I did this analysis to indicate that desert solar electricity is only about 1/10th as good as GEO or Solarsychronous SPS = solar power satellite. One half gigawatt is enough power for a large city. Very large cities need several gigawatts. Lets consider 12 square kilometers = 3 kilometers, East and West by 4 kilometers, North and South. There is a solar power tower close to the Southeast and Southwest corners of the property. They are about 200 meters tall. The land slopes up about 2% from South to North, but is otherwise quite flat. More slope would be better, but more is almost always hilly which is bad. You can see ideal sites are rare. Homes and ranches are practical except very close to the two solar power towers, but maximum height of trees and structures is typically 4 meters, but only 2 or 3 meters in a few spots. Exceptions are the 50,000 steerable mirror towers which range from 12 to 20 meters tall, a few of them with a wind turbine above the mirror. The mirrors average 100 square meters and produce an average of 400 watts per square meter in late June at 1 pm in their beam of sunlight. The heat exchangers near the top of the twin towers absorb an average of 2 gigawatts, and delivers 1.9 gigawatt to the steam turbines. Some additional energy comes from steam super heaters also on the tower. Electrical output is 0.5 gigawatts. With good luck the wind turbines and some photovoltaic panels produce 0.01 gigawatts for a total of 0.51 gigawatt put on the grid. Typically the grid accepts a bit less, so some heat energy is stored in the molton sodium-potassium nitrate in the heat exchanger, and a reserve tank. Sometimes this is sufficient to power the Southwest turbine at reduced power until an hour after sunset which is about the end of the peak demand period. Some additional mirrors may be located to the East of the towers to catch the late afternoon sun. While this would produce some evening reserve, it is likely not cost effective as the towers would be on private property.
As you can imagine, 50,000 mirror towers on 12 square kilometers, is near the limit to prevent shading each other in the late afternoon, especially in December when the sun is low in the sky. Shading is rather severe shortly after sunrise, but that is not very important as the wholesale price of electricity is low mid morning, and it does take about two hours to warm the sodium-potassium nitrate to optimum operating temperature each morning/3 hours typically in December. The installation cannot be enlarged except at diminishing returns, as a 4 kilometer beam produces an illuminated spot bigger than the heat exchanger, even if the mirror is minutely concave. A precision concave mirror is much more costly than plain mirrors and a 500 square meter heat exchanger has considerable heat loss in a high wind. I suppose transparent shutters would help on windy days, and the heat exchanger could be a bit larger than 500 square meters.
The start up crew arrives before sunrise, each morning at the SE tower, for the startup procedure. Typically they leave one technician to monitor, then go to the Southwest tower which will be in a poor position to receive energy until late morning. If there are no problems, all but two are off duty by one pm.
Perhaps ten employees, including two trainees, are needed as the towers need to be observed during start up on Saturday, Sunday, sickdays and holidays as well as week days, so payroll is not a trivial expense, even with the systems highly automated. Please embellish, refute and/or comment. Neil

w says: A receiver and rectenna almost as big would be needed for the microwaves from a solar power satellite. Both would also need operators and repairmen.

Me: w is correct. The satellite and desert 1/2 gigawatt are a bit easier to operate and maintain if they use space mature solar panels without mirrors, turbines and generators, but the huge microwave source and huge antenna will be essentually new technology which likely cannot be fully automated until we have a decade of operating experience. This is another good reason to operate the first several SPS in LEO, so that technicians can live on the satellite. The temperature cycling is likely more severe in LEO than in the desert and it will be an 88 minute cycle instead of a 24 hour cycle. LEO is, however, mostly spared the weather and seasonal cycles, and a semipolar orbit allows beaming to most of the countries of Earth, during part of the peak demand period with as few as ten satellites, I think. Neil