Two terawatts by 2020
January 9, 2008, 10:41 pm
Filed under: Interventions in the carbon/climate crisis

solar use

When, a while ago, I was thinking of a slogan for solar power enthusiasts, perhaps to be used as the title of a blog, what I came up with was “Two terawatts by 2020”. I never started the blog — ignoring this one and the other one takes up pretty much all my blogging time — but the slogan sort of stayed with me.

The idea was that we now run a +/-13TW civilization; that’s the amount of power we generate and use. If we want people in developing countries to live as well as Europeans or the Japanese, and assume that they need to be in the same ballpark in terms of energy use to do so (a questionable assumption, admittedly — but better than assuming it can be done with less energy and finding we’re wrong) then on current figures we need to get to around 40-50TW by mid century. That’s a fairly high end assumption, but I’m fairly pro development.

If we want to get to that sort of energy use without aggravating the carbon/climate crisis unbearably, then 80% or so of that generating target needs to be carbon neutral, with the trend continuing on towards 100%.

With this in mind two terawatts of installed solar capacity by 2020 seemed to me to be a pretty good intermediate target. It would represent 10% or more of total energy use at that point, and form a nice gateway to 35TW of solar or so in 2050. (From 2TW to 35TW in 30 years represents a reasonably staid annual growth rate of 10%).

To get up to 2TW in 2020, though, requires pretty dramatic growth between now and then; today’s installed solar capacity is less than 1% of that. The nice news embodied in the graph at the top of this entry is that, if things follow the trend of the past few years, then that sort of growth is not inconceivable. The Earth Policy Institute’s solar economy figures (via Jim at the Energy Blog), represented in the graph above that demonstrates such a pleasing trend to the perpendicular, say that in 2007 solar cell production rose 50%, producing cells with a total capacity of 3.8 gigawatts. If the 50% annual growth could be kept up, by 2022 we would be adding a terawatt of new capacity a year — and our installed capacity would by that point be more than 2TW, since solar cells stick around. If new growth in our capacity to produce solar cells simply stopped right then then adding a steady terawatt a year would still give us 30TW of capacity by 2050 as long as the cells in question lasted for decades.

None of which is to say that this will necessarily happen. By and large, betting on exponential growth continuing is never very wise in a finite world. But the world is not, in all respects, finite — or rather, its finitude expands. As Moore’s law reveals, technologies can maintain exponential growth over decades if the economic setting is sufficiently expansive, both in terms of the growth of the economy and the growth of demand. As I mention in passing in the book, the x-rays available for scientific research double in “brilliance”, a technical measure of how good the beam is, every 15 months, and have been doing so for some time (see chart here).

In the case of X-rays the expansive economic setting is a willingness to pay on the part of research funders (which is not unproblematic — a willingness to pay for the UK’s very brilliant new X-ray source Diamond is being linked to an unwillingness to pay for work in astronomy and particle physics, to widespread discontent in those communities). In solar cell terms, the growth in use is closely linked to subsidies; that is why Germany leads the world in new installations. In 2006 it became the first country to install a gigawatt of solar capacity in a year. (That doesn’t mean Germany generated a solar gigawatt — night and clouds and winter eat into the capacity. But that lack of availability is a constant term and as long as we’re talking about exponential growth at this sort of rate we can in effect ignore it.)

At subsidy levels of a few dollars a watt the difference between supporting a gigawatt of use and a terawatt of use is fairly noticeable. Subsidies in the trillion-dollar-a-year-range would be hellishly large even by agricultural or military-industrial-complex standards.

That’s why this chart is a crucial one.

The worrying thing you’ll notice is that on this one the rate at which costs were falling flattened out at pretty much the same time that the first graph shows production capacity starting to leap. I have no idea why this is, or indeed if it’s a real effect or some sort of artefact of combining different data sources. Off the top of my head, one possibility is that it was around that time that the bulk price of silicon started to dominate total costs. Another is that it was about that point that subsidies started to take off, reducing interest in cutting costs (I rather doubt that one) [see update, below]. Another is that it was around this point that the costs of the module you put the cells in became an appreciable factor in total cost, and module costs are less easy to cut than cell costs. If anyone who actually knows about this stuff could weigh in, I’d be most appreciative.

Leaving aside the strange inflection, though, the chart still trends down. It doesn’t do so strongly enough to justify the industry rule of thumb that you can expect a 20% drop in production costs every time capacity doubles, at least not to my eye, but maybe that rule is for cells not modules. If that rule does still hold, them by my rough calculations you’d expect 50% growth in production year on year to reduce costs by 50% every five years or so, with costs dropping to 20% of today’s costs by 2020. That would be comfortably below the subsidy level, I think.

