. This would be about triple the annual GDP of the state of California.

But don’t worry, batteries to store power for six months and more and release it without loss on the exchange don’t exist. Maybe someone will invent them in time for California to meet its 2030 renewable electricity targets.

Any reader can feel free to check my math.

I just can’t believe that anybody talks about this as something remotely connected to reality.