The American Wind Energy Association (AWEA), put up a blog post titled Wind helps meet new Texas record for electricity demand. Now, that's something we would expect, except for this part, referring to the fact that the contribution from wind power in the area was 2 GW at the time of record electricity use:
The 2,000 MW were more than double the 800 MW that ERCOT counts on from wind during periods of peak summer demand for its long-term planning purposes, and enough to power about 400,000 homes under the very high electricity demand conditions seen yesterday.System operators officially say that they count on 8% of the rated output of wind power to be help meet demand. That is to say that for every wind farm built, 0.08 times the rated maximum output of the wind farm is that much less capacity that has to be covered by fossil fuel plants. Since 800 MW is the 8% mark, 2,000 MW is 20%, which is the capacity factor the combined wind farms in the area (most of Texas) achieved during the peak hour. Here is a graph for the entire day from ERCOT with the peak hour circled and a line drawn for the 8% of total output:
So, yes, the output during peak demand was higher than the level said to be "reliable", but it had dropped below the 8% level just 5 hours early... that same day. Wind output was at a low around 11am to 12pm and the peak hour was 4-5pm. It's hard to say that AWEA wasn't setting itself up for some backlash. Of course I'm not the only one to notice this. The Capacity Factor blog has a bit of a sharper tongue, writing that Texas grid declares Level 1 Emergency as ten thousand megawatts of wind power stands paralyzed. The Capacity Factor correctly points out that all nuclear plants were operating at full capacity during the peak hour, but commentators at the AWEA blog also correctly point out that thermal plants have a lower thermal efficiency when the temperature is the highest, like on August 3rd.
The Real Story
The reality is that system operators don't absolutely count on 8% of the capacity of wind farms. Some days it dips below that level and some days it doesn't. It is also not the case that system operators "count" on any given plant in order to meet demand. System reliability is based on a probabilistic assessment of all possible failures. The situation is slightly different for wind power because a sudden reduction in output isn't the result of a failure, but a change in the weather pattern. Nuclear plants are some of the most reliable units on the grid, but a loss of a nuclear unit is often the most restricting contingency for planners because they are the largest units.
It is true that wind power increases the grid management challenge, and the most obvious reason is simply due to the fact that other units have to respond to the difference between demand and what wind power delivers. From 12pm to 4pm on August 3rd both demand and wind output was increasing. That's a great situation, because that's less slack that other units have to quickly pick up. These two, wind output and power demand, however, are in no way guaranteed to be correlated and there are times when demand is going up while wind output is going down, and those times are planned for by the system operators as well.
The real story is also that wind power in Texas matters a great deal at this time, and it is one of the critical proving grounds for grid integration of large scale wind. Take a look at the history of the installation of electric capacity to the grid in the United States, via the EIA Today in Energy.
That green sliver is wind capacity added. Roughly 30-35% of the green area is actual energy generation capability added (compared to about 90% for the pink, nuclear sliver), and 8% of the green sliver is reliable capacity added... well, sort of... it's complicated actually. Either way, it's still energy that doesn't release Carbon into the atmosphere. This is why it is neither the case that wind power is bringing the grid in Texas to its knees or aiding in maintaining reliability. It helps meet demand with clean energy, and that's what matters.
Update: I put some more info up on my auxiliary blog.