Anyway, I’m not sure I trust that old rule of thumb in a world of new markets and new technologies. I can imagine costs falling quicker with the advent of continuously processed thin films; with systems optimised for integration into new buildings; and with a lot of the sort entrepreneurialism and technological innovation of the sort found in Silicon Valley (and elsewhere). I wrote a feature about this for Nature in 2006, a piece which, thanks to kind sponsorship, is available here for free. Excerpt:

Amid all the cleantech opportunities, photovoltaics seem to resonate most with Silicon Valley’s history and culture.

One attraction is technological familiarity. Solar power has grown up in the shadow of the chip industry, using its cast-off materials and technologies. The silicon in traditional solar cells comes from the same suppliers who feed the chip market; new techniques to make solar cells often use processing technology, such as chemical vapour deposition, that is already widely used in the production of integrated circuits. Miasolé, a Silicon Valley solar start-up in which Kleiner Perkins has invested, uses expertise derived from the manufacture of computer hard drives.

But there is a broader cultural attraction, too. The potential of solar power to decentralize energy generation — a potential shared, to a lesser extent, by wind power — appeals to a culture that places huge societal significance on the empowering spread of the Internet. And a business community that saw personal computers go from hobbyists’ workshops to almost a billion of the world’s desks in 30 years is not fazed by the small size of the solar market today, but energized by the possibilities of tomorrow.silicon valley at dusk

It’s also a help that Silicon Valley is sunny not just in its outlook; a solar cell in California can produce almost twice as much electricity a year as one in the Ruhr.


A decade’s growth, however buoyant, doesn’t by itself mean that much. That growth needs to last for several decades to change an economy, and needs to accelerate to an even higher level to change the world.

The difference between growing at a more than respectable 25% a year [a widely accepted forecast figure discussed in the article] and at 44% a year — the rate at which volume grew in 2005 — is the difference between doubling in size in just over three years and in just over two. That may not sound a great deal, but over 15 years it means something growing at 44% would outdo something growing at 25% by a factor of eight. Between now and 2050, the difference is a factor of 500. And that could be the difference between providing just 2% of Earth’s energy needs — and 10 times those needs.

The remarkable thing is that the products of the semiconductor industry have grown at a yet faster rate for a similar length of time. If Silicon Valley can apply Moore’s law to the capture of sunshine, it could change the world again.

The company that was a focus for that article, NanoSolar, has just started shipping its thin-film panels; another of the companies, HelioVolt, has just announced manufacturing plans of its own, having raised $100m in capital last year. (Interestingly the two are going in different directions. Nanosolar is targetting installations on sites dedicated to power generation on the outskirts of towns and cities, while heliovolt is intrested in integrating solar cells into construction materials for new buildings.)

But while it seems possible that costs may start to drop more steeply it’s not certain they will. There may be all manner of factors waiting to impose a floor that costs can’t sink below, or for that matter to constrain the overall size of the market in some way.

In short, there’s no cause for celebration just yet, and there’s a long haul ahead — but the news is not bad. For the time being, at least, optimism of the will may be possible without too much pessimism of the intellect.

Update: James Annan points to this interesting story — Japan going for a 30-fold increase in domestic solar power by 2030. Which is encouraging, but also illustrates the scope of the challenge; this ambitious programme would involve growth at only 17% per annum, and would provide in the end only 0.04TW of capacity

Further update: B J Stanbery at Heliovolt tells me that the plateauing of price in ’87 does indeed correspond to the start of large scale subsidies by the Japanese government. Not the only factor, but a real one — it increased demand enough to lower downward trend in costs, and continuing/expanding subsidies there and in Europe have also had the effect of stabilising costs by producing a bottleneck in the silicon market (which may be alleviated, at least in part, by relocation of silicon foundries to places with both capital and cheap labour).

Graphs from Earth Policy Institute, picture of Silicon Valley used with very kind permission of Charles O’Rear


1 Comment so far
Leave a comment

This seems like a very worthwhile ambition. People forget that the challenge of overcoming the climate crisis also generates the opportunity to build a society with a sustainable energy basis.

The transition from relying on a stock of fossil fuels to a flow of renewable energy will certainly be difficult, but it seems like the kind of thing that will only need to be achieved once in human history.

Comment by Milan

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: