A new book, Taming the Sun: Innovations to Harness Solar Energy and Power the Planet, is available from MIT Press. The book touts the wonders of solar energy and tells us that there is an urgent need to invest trillions in more solar energy. The book is filled with outright errors. The author, Varun Sivaram, has a boundless faith in technical progress that he thinks will make solar cheaper and more practical.
Since solar cells and integrated circuits are both manufactured using silicon wafers, the author assumes that something like Moore’s Law must apply to solar cells. Moore’s law predicts the halving of the price of integrated circuits every 18 months or so. But integrated circuits become cheaper because they become smaller, taking less real estate on the wafer. However, solar cells don’t become smaller, because they need to collect sunlight, and the amount of energy they collect is proportional to the area they occupy. In any case, the cost of the silicon components of solar energy is rapidly becoming negligible. At least 75% of the cost is for mundane things like concrete, steel, power lines, land, etc.
Solar has the huge problem that it doesn’t work at night. It doesn’t work during the day if it is cloudy. Even the sunniest city in the USA, Yuma, Ariz., has 50 cloudy days a year. If there were a cheap and scalable method of storing electricity, the prospects for solar would be better. But the best method of storing electricity, pumped storage, is expensive and dependent on favorable terrain. Neglecting many costs, utility-scale solar farms cost about $2,200 per kilowatt of nameplate capacity. Under favorable conditions, the average power produced is about 25% of nameplate capacity. If a realistic computation of the unsubsidized cost of electricity is made, solar electricity costs 12 cents per kWh, under good conditions in sunny locations. Natural gas can deliver electricity for less than 5 cents.
