I recently asked physicist Richard Muller whether he thought the price-performance of solar electricity generation would follow a Moore’s Law-type curve. He said that this would not occur due to improving the efficiency of solar collection, as the current levels of efficiency – 20-40% – are reasonably high. However, he added
“I do expect the price to drop by a factor of 10, so we will have lots of solar.”
Well, in the nature of things, there’s definitely a limit to how much energy a solar PV collector can get from a square meter of sunlight. (There’s about 1kw of energy in a square meter – as I learned in Physics For Future Presidents, by Professor Muller – so we can expect to get 400 watts or less.) The amount of this energy per square meter we can collect will go up, but asymptotically approach (at best) the physical limits.
On the other hand, I’d argue that the cost of collecting it can go down a nearly unlimited amount – certainly multiple orders of magnitude. So what will solar PV look like in 2018 – ten years from now?
As we contemplate the future of energy, and the combination of utility-level and distributed energy, and of different types – solar PV, solar thermal (heat your own hot water for showers), wind, etc., one question I have asked myself is how much energy can realistically be produced by the solar collectors on the roofs of our houses and office buildings in the U.S.?
It turns out the United States government has done some research on this! There’s a very interesting set of Department Of Energy reports, including one (PDF) on the market opportunities for grid-tied distributed solar PV. It figures out, state by state, how much roof surface is available, how attractive the incentives and infrastructure are (e.g., is there net metering?) and uses some simple algorithms to come up with an expected market penetration for solar PV on commercial and residential roofs. The resulting amount of electricity generated in this distributed fashion is amazingly high. Their best case scenario has installed MWs of rooftop solar PVs rising from about 2,000 in 2008 to almost 25,000 in 2015, more than a factor of ten increase over seven years.
The report uses conservative numbers for solar PV cost improvements – breakthoughs and innovations like the ones mentioned in Technology Review every week (like this one), will make the market penetration even faster (and higher) as they come to market.
I was pleased to see that our government has done this kind of research. Think what could be done if funding for renewable energy research and development was an order of magnitude higher!
Moore’s Law depended (and still depends) on a constant flow of breakthrough technologies, processes, scale, and designs. You can’t necessarily predict how Moore’s Law will continue to hold two years from now, or five years from now, but you can be confident that through some combination of technologies, processes, and designs, the price/performance of IT will continue to decline at an exponential rate.
The top five green energy stories of 2008 give an indication that the same types of forces are at play in the green energy world. Numbers 1, 2, and 3 each represent a potential 10x reduction in the cost of the most expensive part of a particular energy flow. For number 4, Gore used the bully pulpit of a Nobel Prize and Oscar (and, oh yeah, he was nearly president) in a most constructive way. And number 5 illustrates that green energy technologies are on a growth rate of doubling about every 18 months.
Did these stories excite you as much as they did me? Were there other green energy stories in August that you feel are more important?
In his galvanizing speech a few weeks ago Academy Award and Nobel Prize-winner Al Gore exhorted the United States to “produce all electricity from “carbon-free sources” by 2018.” This is a pretty abstract goal, in those terms – Gore (appropriately) didn’t go into great detail about how this should be done or even what it means in specific practical steps. Depending on your point of view and background knowledge about energy, the goal may seem easy or incredibly difficult, or even impossible, especially without further analysis.
So I thought it would be interesting to run some numbers on the goal. The idea is not to define how it should be done, but to look at some very simple scenarios for how it could be done to get a sense of the scale involved. The calculation is based on the Topaz Solar Farm project, which California’s PG&E utility just contracted for – a 550 megawatt solar generating station.
My initial calculations makes some gigantic simplifying assumptions, so it’s not “correct” – but it should be the right order of magnitude. For details of the calculation, see the analysis below. The conclusion is as follows:
As a very rough estimate, we would need about 800 Topaz-sized plants, total cost about $1 trillion, to meet the U.S. electricity demand. And it would require about 8,000 square miles of sunny land.
1 gigawatt: Generating capacity for a “large” coal-fired generating plant
50 GW: California’s typical peak energy demand
24%: portion of PG&E’s currently contracted generating capacity that is renewable
For the purposes of this analysis, I’m making a few simplifying assumptions. These make the analysis “invalid” from a technical sense, but allow us to quickly see the big picture:
Electricity demand will stay constant: This may or may not happen – in California energy intensity (the energy used per person) is going down, and this summer absolute energy use went down. Amory Lovins of the Rocky Mountain Institute believes we can cut energy intensity by 50% via efficiency, which would definitely cut energy use. On the other hand, most scenarios dealing with energy use assume it will continue to grow.
