A Sign of Sanity in Solar Manufacturing

I've been writing for some time about the chronic overcapacity in global solar manufacturing and the consolidation this is likely to produce.  Now here's a sign that at least one company realizes how bad the situation is.  GE is apparently delaying the construction of its previously announced Aurora, Colorado, thin-film solar panel factory, and "taking this opportunity to re-look at our solar strategy."  I couldn't find a GE press release to back this up, but it's been reported by RECharge and confirmed by Forbes.  It's easy to read too much into a single event, but I think this looks significant, particularly in the wake of Monday's Chapter 7 bankruptcy filing by Abound Solar, incidentally another recipient of a sizable federal renewable energy loan guarantee.

If this information is correct, GE is backing away--for at least 18 months--from building a 400 MW thin-film photovoltaic (PV) solar line in Colorado.  That suggests that they have concluded that even a brand new facility using the latest technology and large enough to compete on scale with thin-film leader First Solar wouldn't be able to earn an attractive margin in this market.  And as a global competitor, GE would presumably regard the new US tariffs on China-based PV manufacturers as insufficient to resolve global PV overcapacity that appears to be stuck at about the same magnitude as demand, despite the continued rapid growth of the latter.

In the last year I've seen numerous articles and blog posts attributing the recent PV price declines to the predicted scale-related effects that have long anchored the industry's central narrative: If we build and deploy enough PV, the cost will fall to the point at which it will be competitive with conventional electricity generation.  That may still be true in the long run, but few of these advocates seem to have understood that the industry was getting ahead of its own narrative--that a big slice of the recent price declines was the result of intense competition among producers who over-expanded and whose margins have contracted sharply or turned negative in the process.  That's a good reason for GE to hit the pause button and focus on improving its technology in the lab, rather than the fab, while other, less well-capitalized firms struggle to survive long enough to participate in the expected growth surge when solar reaches "grid parity" on a sustainable basis.

PV is an important energy technology with a bright future, but its present doesn't look so great.  It's not unusual for manufacturing industries to experience boom-bust cycles, though in my experience those are more common in commodities like chemicals and fuels.  However, it is distinctly unusual for governments to contribute so much to the inflation of the boom part of the cycle through a wide array of incentives, loan guarantees and loans to manufacturers and with subsidies--in some cases extravagantly generous ones--to the industry's customers.  Such interference may have been necessary to jump-start PV supply and demand, but it will almost certainly make for a harder and messier landing for companies, investors and employees, and in cases like that of Abound Solar for taxpayers.  

Commodity Cycles and Renewable Energy Costs

It's an article of faith for many that the costs of renewable energy sources will continue to decline, even while the costs of many other forms of energy increase. This view is supported by the impressive experience with both wind turbines and solar photovoltaic (PV) installations over the last couple of decades. The latest survey of US solar cost trends from the Lawrence Berkeley National Laboratory (LBNL) showed that between 1998 and 2009, the capacity-weighted average cost of PV installations fell by more than 30%, from $10.80/Watt to $7.50/W, with another drop of roughly $1/W occurring during 2010. However, it's important to think about what's behind these trends when assessing their future progression. Continued cost declines are subject to many uncertainties, not the least of which is the current rising tide of commodity prices as the global economy recovers.

The rationale for continued cost declines that I encounter most often is based on volume: If we install more wind and solar capacity, costs will fall in a virtuous cycle, making subsequent installations cheaper and prompting even more of them. The underlying logic behind this argument derives from empirically observed "experience curves", in which cost components such as manufacturing fall by a set percentage for each doubling of cumulative output. The problem with these curves is that they tend to flatten out fairly quickly, delivering their maximum effect in the early years of a technology, when doublings are frequent. Then they slow significantly as the technology gains scale and the interval between doublings grows. A look at global wind turbine capacity shows that it doubled four times between 1996 and 2008. At the currently expected pace of additions, the next twelve years could produce just two or three doublings. And as the experience curve effect slows, other factors including commodity costs can overwhelm it.

We saw that a few years ago, when the boom economy stoked by the inflating global financial bubble drove up commodity and construction costs to such a degree that wind project costs stopped falling and began to rise. These pressures eased after 2008, mainly because the recession and financial crisis reduced overall demand for construction and raw materials, while the rapid growth of renewable energy equipment manufacturing, especially in Asia, led to overcapacity and stronger price competition. That probably accounts for much of the effect that LBNL saw in its solar trends in the last two years, rather than the sort of scientific and engineering improvements that drive experience curve effects.

