Friday, June 20, 2008

America's addiction to oil, part two

I wrote a lengthy article yesterday on the oil crunch. A couple of people pointed out the Tesla Roadster as an electric vehicle option and felt that I unfairly glossed over it. The Roadster is a really neat car, no question about it. However, it is extremely pricy, $109,000, and I still have doubts about its lithium-ion batteries. Also, it requires a 70A charging circuit, which is more than I have available in my house (we only have 100A service here, and putting 70% of that into charging my car would leave insufficient reserve to run the rest of the house). I think the Roadster is a great proof-of-concept vehicle, as is the equally impressive Aptera, but neither of these cars is quite "ready for prime time" and I left them out of my discussion because of that.

Another person asked me about solar power, specifically photovoltaic power. While I think PV power is going to be useful as a spot source, and to provide daytime surge power, there are serious issues that prevent it from being the backbone of our power grid. Last year (2007), the total electrical generation in the United States was approximately 14433 petajoules (see below). One square meter of solar cell, at 40% efficiency (which is about the best anyone has made so far), will yield, under average insolation conditions, about 5.66 megajoules a day, or 2065 megajoules a year. That means we'd need about 6.99 billion square meters of land completely carpeted with photovoltaic cells to generate that 14433 petajoules. That's 2698 square miles. Under more realistic efficiency values (around 8%) we'd need five times that, or nearly 14,000 square miles. We can do that (this is about 15% of the land area of Nevada, most of which we're not really using for anything anyway), but there are several other catches here.

First, photovoltaic power is only available when the sun is visible in the sky. This isn't the case at night. We'd have to find some way to store excess power generated during the day for use at night. There are a number of ways to do this (batteries, pumped hydroelectric, supercapacitors), but none of them is terribly efficient. So that reflects significant losses, which mean even more Nevada desert gets covered by refined silicon. Also, it turns out that the areas in the country that use the most power tend not to be those that have the best insolation. This means that we'd have to generate the power being generated in sunny, empty areas like Arizona and Nevada and transmit it to the areas that use it, like New York and Boston. Long transmission lines have high losses, as much as 50% for applications like this. This is why we typically generate power near where it will be used, and it's why electrical power is so much more expensive in the Northeast. If we tried to power the entire United States using a solar farm in Nevada, we'd probably have to cover most of the state with refined silicon.

Also, photovoltaic cells are expensive to make. The materials required to make a PV cell have to be very pure and must be constructed using very carefully controlled methods that require a good deal of energy. Right now solar cells cost something like $120 per kilowatt of generating capacity to make. At that rate, it'll cost around 500 billion dollars to make the solar cells required, and I'm not even accounting for losses due to inefficiency in storage and distribution. That also represents about two million tons of semiconductor-grade silicon - a couple orders of magnitude times the amount currently available or predicted to be available in the next several years. Maybe I've made a mistake in my math somewhere, but these numbers just lead me to believe that chasing photovoltaic as a prime source of electrical power is a mistake. I think PV as a "boost" source, and especially for microgeneration at the point of consumption, is potentially a good idea, but it's not the solution by itself.

Photothermal power is actually more appealing. The direct efficiency is about the same as photovoltaic, but there are several major advantages. First, the use of salt as a circulating fluid offers a relatively simple way to store energy for the nighttime hours; there's no need for batteries or pumped hydroelectric storage. Second, the design does not call for any significantly expensive materials; no need for millions of tons of semiconductor grade silicon, just ordinary concrete/metal construction and other technologies we've already mastered in existing power plant technologies. It appears to me that photothermal power systems can generate as much (or possibly even more) power than photovoltaic power for the same land footprint, at a fraction of the cost. So in response to the individual who asked me about photovoltaic power, I'd say that you should look at photothermal instead. We still need to carpet Nevada, but this time it's with plain glass mirrors with a thin layer of aluminum, not with refined silicon. A lot cheaper to make, and to fix.

However, allow me make another point about power generation: The main substitute for oil in the American economy is clearly going to be electricity, which we currently generate by coal (7212 petajoules per year), natural gas (2932 PJ/y), nuclear fission (2905 PJ/y), and hydroelectric (886 PJ/y), with a total generated electricity of 14433 petajoules in 2007 (all numbers derived from 2007 government reports). Our total fuel consumption for consumer motor vehicle transport in 2001 was 14876 petajoules, or slightly more than our total electricity production in 2007. That doesn't count fuel usage by commercial vehicles, mass transit, airplanes, or trains. We're just talking about consumer use here. The obvious conclusion from these numbers is that if we're going to replace our current fleet of gasoline-powered cars by plug-in electrical vehicles, it's clearly obvious that we're going to also have to, at a minimum, double our generation capacity, and probably closer to triple it to deal with losses. Yet another reason to restructure our lifestyles to reduce the distance we travel on a daily basis.

This post is long enough, so I'll wait for a subsequent one to talk about where hydrogen fits into the picture.