Michael Purdie |
Major cities in Texas were subject to numerous 100 degree-plus days this month. Houston, Dallas, San Antonio and Austin all had record setting electricity demand. If you have ever been to these cities during one of these days, it’s hot and there is very little breeze to cool you down.
The most extreme day for the grid was August 13, when power prices peaked above $1,000/MWh. When this occurs, the grid operator (in this case, ERCOT) takes action. ERCOT called for conservation measures because electricity reserves were below 2,500 MWs during the peak.
Why did this occur? Simple. The wind generating units in Texas produced less than 20% of what they’re capable of providing. By operating at less than a 20% capacity factor, wind units provided 633 MWs of power less than what ERCOT predicted during the daily peak demand. The chart below depicts the planned and actual wind generation during hours of the day. The power price curve is positively correlated with electricity demand. This graph shows that when wind resources are most plentiful is also when the electricity is least valuable.
Source: Platts Megawatt Daily, Aug. 17, 2015 |
A representative of a fossil fuel generator told Platts Megawatt Daily that gas and coal were operating at approximately 90% of their potential. What did better? Texas’ four nuclear reactors (two each at the South Texas Project and Comanche Peak) operated at 100% for the whole week. These four reactors provided nearly 5,000 MW of electricity when Texans needed it most. Assuming a 90% capacity factor over one year, the four Texas reactors provide power for 2.74 million people. This is roughly equal to the population of Dallas and San Antonio combined.
Comments
As a result of the challenges, ERCOT has moved to a multipart assumption that gives coastal wind turbines credit for a 56% capacity factor during summer peak demand (due to onshore sea breezes) and turbines installed on inland planes a 12% CF.
http://www.ercot.com/content/gridinfo/resource/2015/mktanalysis/ERCOT_ReserveMarginAnalysis42302.pdf (page 17)
Do you know how well those assumptions worked during the period analyzed?