So what is likely to happen as the economy rebounds and the slack that developed is taken up? We're already seeing the early results in higher prices of raw materials such as steel. And higher prices for oil and coal, which are important inputs in the extraction and processing of many other raw materials, should intensify the pressure already being felt from the demand in developing Asia. For renewables, this could be further complicated by tightening supplies of critical materials such as rare earth metals, Tellurium, Lithium and other key ingredients of electric motors, thin-film solar modules, and electric vehicle batteries. Toyota and other makers of hybrid and electric vehicles are working on ways to circumvent these materials, but the results might not come fast enough to prevent a crunch.

The renewable energy industry is counting on a return to economic growth to boost demand for electricity. Together with government-mandated renewable energy targets, that should translate into increased demand for equipment like wind turbines and solar panels. However, if the underlying source of that demand increases their material and construction costs just as governments are coming under increasing pressure to reduce subsidies for the industry, the result might be that the end-users of these products could see their effective costs go up for the first time in several years. That would be a rude shock for those who believe that these costs must inevitably fall as installations grow.

Misguided Incentives

Today's Wall St. Journal includes an interesting article on the emerging controversy concerning Germany's subsidies for solar power and their unintended consequences for that country's solar industry. It seems that solar incentives there have been so generous that they have discouraged German solar manufacturers from focusing on becoming competitive, rather than merely bigger. As a result, a growing share of the incentives is going to foreign firms that can sell these products cheaper. The hue and cry about this suggests that perhaps the original motivation behind the subsidy program, which not long ago was paying as much as a dollar per kilowatt-hour for power generated from solar panels, had at least as much to do with industrial policy as protecting the environment. In fact, Germany may have harmed the environment by wasting money on an impractical solution for such a cloudy place, when the same funds could have bought much greater emissions reductions in other areas of the economy. This should serve as a cautionary tale for those who are promoting similar incentives here, and for columnists--even those with a Nobel Prize in Economics--who argue that going green will be cheap. It won't be if we encourage the wrong technologies with bloated incentives.

At the heart of the solar debate in Germany is something called a "feed-in tariff" or FIT. It requires utilities to buy the output of qualifying solar power installations at a guaranteed fixed price well above the prevailing price in the power market. What's unique about the FIT compared to incentives such as the US federal renewable Production Tax Credit of 2.1 cents per kWh is that the funds to pay this green premium don't come from the government but from each utility's ratepayers. In other words, it is a mechanism for redistributing wealth from utility customers to the owners of solar installations, whether the affected ratepayers receive any solar power or not. The paradox of the FIT is that it makes the most sense when a technology is at its very earliest stages, producing so little energy that the cost to average utility customers is just pennies a month. The more solar power is produced and bought at inflated prices, the higher utility bills go and the less competitive the entire economy becomes.

So far, this just sounds like a political matter. Germany decided to nurture a large industry to build and install solar products and chose to pay for it by sending the bill to utility customers every month. That might even make a certain amount of practical sense, if not for two facts. First, the subsidy remains extravagantly generous, even after having been significantly reduced in recent years. It currently stands at a range of 34-43 €cent/kWh, depending on the kind of installation involved. At current exchange rates, that equates to $0.50-0.635/kWh. A recent study comparing levelized power costs for a variety of power technologies puts the cost of unsubsidized solar power between $0.26-.32 for the crystalline silicon photovoltaic cells that most German solar firms produce, based on an average capacity factor above 20%. After adjusting for Germany's much poorer solar intensity, the cost of solar power might rise to as much as $0.40/kWh, still well below the level of the FIT. This makes un-sunny Germany a remarkably attractive place to sell solar panels, and German companies haven't been the only ones to notice this. Suddenly the FIT looks like a means for Germans to subsidize Chinese solar firms, and that is not going down quite so well. More importantly for the success of Germany's solar industrial policy, the Journal indicates that the head of one of the country's largest solar module manufacturers is now arguing that German suppliers will not become efficient enough to compete in the global market for solar panels unless they are weaned off such generous support.

