(Photo credit cloudfront.net)
The solar eclipse of March 20th, 2015 and the havoc it threatened to wreak on the German grid generated a lot of portentous web coverage before it happened:
Will next week’s partial solar eclipse turn off the lights in Germany? Experts say the country’s electricity grid, which relies increasingly on renewable energy, faces a crucial test on the morning of March 20, when the moon will pass in front of the sun and block up to 82 percent of its light across Germany.
Some nail-biting web coverage while it was was happening:
10:54: Here comes the big stress test. The greatest change in PV power generation takes place between 11:00am and 11:15am.
And quite a bit of triumphant web coverage after it was over:
Clean Energy Wire: Energiewende passes solar eclipse stress test
Grid operators had warned the astronomical event would be “an extreme challenge” that will “stress test the power system” but in the end, all went “wonderfully”. Germany has passed the solar eclipse stress test with flying colours.
But was the eclipse really an “extreme challenge” for Germany’s grid? And what does it tell us, if anything, about the Energiewende? Let us examine the data.
The Fraunhofer Institute publishes an interactive graphic of German grid performance covering the week of the eclipse, and in this post I present some charts developed from this graphic. Figure 1 shows German solar output over the week of the eclipse, which caused the incision in the leading edge of the solar curve on March 20th:
Figure 1: German solar output, week of March 16 – 22, 2015
Figure 2 shows solar output on March 20th in more detail (readings are at 15-minute intervals):
Figure 2: German solar output, March 20th, 2015
Between 9.45am and the high point of the eclipse at around 10.45am solar output fell by about 7GW in an hour and between 11am and 12am it rose by about 13GW in an hour. The problem wasn’t the size of these swings – the German grid routinely handles larger ones – but the rate of change. Could Germany’s load-following plants ramp down and up this quickly, and what would happen if they couldn’t?
As it turned out they never had to. The bulk of the load-following burden was shouldered not by Germany’s load-following plants but by interconnector flows to neighboring countries. Figure 3 gives details. The red line shows solar generation at 15 minute intervals over the period of the eclipse, the black line shows Germany’s power exports over this period and the blue line shows what was left for the German grid to handle after subtracting exports. By reducing its exports by ~6GW at the height of the eclipse Germany lowered ramp rates to potentially manageable levels before its load-following plants even got involved. (Note that there is a problem with the times given by Fraunhofer, which show the eclipse peaking at around 3.30am. Figure 3 adds 7 hours and 15 minutes to these times to match them to the actual peak of the eclipse, which occurred at around 10.45 am):
Figure 3: Solar generation, exports and remaining requirement after subtracting exports from solar, March 20th solar eclipse
Ramp rates were then lowered further by a 3-4GW reduction in electricity demand during the eclipse, which was caused partly by some industrial consumers agreeing to cut their electricity usage during the eclipse and partly by people turning their TVs off and going outdoors to watch the spectacle. The reduction is illustrated in Figure 4, which plots Germany’s total generation over the period of the eclipse (I assume here that total generation over this period would have matched demand; Fraunhofer gives no demand numbers.)
Figure 4: Total generation and generation from “conventional sources” during March 20th solar eclipse
The green line in Figure 4 plots actual generation from German “conventional” plants, which include everything except solar, wind and exports/imports, during the period of the eclipse. (Wind generation in Germany during the eclipse was negligible, averaging around 0.5GW.) To meet demand these plants had to ramp down at a fairly constant rate of around 3 GW/hour for two hours. The German grid routinely achieves such ramp rates.
Figure 5 is another plot from Fraunhofer showing generation from all sources between March 16 and 22nd (imports/exports are not shown). The eclipse is defined by a small coincident dip in total generation and a small coincident spike in total generation from non-solar sources. Except for these wrinkles March 20th was a weekday much like any other weekday on the German grid:
Figure 5: Generation from all sources, week of March 16 – 22, 2015
The three graphs in Figure 6 compare solar generation during the eclipse with generation from gas, hard coal and pumped hydro, Germany’s main load-following sources. All three were ramped down when solar generation began to recover after the eclipse, but none of them at a rate that was visibly different to the rates observed at the same time on previous weekdays. Note also that most of the ramping was done by coal:
Figure 6: Generation from load-following capacity, week of March 16 – 22, 2015 (note the different Y-scales)
These results show that the solar eclipse of March 20th, 2015 was effectively a non-event on the German grid. It certainly wasn’t a “stress test” of the Energiewende, as has been claimed (Energiewende 1, Solar Eclipse 0, as Clean Energy Wire put it). The ultimate success or failure of the Energiewende will be judged by stricter and more meaningful standards.