Disregard base load issues: The sun don’t shine at night, but people still use electricity then. This is called “base load.” You often hear that “solar can’t provide base load,” which may or may not be true in the future, depending on storage technologies that might be developed. In any case, I’m not considering it in this analysis – I’m assuming “a megawatt is a megawatt.”
Disregard transmission issues: We’ll assume that if the energy is generated somewhere in the U.S., it can be used anywhere else it’s needed.
Disregard technology improvements – this calculation is based on the technology planned for the Topaz Solar Farm
We now have enough data to make the most simplistic conceivable analysis. How many Topaz Solar Farm equivalents (TSFs) would we need to supply total U.S. energy demand (given the assumptions above)?
Conclusion: In our simplified energy world, we’d need about 800 Topaz-sized plants, total cost about $1 trillion, to meet the U.S. electricity demand. And it would require about 8,000 square miles of sunny land.
Now, there are many ways that this analysis is “wrong” – since my assumptions simplify the world quite a bit. So it could easy be off by a factor of 50% or more. But, because the assumptions also tend to cancel each other out, it’s not off by a factor of five, say. For example, I’m not considering base load (which solar PV today can’t provide effectively), but on the other hand, solar PV is the most expensive energy source. We will probably need more than 800 plants, but a lot of them will be cheaper, per megawatt, than the Topaz Solar Farm.
In future posts I will expand this model to make it less simplistic and more realistic, and to take into account technology improvements, base load requirements, the ability of energy efficiency to change the demand line, and lots of other details that are just dropped on the floor for this analysis.
I’m very interested to hear your comments on this analysis. In particular, I hope for some constructive guidance on the next steps for making it more realistic. I want a simple model that’s easy for the layperson to understand, but which doesn’t over-simplify too much (as this model does). I’d consider this a “zero-order” approximation – the next one should be a “first-order” approximation.
The New York Times’ Dot Earth blog posted the text of Gore’s speech and allowed commenters to annotate it – interesting reading if you have a few hours to get through all the comments!
I wrote on Monday about why I am optimistic that we will come out of this energy mess in excellent shape. But, my optimism is not unalloyed – there are a lot of questions still to answer.
Is there truly enough capturable solar energy streaming down on the Earth to power a good lifestyle for all 9 billion of us in 2050? Clearly not, at least at the U.S.’s current per capita energy intensity. What about at 50% of our current energy use? That’s a target that many think we can accomplish here in the U.S., so why not around the world?
What about all the C02 we’ve stuck up there already? Can we do something about it that won’t end up causing as many problems as it solves? Certainly sensible steps like reversing deforestation will help a lot, but do we have time, and do we know how? Can we grow a rainforest from a burned-out meadow, even if it use to be a rainforest? This is not clear – but we should figure it out.
Can we do any of this fast enough? I’ve argued that the technology and knowledge are here for reducing our energy footprint in the U.S. by 50% and replacing all of the remaining energy needs with renewables, but is there time and will to do it? The sheer manpower that it will take? Even if owners of commercial real estate were willing to do the necessary retrofits to achieve the goals, because they are cost effective? More importantly, if every one agreed to do it, are there enough architects, contractors, HVAC installers, and electricians to do the work?
There’s a similar question for residences – most residences get enough solar energy flux on the roof to offset a good portion of their electricity use – but even if the cost were free, after first year saving, who would do the 100 million installations? Even if spread over ten years, that would keep 25,000 installers busy every day.
There are many more such questions – can we successfully combine distributed power generation (e.g., on residences) with utility energy on a gigantic scale? Where do all the materials to do these installations come from?
I’d love to hear your questions and comments about whether you’re optimistic, the obstacles you see in the road ahead, and your ideas on how to overcome the roadblocks.
Here are a few tips to help reduce your carbon footprint:
If you are going to drink wine anyway, consider drinking one of Far Niente’s varietals. They’ve installed a 400kW solar PV system (PDF of SF Chronicle article) that results in a net-zero energy bill and offsets a large percentage of their CO2 emissions
When flying, which we know is one of the worst activities from a carbon standpoint, you can at least connect through Denver International, which just dedicated a 2MW solar system (PDF of Sharp Energy press release).