The high effective cost of the emissions reductions these subsidies are buying ought to be of equal concern to German policy makers. Even if you assume that each kWh of power generated by FIT-subsidized solar panels backs out a kWh generated from coal, the extra premium over the cost of other low-emission power sources such as wind is enormous. The difference in the average solar FIT vs. Germany's FIT for offshore wind of 13 €cent/kWh ($0.19/kWh) yields an effective cost of CO2 reduction from solar of about $400 per ton. That compares to a current price for emissions credits on the European Climate Exchange of around $19/ton CO2. The more you pay for reducing emissions, the less of them you can afford to reduce, even in a prosperous country like Germany.

At the end of the day, German politicians appear to have spent billions of Euros of German consumers' and businesses' money to build a solar industry that has thrived on the installation of high-costs solar panels in one of the least suitable countries for solar power imaginable, and that may not be able to compete internationally without drastic restructuring. This initiative has also failed dismally as climate policy, purchasing less than 5% of the emissions reductions that could have been bought had this money been spent on other, more cost-effective power technologies or on energy efficiency. The further irony is that much of the German investment in solar technology to date would have to be written off should it turn out that the current generation of technology can't be made cheaply enough under any circumstances, and crystalline silicon cells ultimately give way to cells relying on non-silicon thin-film techniques or novel nanotech-based designs. These are the perils of industrial policy masquerading as environmental policy, and it is hardly a winning case for the application of a similar FIT in the US.

The Sun or the Atom

It looks like we might finally see a solar power installation built on a large enough scale to enable meaningful comparisons between it and our current largest low-emission energy source, nuclear power. Tuesday's announcement by First Solar, Inc. that it had negotiated a memorandum of understanding with the Chinese government to install up to 2,000 MW of capacity in Inner Mongolia could take solar out of the world of rooftops and into direct competition with central power plants. It also provides an opportunity to assess this technology on a more consistent basis with our other full-scale energy options, even if it's clear that they ultimately serve somewhat different portions of the power market.

For years I've been reading suggestions for covering large swaths of desert with solar panels, and this looks like the largest realistic proposal so far, involving 25 square miles of desert near the city of Ordos, China. While the Desertec project in North Africa might ultimately be much bigger, that still looks like a much more remote possibility, at this point, though it was also clear from First Solar's press release that their project in China will start at a modest scale of 30 MW and work its way up from there. I can't help wondering if some of the project's later phases might be contingent on continuing to reduce the cost of solar power from today's levels. As I noted recently, even with solar module costs below $1/Watt for First Solar's thin-film technology, non-module costs can still push total installed costs above $4/W. That would put the Ordos project in roughly the same category as a new nuclear power plant in terms of total cost, not just notional output.

And while we're looking at output, we ought to consider how comparable an installation of 2,000 MW of solar panels anywhere on earth would really be to two coal-fired power plants, as was mentioned in several news reports on this story. Although I couldn't find data specifying the actual annual number of peak-sun hours for Ordos City, a glance at this solar irradiance map of China suggests that this location gets around 6 kW/m2/day, equal to 6 peak-sun hours per day, on average. That gives this project an average capacity factor of 0.25, which means that 2,000 MW of peak solar power would generate roughly the same amount of electricity annually as one 700 MW coal-fired power plant or a single 550 MW nuclear power plant, if there were such a thing. Perhaps this project's biggest benefit is in its scalability. Unlike a new nuke, which would probably take about as long to build, the Chinese won't need to wait until the project was completed in 2019 to get useful electricity from it. Each sub-project would stand on its own, and the first one could start generating within a year or two.

On balance, then, this huge solar project could produce as much peak power as a pair of nuclear power plants (or large coal-fired plants)--though still quite a bit less than either of those technologies over the course of a year--while costing about as much as one large nuclear reactor, even allowing for significant cost improvement between the time the first and last solar panels are installed. How useful such a facility will be is largely a function of whether that region of China has a greater need for lots of power when the sun happens to shine, or reliable power around the clock. I'm as pleased as anyone that China appears to be diversifying its energy mix away from coal, even to a modest degree, and there's certainly nothing about building such a facility that precludes expanding nuclear power, since in the long run the country is likely to need lots more of both. Still, in terms of bang for the buck and without factoring in what this one project might do to help bring down the cost of solar power elsewhere, it hardly seems an obvious choice.