I just got back from Lynne Kiesling's excellent blog, which has a number of good entries related to electric power. Then I remembered that I write about that stuff too. Cool.
Anyway, the post of interest mentioned that it takes time to build more electrical generating capacity. Very true. But even if it's built, you don't get power from a plant immediately. You have to warm it up first.
My observation seems to be that if the temperature is decent, people hop in their cars, start them, and are in gear and moving seconds later. They wouldn't make good power plant operators. Unless you want to risk potentially very costly maintenance problems, you start up slowly.
In particular, steam turbines are touchy about uneven heating. The ones in use at big commercial nuclear power plants are quite large, and the shafts are very long. So long in fact that one of the first things you do when starting up a turbine is to put it on turning gear, so any heat it is exposed to is evenly distributed and you don't make the shaft bow from the weight of all the blades at operating temperature. Even small amounts of bowing of the shaft will cause vibrations which can destroy the turbine.
Piping runs are also very long, and the temperatures in them may vary over several hundred degrees. These factors cause great thermal expansion to occur during startup, accompanied by noises. Heating these pipes too rapidly can cause undue stresses or even failure.
Heat exchangers (see figure 1) like feedwater heaters are also big enough to expand by several inches as they heat up to operating temperatures. The hotter stages will be constructed with one end on wheels to permit the motion required. Meanwhile there are hundreds of interference fits between tubes and the bulkheads they pass through (called 'tube sheets') in these heat exchangers - any of them can leak, lowering the thermal efficiency of the plant.
The part of the plant that produces the steam takes time to warm up too. For nuclear reactors this is further complicated by the need to pull the fuel rods out in special patterns as directed by nuclear engineers to optimize the use of the reactor's power potential. You don't want to do things rapidly with a nuclear reactor unless it's an emergency shutdown (which is called a 'scram').
Once the plant is at operating temperature it becomes more flexible. It can be synchronized to the grid (another post later), and then its load can be varied over a certain range without affecting efficiency and reliability intolerably.
A little consideration of the above tells us something about how electric generation utilities are forced to operate. If they are to be able to pick up loads rapidly and reliably and to accomodate disturbances such as trips, they must always have more generators running than they need to carry the load. And because of the time it takes to heat up, these generators must be running well in advance of the load change. This might force them to run, say, 5 plants at 80% capacity instead of 4 plants at 100% capacity (this of course assumes you have 5 plants of identical capacity using the same fuel, or else there are several other things to consider).
But running 5 plants instead of 4 is expensive - the power to spin the turbine costs money whether it's generating salable electricity or not, and the 4 plant scenario is cheaper than the 5 plant one. So a utility must in a sense act like an insurance company, trying to balance reserves against anticipated demands and still make a profit.
Whoops, I promised another post. Terrific - I can think of about 3 others I've pledged, and one of them is about 2 months old now. The synchronization one will probably come soon though - I'm just getting warmed up.