In his call to action two weeks ago, Al Gore compared the future development of solar electricity sources to the development of the semiconductor industry. His implication was that Moore’s Law, which reliably predicted that the price/performance of semiconductors doubled every 18 months, would also apply to photovoltaics.
“But does Moore’s Law also apply to the solar energy industry? The short answer is no. As with microprocessor technology, the price and performance of photovoltaic solar electric cell is improving. And Gore can clearly point to price drops of solar cells to make his case. But the efficiency of those solar cells — their ability to convert sunlight into electric energy — is not increasing as rapidly.”
The article goes on to suggest reasons that Moore’s Law might not apply – there’s a lot more to solar panels than just silicon, while the price/kilowatt has been coming down, it doesn’t seem to be coming down fast, etc.
“We are not that far away from a tipping point where energy from solar will be [economically] competitive with fossil fuels.”
Kurzweil characterizes solar energy technologies as “information technologies,” especially as nanotech gets into the picture.
“We also see an exponential progression in the use of solar energy,” he said. “It is doubling now every two years. Doubling every two years means multiplying by 1,000 in 20 years. At that rate we’ll meet 100 percent of our energy needs in 20 years.”
I think we may be at one of the most interesting points in human history, when technology is changing so fast around us that in twenty years the world will almost literally be unrecognizable compared to today. (One of the side effects of the Law of Accelerating Returns is that the world changes completely on a regular basis – it just gets faster and faster!)
The New York Times has a story today about the big box stores rushing to get solar cells on their roofs before a Federal tax break expires at the end of December. The article’s analysis is that they are primarily doing it for PR purposes, since PV-based energy is still a lot more expensive than conventional. The benefits of being able to say they are green are compelling. But the companies put a slightly different spin on it:
But retailers believe that they can achieve economies of scale. With coal and electricity prices rising, they are also betting that solar power will become more competitive, especially if new policies addressing global warming limit the emissions from coal plants.
Retailers, hoping to create a bigger market and positioning themselves at the forefront of a national shift toward renewable energy, are encouraging one another to join the bandwagon.
Solar energy will cost the same as power produced by coal, natural gas and nuclear plants in about a decade, a report released Tuesday suggests. By then, the price parity could propel solar adoption so that it accounts for 10 percent of U.S. electricity generation by 2025
If you listen to this kind of thinking, solar energy (which is defined as what, by the way?) is still far more expensive than other kinds. But solar energy, even today, has a finite payback time – if I put solar collectors on my roof, for example, eventually they will pay for themselves.
So that’s one way it’s wrong.
Secondly, the study assumes that conventional energy prices will go up by 3% per year. That could be a slight underestimate. Didn’t we just experience a three month period where gas prices nearly doubled? (That’s 100%, folks!).
I can’t make any argument about the assumption that solar energy prices will come down 18% per year. That’s a lot, by one metric, but we’ve certainly seen large and faster price drops in high tech in the past. Even the iPhone last month, which dropped in price by almost 50% in less than a year. Sure, that was partly through some magic AT&T financial pixie dust, but to the user, it’s a clear 50% price cut. There’s no reason similar magic pixie dust, whether from the government or from the utilities themselves, won’t contribute to market price declines.
The claim that solar currently accounts for less than 1/10th of a percent of the U.S. energy supply today is fine. But the assumption that it will still be less than 1 percent in 2015 (seven years from now) is curious. If we start at .1 percent, and double our solar usage every year, we end up at 128 times as much – 12.8% of today’s total. This is the amazing power of Ray Kurzweil’s “Law of Accelerating Returns.” Even if it takes two years for each doubling, we’re still up a factor of 32x in seven years. That means 3.2% today’s usage. Our total energy usage may also go up (although there are very good reasons to think it may not go up much and and will be starting a downward trajectory), but for a 32x increase in solar supply to translate to 1% of our total energy use, total energy use would have to double. Not too likely in the U.S., where population growth has stopped, and SUVs are starting their long decline.
Finally, there’s good reason to believe that solar energy will actually have a much larger share of U.S. energy usage, due to the power of “negawatts” (as explained brilliantly by Amory Lovins in this series of talks at Stanford in 2007), in which efficiency turns out to be the most cost effective way to power industry and create profits. Oh, and by the way, it significantly reduces our energy usage, by as much as a factor of five to seven!
The article combines a couple of types of fallacious thinking – that technological progress is linear, for example, rather than geometric, and that other factors, such as the desire to reduce greenhouse gases or realizing the benefits of negawatts throughout the economy, don’t have an additional accelerating effect on technology changes.